Faculty of Arts, Computing, Engineering and Sciences
DEPARTMENT OF COMPUTING
Undergraduate Computing Programme Project Report as part of the
BSc/BSc Honours Degree in (please tick) :
Business Information Systems
Computer and Network Engineering
Forensic and Security Technologies
Games Software Development
Games and Interactive Media Technologies
Internet and Business Technologies
Interactive Media with Animation
Web Information Systems and Services
Web Systems Design
Other degree (please specify):
Project Title: Recommendations for the suitability of Cloud Based
Business Intelligence tools for SME’s
Author: Geoffrey Fenwick Supervisor: Stephen Lofthouse
Date of Submission:
Type of Project (please tick)
Application Artefact Investigation
Confidential (please tick) YES NO
SHEFFIELD HALLAM UNIVERSITY
the suitability of Cloud
Intelligence tools for
Table of contents Page
Executive summary v
1.0 Introduction 1
1.1 The Problem identified 1
1.2 Project Aims 2
1.3 Project Objectives 2
1.4 Project Justification 3
1.5 Expected Results 3
2.0 Research 4
2.1 Literature Review 4
2.1.1 BI and Why it is required? 4
2.1.2 Decision Support Systems(DSS) 5
and their relationship with BI.
2.1.3 How BI can be used 6
2.1.4 Evaluation of BI tools 9
2.2 Research Methodology 11
2.2.1 Induction vs Deduction 11
2.2.2 Research Philosophy 13
2.2.3 Quantitative vs Qualitative 15
2.2.4 Methodology Justification 16
2.3 Research Tools 16
2.3.1 Methods of data collection 17
2.3.2 Different types of data collection 18
126.96.36.199 Action research and case studies 18
188.8.131.52 Experimentation and Observation 19
184.108.40.206.1 Experimentation 19
220.127.116.11.2 Observation 20
18.104.22.168 Questionnaires and Interviews 21
22.214.171.124.1 Interviews 22
126.96.36.199.2 Questionnaires 23
2.3.3 Issues within data collection 25
188.8.131.52 Quantitative Research 25
184.108.40.206 Qualitative Research 26
2.3.4 Choice of data collection method and justification 27
3.0 Research Implementation 28
3.1 Aims of the Interview design 29
3.2 Aims of the questionnairecomparison chart data collection 29
3.3 Justification of research implementation aims 29
4.0 Analysis and Results 30
4.1 Analysis Method for questionnaire 30
4.1.1 Interpretation of the questionnaire 32
4.2 Analysis Method for interview 32
4.2.1 Interview with Alan France 32
4.2.2 Interview with Max Willis 33
4.2.3 Interpretation of the interviews 33
4.3 Data collection analysis 34
4.3.1 Quantitative analysis 34
220.127.116.11 BIRT onDemand 360 34
18.104.22.168 Jaspersoft Live 37
22.214.171.124 Microstrategy Cloud Enterprise 39
126.96.36.199 Quantitative summary 41
4.3.2 Qualitative Analysis 42
188.8.131.52 Interview with Max Willis 42
184.108.40.206 Interview with Alan France 43
4.3.3 Analysis Justification 44
5.0 Conclusion 45
5.1 Understanding CBBIA suitability for SME's 45
5.2 Suggesting CBBIA usage for SME's 46
6.0 Critical Reflection 49
6.1 Positive factors 49
6.2 Negative factors 50
6.3 Exception handling 50
6.4 Further research 51
I would like to thank Steve Lofthouse for all of the support throughout the project and
enthusiasm also thanks to my family for their moral support when things started going
I would also like special recognition to dropbox as without them my project would have
probably been lost a few times.
I would also like to show gratitude to Chelsea Palmer whose English writing skills help
me communicate the written word properly.
Finally I would like to thank Guy Conway for his advice and information that gave the
project a more professional edge.
Cloud computing is a new area of Information technology where data centres can hold
data remotely and enables access from anywhere with an internet connection.
Combining this with Business Intelligence (BI) seems like a logical step forward
considering that all of the processing can be done remotely. This enables BI to be used
by smaller companies that otherwise would not have the resources to utilise a BI tool.
This project is aimed towards Small to Medium sized Enterprises (SME’s) to understand
how they can utilise this new service that has been made available. The project
attempts to understand what these Cloud Based Business Intelligence applications
(CBBIA) have on offer for SME’s and whether this is actually useful. As opposed to this
the project will also look at what an SME will require from a CBBIA in order for them to
be persuaded to use one. The idea of this project is to deliver suggestions that will
discuss what the issues are and key points that SME’s have to consider. It will then
deliver a report on whether a CBBIA can be utilised for an SME. This report can be used
to understand the subject of CBBIA’s and how the different applications can be used for
an SME. It will also give information about the precautions and some requirements that
an SME will want from a CBBIA implementation. There has been no formal study
between the relationship of SME’s and CBBIA’s as cloud computing is still a new area
for study. Therefore this project is more of a pre study as further research will give
some interesting results in CBBIA’s and SME’s.
1.1 Problem Identified
To understand what Business Intelligence (BI) is, a good place to start would be
discussing why it is required and why it exists. The idea behind the use of BI is simply to
“help make better decisions faster” (Vitt et al 2002). BI is all to do with analysing data
and giving answers fast which lead to making better business decisions. However the
first step involved with creating a BI system is asking what data is required to answer
the question. If wrong data is chosen then the answer may not making any sense when
looking at the bigger picture. If BI is used correctly then it will enable a user
manipulation of results in many different ways for example using drill downs, pivot
tables and calculated metrics. From the analysis there will be a number of questions
around the results found; this will be insightful towards answering the business
question. This insight is where patterns can be seen and recognised to be acted up,
possibility to pursue with further business questions. Once the business questions have
been answered substantially it is then possible to act upon these findings.
There are many tools which can be used for Business Intelligence (BI) which are
exclusive to large enterprises with a budget of millions. This is because of the
supporting IT infrastructure that has to be built. However there is a problem for Small
to Medium sized Enterprises or SME’s; they do not have the budget to support the IT
infrastructure to use business intelligence. Therefore they lack the decision making
capabilities that can be gained when using business intelligence. This gap between
large enterprises and SME’s is now less prevalent because of cloud computing. The
traditional way of using business intelligence is to build a data warehouse or data mart
and then extracting data using an Online Analytical Processing (OLAP) method. The new
cloud computing architecture means that all of this expensive hardware required for
business intelligence can now be used through a portal or web browser. This has
increased the amount of business intelligence software applications available because
they are cheaper to make and deploy on the cloud. With there being so many business
intelligence software applications on the cloud now, which one is the most suitable? Is
there one that can fit all businesses or does each application have its own purpose?
1.2 Project Aims
The aim of this investigation is to create a number of reviews to give a comprehensive
outcome on which tool is best for the job. Each review will be based upon a Cloud
Based Business Intelligence Application (CBBIA). They will then be evaluated and given
a scenario for which they would be best suited, considering the cost of the service,
most effective area and whether they provide the same results. Using the information
that will have been concluded, the investigation will try and apply a business
intelligence tool onto a real world scenario and gain feedback on what they would think
about the solution.
1.3 Project Objectives
The project will be following a logical order so that an appropriate conclusion will be
made. Firstly an investigation will be made into the different business intelligence tools
used for SME’s if any. This will give information to base decisions on how CBBIAs can
replicate these applications. The investigation will then identify current CBBIAs on the
cloud ready for comparison.
Before an in depth analysis can be made it will be logical to analyse how to compare
tools and define how the tools should be compared defining measures. This will give
me a resulting score and firm basis to form an opinion on which tool should be used
and where. The tools will then have to be critically evaluated and the strengths and
weaknesses drawn out for a comparison. This comparison will then be evaluated giving
a comprehensive report on recommendations for SME’s to possibly follow when
considering CBBIAs. This will also be delivered as a case study on a specific organisation
to how my conclusions can be applied in the real world with feedback.
1.4 Project Justification
The reason for this project is to be able to offer a set of recommendations comparing
the different types of CBBIA tools that are available and where they are best suited. At
the moment there is no review that covers the different BI tools and compares what
they can offer from a general perspective. Most organisations when approaching BI will
do their own assessments which will take time and resources to do this investigation.
This project will offer a more specific way in how a tool should be used and where it
could be best suited.
1.5 Expected Results
The results from this investigation should deliver answers to the following questions:
Is a CBBIA needed for an SME?
What types of CBBIA’s are available for SME’s?
What requirements does an SME have from an CBBIA?
How is business Intelligence used within organisations at the moment?
Can this be replicated using a CBBIA?
What requirements does a CBBIA need?
2.1 Literature Review
The Aim of this section of the project is to provide further information about Business
Intelligence (BI) and its development. It is hoped that this will facilitate a fuller
understanding of the research question and define why it will develop new knowledge.
This will be accomplished by defining what BI is, describing why BI is required and its
origins, together with the advantages and disadvantages of its application this will then
be tied together through a discussion about its relevance to the project.
2.1.1 BI and why it is required?
BI is the process of utilising information to answer business decisions. Loshin describes
BI as a type of practice which means that without the personnel BI tools are of little
value (2003). The three areas which are covered within Business Intelligence are
helping make better decisions faster, converting data into information and using this
knowledge to approach management (Vitt et all, 2002). These areas cover the higher
level concept of what business intelligence is and suggests that it is done to support or
inform managers to make strategic decisions. Turban et al describes BI as an evolution
from a Decision Support System (DSS) which is the name given to a selection of tools
that can directly help managers to utilise technology to help make intelligent business
decisions faster. It sounds very similar to the BI description given by Vitt which suggests
that DSS has simply been adapted to the new name of BI. This could mean that the
same concepts and designs that originally are used for DSS can be used for BI. Turban
(2011) furthers the definition of BI by saying that “BI’s major objective is to enable
interactive access to data, to enable manipulation of data, and to give business
managers and analysts the ability to conduct appropriate analysis”. This enhances
previous definitions because it suggests the use of technology to make the data easy to
manipulate, therefore enabling non-technically inclined managers to use tools which
are highly complex without technical know-how. It does this by analysing historic data,
situations and performances in order for a decision maker to help gain the best
possible outcome through previous happenings.
2.1.2 Decision Support Systems (DSS) and their relationship with BI
As was previously stated BI was developed from DSS which has benefited from the
technology available to make more intelligent decisions. Shimizu et al suggests that
organisations make a decision based on successful past experiences and refine them as
they are known to be successful. He goes on to state that decisions are formulated and
selected by quantitative methods of Operations Research and qualitative methods for
the use of artificial intelligence formulated through expert systems. This means that all
organisations will have some sort of method of considering decisions for strategic
choice. This way helps an organisation refine procedures instead of re-create the same
experiences. Using Artificial Intelligence it is possible to run models where based on
previous experience certain outcomes can be decided quickly and with the most
amount of support. Marakas defines a DSS as “a system under the control of one or
more decision makers that assists in the activity of decision making by providing an
organised set of tools intended to impose structure on portions of the decision-making
situation and to improve the ultimate effectiveness of the decision outcome”. This
means that a “decision maker” can use a DSS to control certain areas to provide
outcomes that are possible decisions to choose from. The DSS will help the decision
maker use the data previously gathered and forecast what outcomes can happen. This
will improve the chances of making the best choice in a scenario. This definition is more
like a generalisation of what a DSS means because it is not widely agreed upon. Turban
talks about DSS as an umbrella term and states that any computerised system within an
organisation is there to help improve the quality of decision making. He also suggests
that DSS can be considered as an application specific concept, in that an organisation
will build a model to consider the different outcomes from a scenario.
The reason that BI exists is because of the foundation of DSS and the ability to make
more informed decisions with a bigger chance of success. The key difference between
these two systems is that “a DSS application is typically built to support the solution of
a certain problem or to evaluate an opportunity” (Turban 2011) which suggests that a
DSS is purpose built for an event and will not apply for many situations unless they are
the same, whereas BI systems are built to monitor data and identify problems using
analytical methods. Using BI will create reports and updates on how the data is
performing and can offer predictions into possible scenarios however it is the user who
will be required to be aware of any situations that arise which can cause problems. DSS
applications can therefore be used in a different context, which are more specific to a
2.1.3 How BI can be used
There are a number of ways in which BI can be used but there are many constraints on
what type of BI can be used within an organisation. This is why there is no single
answer that can deliver a result; BI systems have to be tailored to the organisation and
depending on their usage integrated, Vitt (2002) talks about enabling BI within an
organisation and the key factors which can limit the analytical capabilities available.
Considering when the book was written technology has changed dramatically but the
factors have remained the same, they are: Technology, Processing Power, Data
Volumes, Network Technologies, Standards, BI software, People, Culture.
The new technology available for the latest BI systems has enabled interaction through
a mobile device for example a tablet or phone which covers all of these aspects (SAP a
2012). The technology is there to implement the platform, the processing power that is
available can be negated if built on the cloud, and because it is remote the data will be
stored elsewhere not on the mobile device, providing the data through wireless
methods is a common method of connecting to a network. This method that SAP
provide allows organisations already using SAP as their Enterprise Resource Planning
Application (ERP) to utilise this type of BI easily. This is because the other factors that
Vitt suggests are already considered; they have adopted the culture, the people and
standards that would normally cause issues with ways of working. From here it is
possible to look at the types of BI applications that Loshin (2003) has identified. He
suggests that all BI applications can be based upon two areas, Strategic and
Operational; Operational application “helps run the business” and Strategic
implementation “helps improve the business”. Some of the most common applications
are described below and it will be important to know what types of BI are used within
Customer Analytics which involves the analysis of data gathered from customer
relationship management (CRM) software to improve customer relations identifying
characteristics that are most likely to help run the business. Some of the analytical
benefits of implementing BI for a CRM is that it is possible to do customer profiling,
targeted marketing, personalisation, collaborative filtering, these features will be able
to increase the customer satisfaction, lifetime value and loyalty. Other features will
allow monitoring of customer interactions with an organisation and help to create
more business. (Loshin, 2003)
Human Capital productivity Analytics is looking inward towards the organisation and
helps to “streamline and optimise people”. Some of the type of analysis the can be
applied here from BI is monitoring the productivity of employees by implementing
measures. For example a call centre could monitor the amount of time that a customer
spends waiting on a call centre with the intention of being able to improve the
customer satisfaction. (Loshin, 2003)
Business productivity analytics is where organisations productivity is monitored from a
whole perspective implying that all areas are linked together somehow and using BI it is
possible to navigate through data to find the problem. Some of the analysis techniques
that can be done to find out problems within an organisation are: Defect analysis,
capacity planning and optimisation, financial reporting, risk management, asset
management and resource planning. Another concept that can be adopted with this BI
analysis is Just-In-Time (JIT) where suppliers deliver ingredients to be developed just in
time. Therefore the production will suffer a minimal amount of idle time as possible
through efficiency of vertical integration. (Loshin, 2003)
Sales channel Analysis is a subset of business productivity however it is specific within
the type of analysis. The sales channel analysis is focused on how to make the best use
of marketing and sales performance. (Loshin, 2003)
Supply Chain Analytics are to do with managing the suppliers and assessing their
performance through internal data. As with the CRM it can be just as important to
manage the suppliers to lower costs and cut completion times. It is possible that some
organisations do not keep track of every supplier and using BI it would be possible to
assess whether a certain supplier is worthwhile. It would be possible to view the
schedules for different products and what affects it can have on the final product
completion. Just as important as knowing how quickly an item can be replenished is to
know how many items are in stock so inventory control can be considered within BI.
The final part of Supply Chain Analytics is to know how long it will take to get certain
finished products to certain areas around the world. For example working out how long
it would take to make 1,000,000 phones in china and to distribute them to the UK.
Having this information it would then be possible to act upon any changes that can be
made and what affect this will have to the lead time. (Loshin, 2003)
Behaviour Analysis is to discover and predict what will happen for a certain scenario.
Looking at historical data about previous buying habits can provide information that
can be used to find out any trends that could be happening. An organisation can then
react upon this information to give them a competitive advantage. Web activity can be
monitored to find out what products are bought when buying another product like
Amazon.com which has an recommendation algorithm that can recommend products
from what other customers have bought (Amazon.com, 2012). Another use of
behaviour analysis is that it can be used to detect fraud and abuse; this would involve
looking at data that is suspicious and can find patterns where they have been
fraudulent. (Loshin, 2003)
An Intelligence Dashboard is what can be used to integrate all of the other BI types
together so that it can be summarised and understood immediately. It is usually
constructed using Key Performance Indicators to develop a conceptual scorecard which
will monitor the performance of the organisation. An Intelligence Dashboard will
overview all of the core competencies and let the viewer know whether the
organisation is failing or succeeding. The overview will tend to allow the user to drill
through the data and show how each core competence is made up. The whole idea of
having the dashboard is to give as much information as possible without getting
technical and having to understand what everything means. (Loshin, 2003)
2.1.4 Evaluation of BI tools
Thompson (2004) claims that when implementing BI tools there is a clear benefit
because 19 per cent of companies have met or exceeded their business goals. This does
not sound like a lot but 60% of the companies have achieved most of their goals. It is
suggested that the problem when implementing a BI tool it is the return on investment
(ROI) which is hard to value whether it has been successful or not. Thompson went
further to detail about the companies that realise a benefit from BI tools, below is a
table that shows that percentage of companies that recognise the benefits.
These figures are based upon the companies already having an ERP already
implemented and therefore these companies did not exploit their data as much as
possible. The results show that the most recognisable benefit is the faster more
accurate reporting which is to be expected as this is what BI is for and the least realised
benefit is the IT savings.
Some of the challenges that Thompson states are not always to do with the tools
themselves. The biggest challenge not to do with the product is “company politics”. If
an organisation does not have integrated business processes that can adopt new work
processes it can make it difficult to know how the business is operating. This means if
the foundations are not correctly implemented the structure built upon them will not
be accurate. Another problem with BI implementation is user satisfaction. Sometimes
the “inability to get users to agree on requirements” (Thompson, 2004) at the
beginning causes the project to be implemented wrongly and therefore cannot be
called a successful project. The final problem that Thompson suggests is the response
Benefit % Companies Realizing Benefit
Faster, more accurate reporting 81
Improved decision making 78
Improved customer service 56
Increased revenue 49
Savings in non-IT costs 50
IT savings 40
of analytics which is usually because of there being problems at the start to do with
architecture issues with the variety of data sources.
When Thompson (2004) collected information about the BI tools from companies he
also asked what the most important factors within product selection were, below is the
list that summarises the priorities.
2. End user ease-of-use
3. Integration to existing applications
6. Ease-of-use for application builders (Thompson, 2004)
This summary will be very useful when looking at CBBIA’s however this project will be
based upon SME’s and the fact they are small organisations that cannot usually afford
large systems. Therefore it will have to take into consideration that price will be a more
important factor or possibly if CBBIA’s are cheap it could be less of an issue.
This project will be using the information that has been gathered about Bi’s to
categorise different types of BI tools and to analyse any features that can be specific to
certain industries. The concept that the BI’s are based upon the cloud should not
change how they work differently from traditional BI tools and using the measurements
above will give a way to compare the different features and potential usage. When
analysing the CBBIA’s it will be interesting to find out which ones could be used for
what type of BI for example some maybe more directed towards CRM analysis.
2 .2 Research Methodology
For a project to be successful it requires a research methodology so that it can be
carried out in a systematic way. To better understand which methodologies will be
most appropriate for this project a number of different methodologies will have to be
investigated. This section will be assessing which methodologies are best suited for the
project. This will be done by defining, explaining and comparing strengths and
weaknesses from a chosen set of methodologies.
2.2.1 Induction vs Deduction
Lee and Lings describe deduction as “the process of drawing conclusions from rational
and logical principles” (Lee and Lings 2008). This means that there cannot be a situation
where a principle is false and a conclusion is true. Deduction is when a theory or
hypothesis is created and then is tested out to see if it can be supported through
evidence which will then be collected. If there is no evidence to support the theory or
hypothesis, they are altered to reflect the findings to create a new one. This will then
be tested and the cycle will continue until the researcher is satisfied that their theory is
correct or cannot be proven to be true. Below is a diagram (Figure 1) showing this
approach. It would also be possible to change the word hypothesis or theory to
problem, this would then be applied research because a problem would have to be
identified first and the result would be a solution to the problem.
Testing of theory through observation of the emprical world
Falsification and discarding theory
Creation of as yet unfalsified covering - laws that
explain past, and predict future, observations
Operationalization - translation of abstract concepts into indicators of measures that
enable observation to be made.
Figure 1. The process of deduction (Gill and Johnson 2010)
This type of research project that start out with a set of problems or issues which the
research will attempt to address can also be called “consultancy-based research”
(Lancaster 2005, p23).
It is also possible to think of deduction as a ‘Top-down’ (Horn 2009, p108) approach
which starts with a theory and ends up with a hypothesis to be tested. Horn also states
that the Induction philosophy can be described as ‘Bottom-up’. This means that instead
of narrowing down to be specific, this method is to end up with a generalised theory
that can be applied to many situations.
Induction follows the opposite direction to deduction, in that the researcher “develops
hypotheses and theories with a view to explaining empirical observations” (Lancaster
2005). A theory which follows the induction method can be based upon personal
experiences and many other factors that could influence them. This means that it is not
constrained to have a base of prior knowledge and is therefore flexible as it can be fit
into many situations. However to fully understand a problem it may be necessary to
use both methods of induction and deduction. Figure 2 shows how they are linked
together throughout the theory or hypothesis creation stage.
Figure 2 Induction and deduction in social science theory. (Lee and Lings 2008)
Collect data in
The diagram in figure 2 suggests that only one research method can be applied to a
study; however this depends on what type of research is being done. There are
occasions when it is necessary to use deduction and Induction on the same study.
Within this project it may prove useful to use induction based upon knowledge already
known and then use deduction to find out whether my assumptions are correct. This
will give a base to start from and will help when writing recommendations based upon
2.2.2 Research Philosophy
Positivism is a philosophical idea where it is believed that if there are universal truths,
they will have to apply to a research project. A researcher who is positivist believes that
a project should be carried out with a set of rules and guidelines in order for the
research to be followed. This can enable certain areas to be criticised and questioned
proving the research to be fallible. Jankowicz (2005, p110) suggests that another name
for positivist could be ’realist’ as this methodology takes into account that being human
means that mistakes happen. For a positivist project to be successful it will have to be
checked in a way that will provide similar results to what has been found. This can be
done by having it verified or drawing on empirical data which avoid the dangers of
stereotyping, myth and superstition (Jankowicz 2005, p110). This means that social
science is not accepted as verifiable because it is not empirically tested, therefore if it
cannot be proved every time it cannot be accepted as a positivist point of view (Lee and
Lings 2008, p31). This philosophy is the most widely used and commonly adapted to
modern science because of the black and white answers that can be reproduced
repeatedly. This is not the way in which the project will progress because it will follow a
set of guidelines that have been set by an established organisation (KPMG Appendix 1).
These guidelines are interpreted by an individual and therefore there is no definitive
answer to the question, only best recommendations. This means if a more experience
researcher did the same research it could very well result in a completely different
Interpretivism is where it is necessary to understand the “differences between humans
in our role as social actors”. (Saunders et al, 2007) This means that there is a different
point of view from each human exists and this is affected by what role they ‘play’ in the
real world. A researcher who uses the interpretivist methodology will be of an opinion
that the positivist way of thought is too rigid in its understanding and that management
is too “complex” to be simply defined in the same way as a physical science. This
project will certainly be of a similar understanding because it will be trying to best fit a
CBBIA for an organisation. This will be a point of view which will be understood from
talking to a member of the chosen organisation. To do the research thoroughly it will
have to be done using a group of employees from the organisation to give a fuller
picture on how to make sense of the “social world” that they exist. (Saunders et al,
Positivism is one end of a spectrum as it is the way it covers facts and what is actually
happening; the other end of the spectrum is normative. This type of theory is to do
with the “right way to do things” (Lee and Lings 2008, p13) and deals with what is
expected to happen and how to make right decisions in a situation. Normative theory
can sometimes help with discovering why a problem occurred or why there is a
requirement for research. However if this project followed only one philosophy it
would not give a true perspective of what is happening or what should happen. This
means that with these two philosophies identified it will be necessary to bring in some
of the other philosophies that could be involved in the project. This project will not be
giving an explicit answer because of the nature of the research question and the results
that will be collated.
Saunders et al (2007, p110) states that pragmatism is the solution to this situation
because it is “unrealistic in practice” to suggest only one philosophy can be applied.
The pragmatist theory suggests that it is relevant to the research question because it is
not constrained by following a certain perspective. The project would be more flexible
adopting this approach allowing more control over what the project can achieve. Doing
the project with a pragmatist approach may not give a definitive answer but it can be
adapted to the direction that the findings will suggest. This will be the most optimist
view as it will be most likely to achieve a positive result.
2.2.3 Quantitative vs. Qualitative
Quantitative data is defined by Lancaster (2005) as data that “can be expressed
numerically or classified by some numerical value”. This means that quantitative data
can be used to support evidence through statistical analysis using graphs and charts to
map the data. This type of data is explicit in that it cannot be questioned because it is
objective; it is done for a particular reason. If a project uses this method to gain results
it means that the project is able to be measured and quantified. The most common
gathering technique of quantitative data would be from a questionnaire with a set
number of answers. Wisker (2008) states that a questionnaire will be used if the
researcher requires a response from a large number of respondents because they can
be counted measured and statistically analysed. This gives the researcher a definitive
answer from the subject and can be quantified on the number same answers.
In contrast qualitative data is where there are “descriptive accounts of observations or
data classified by type” (Lancaster, 2005 p 66). This means that although a qualitative
method would be more in depth and thorough in answering a question it will be to do
with perspective and observation. Wisker (2008) talks about qualitative methods
involved “capturing people’s opinions, feeling and practice, their practice and the kind
of atmosphere and context in which they act and respond”. This method would be
most appropriate for the project because the basis of the project will be following the
techniques that KPMG (Appendix 1) employ to discover appropriate recommendations.
The style that this project will follow because of its pragmatist / applied research
method will require more emphasis on the qualitative methodology; this is because it
will require real world answers to the research question. Using the qualitative
methodology will help constrain the amount of data that will be gathered but will
increase the quality of data. This is required from the project in order to achieve the set
of recommendations as a result. Also this project will not involve data that can be
quantified which is the biggest factor at choosing qualitative over quantitative.
2.2.4 Methodology Justification
After reflecting on the types of methodologies that can be used it has been decided
that for this project it will be required to use a pragmatist approach. This is because the
project will be bias without covering many areas of interest; this is why it will be
necessary to gain different points of view based on the same recommendations. The
Interpretivist view will have to be used at this point to gain information from 'social
actors' which will give a more thorough input into how successful the
recommendations will be. Whist deliberating recommendations it will be important to
use the normative perspective which will give an ideal world scenario of a CBBIA.
Together with these philosophies the project will require a deductive approach so that
in the end a hypothesis can be made about the right CBBIA for an SME. This could then
lead to suggestions and falsification which will be required to give an appropriate
The final stage of the project will be confirming assumptions that have been made
which will require measures and a set of tools to analyse the data. Qualitative data
collection will be required meaning that interviews will have to be done to get in depth
insight on whether the right recommendations have been made. The person to
interview for this part will be important to the investigation because it will either prove
whether the recommendations are useful. The information gained from this interview
will also be biased towards their experience and could vary from person to person.
2.3 Research Tools
This section is to discuss and analyse what tools are available for data collection to be
used within the project. It will concentrate on the design and structure of the research
method resulting in useful data to formulise a detailed recommendation. It is important
to understand which methods of data collection can be used in order to gain data
which is useful and relevant. The result of the research tools section will detail the
method of data collection which will be the most relevant for this project. It will also
include the design approach of the chosen method and how it can be adapted to this
2.3.1 Methods of data collection
There are many methods of data collection and all of them follow either a qualitative or
quantitative philosophy. Typically qualitative methodologies are interview based
because the data collected is more in depth and it generates non-numerical data. In
contrast a questionnaire is typified as quantitative this is because the data gathered
can be quantified and presented in a graphical way. Using one of these methods can be
defined as “mono method” because it will only use one technique in order gather data.
This is opposed to “Multi method” which uses more than one technique towards the
research question (Saunders et al 2007, p145). An example of this would be to do a
survey and questionnaire to gain better results for analysis. However multi method is
confined to one methodology therefore if multi method is used it could only use either
qualitative or quantitative. This means it is possible to have a multi method qualitative
or quantitative study which gives more thorough results but requires more time and
planning. There is a method which can use multiple methodologies and can be very
useful if adapted correctly and this is called “Mixed method”. It can use both
methodologies and it will give the study benefits from the quality from qualitative
methods and the definitive answers from qualitative methods.
For this project it will require a focused approach on the research question to give the
best recommendations. The research will require data from volunteers who are
qualified within the research area to give the most relevant answers. It would also be
important to design the research in a way that will be completely open ended so that
all data gathered can influence the final recommendations.
2.3.2 Different types of data collection
There are many taxonomies of data collection and they can be used in different ways
dependent upon the research question is asking. They are listed below and can be
grouped based upon how they are collected (Lancaster, 2005).
Secondary data collection
Interviews and questionnaires
The data collection types can be grouped by how they are done, for example action
research and case studies are based upon the subject of a particular case.
Experimentation and observation would require a situation or test to be set up for data
to be gathered. Interviews and questionnaires are collecting the opinions from
participants and can differ on a whole many different variables.
220.127.116.11 Action Research and case studies
The first group to explore will be case studies and action research. These types of
research design are similar to each other because they both involve ‘cases’, where they
differ is what they do with the ‘cases’. Lee and Lings (2008) describe a case as anything
you want it to be” because it depends on what the research question is about.
Therefore a case could be a house, an organisation, a person or even a social scene.
Therefore a case study could be based upon anything at all and not confined to a
particular type, as long as the research is thorough. Action research will treat the case
in the same respect but researcher will actively become a part of the research. This
means that the action research becomes more of an experiment or observation to test
reactions to certain conditions. A case study is often used in medical research because
an in depth examination can be done into symptoms and ailments. This can be used to
link certain conditions or to diagnose patients with similar symptoms.
For a medical case study the ‘case’ in question would be the patient and the research
question would be what is happening to this patient (Lee and Lings, 2008).
A case study can use many methods to answer the research question, involving both
qualitative and quantitative research. This is the same for action research but the
method that action research uses is more of a practical method. Lancaster (2005)
describes the characteristics of action research as problem centred participation,
cyclical, co-operation and professional development. This means that for action
research to be successful it will require participation of the organisation involved or
whatever the case maybe. When Lancaster describes it to be cyclical he means that the
research should involve a feedback loop which can be evaluated and changes made in
order to carry out further research.
If time was not limited on this project a case study or action research could be used
because it would give the best most useful results; however the scope would be very
large and would require feedback from all cases. This project will be more like a first
stage of a case study and once some recommendations have been made some further
research would be to ask some organisations to partake in the study.
18.104.22.168 Experimentation and Observation
The next group of data collection techniques to be explored are experimentation and
observation. Both of these data collection techniques are collecting primary data and
are controlled by the researcher; however experimenting is done in a proactive way
trying to determine the cause and effect of certain defined variables. Saunders et al
(2007), defines an experiment as the study of “causal link” between two independent
variables. It is mainly for exploratory and exploratory research in order to answer the
‘how’ and ‘why’ behind a research question.
For an experiment to be successful it will have to be separated out into 2 groups, the
control group and the experimental group. The reason for a control group is to remove
the possible effects of alternate. This means that the control group is required to
explain the findings gathered and ensure that the results don’t occur through nature.
Experiments are often done in a lab and this is so that all of the elements involved with
the experiment can be controlled. Once an experiment is controlled it can then be
classified as internally valid which means that any findings can be attributed to the
experiment and not from any research design flaws (Saunders et al, p 600, 2007).
A summary of an experiment is listed below which shows the different areas that an
experiment can cover:
1. Definition of hypothesis
2. Selections of samples
3. Allocation of samples to groups, control and experimental
4. Introduction of planned intervention to one or more variable
5. Measurement on a small number of dependant variables
6. Control of all other variables.
(Saunders et al, p137, 2007)
As opposed to experiments where they are introducing change to a specific case,
observations can be used to observe the natural world “As Is” (Conway, 2012) and then
a recommended course of action can be taken.
The observation data collection technique is the best way to understand how the world
operates. The main problem with observation techniques is that they require long
periods of time to study, however there are two methods to allow for some control
over time. These are continuous and time frame; time framing is where only a specific
measured time is allowed for example 1 hour and then the research will be based
around the observations made in this one hour, a continuous time period is where the
researcher will determine when enough data has been collected and suitable results
have been obtained. When time has been considered for the research it can then be
decided which type of observer should be adopted. This will vary on the type of
An objective observer type is when the researcher does not partake in the activity and
therefore observes the activity directly without any involvement.
This method requires the participants to know that they are being observed and for
them to act as if everything was normal. This becomes a problem because as soon as
the participant knows they are being observed then they typify their actions. This
means that the data collected will be false because it will be based upon what the
participant thinks they should be doing instead of what they actually do. Once a
participant has settled data will become more accurate as they will become use to a
routine and this is why it requires a large amount of time.
An un-obtrusive observer type is done in the same way as an objective observer but
with one major difference. The participants are unaware they being observed. For this
type of observation it will require approval from an ethics committee to deem whether
it is ethical for the observation to take place. It would be used to avoid the scenarios
where the participant’s actions are typified.
The final type is participant observation and this is where the researcher takes part in
the activity making notes on a day to day basis. A journal is usually kept in this scenario
and will detail what was done throughout the day. This type is usually linked with un-
obtrusive because it is possible the other participants will not know you are a
researcher. This will however require ethical considerations and be submitted to an
After exploring this group of data collection techniques it seems like they would not
suit the project because of the time frame and resources. However if more time was
allocated it would be suitable to suggest that the experiment technique would be
appropriate. This would be useful for a project that would investigate the findings of
this project and applying them to organisations practically.
22.214.171.124 Questionnaires and Interviews
The final group of data collection techniques are both different approaches on how to
ask questions to gain data about the research question. These are interviews and
questionnaires, interviews are aimed at gathering more in depth data which tend to be
Interviews can be adapted for any research question and therefore are commonly used
in research projects. Flexibility also allows them to be fit around the researcher’s time
and can deliver the necessary requirements. Some of the different ways it can be
adapted is the structure, this is also the main difference between an interview and a
questionnaire. An interview can either be unstructured or semi structured, once an
interview has become structured it is classified as a questionnaire. Lee and Lings (2008)
describes the differences between unstructured and semi structured interviews as a
spectrum with ideal interviews being somewhere in the middle.
Figure 3. Interview Structure Spectrum
An unstructured interview is where an interview has no direct questions and should be
used when the topic area is not fully understood. It will give more data around the
subject area but may not give the data required to give a definitive answer. It can be
used in situations where the interviewer may cause an imposition on the subject and
affect the answers making them bias. The biggest problem with a completely
unstructured interview is that there will be a large amount of irrelevant data collected.
It will also be a distraction from the research question and most of the time will be
wasted. This approach can also be called freewheeling due to its nature of allowing the
interviewer to follow up questions and approach from different angles depending on
what answers are given.
On the opposite side of the spectrum there are semi-structured interviews, which have
some elements of a structured interview but are not constricted like a questionnaire.
The most appropriate use of a semi structured interview is when the interviewr has an
understanding of the subject area and more direct questions can be asked about the
research question. From this understanding of the subject area a guide can be drafted
to aid the interviewer. The guide will include some questions that will be required in
order to gain specific data and open ended questions to collect more in-depth data.
(Fenwick, G, 2012)
Unstructured Semi-StructuredIdeal Interview
The idea behind a semi-structured interview is that some sort of control can be
established giving information about the research question and limiting the collection
of useless data.
It is often conceived that interviews are purely qualitative data collection but the
adaptability of using interviews enables the collection of quantitative data. This can be
done if one or more person is interviewed and the same question is asked to compare
results. This would allow for data to be quantified and given as evidence, if found to be
useful. This is compared to a common perception of an interview which should be in
depth and adapt to the subject and what they say.
Questionnaires are the most common quantitative research method and are the
opposite end of the spectrum from an interview. This means that a questionnaire is
fully structured or semi structured depending on the research question. The data that is
gathered from questionnaires is valuable because it can be used in many different ways
depending on the context of the research question. They are the most flexible data
collection technique because they fit the research question exactly and the participants
who are questioned do not have to have previous experience in the research area. This
allows for a wider spread of data collection which may provide different angles to the
research direction. (Lancaster, 2005)
When constructing a questionnaire certain areas have to be considered. This is to make
the questions relevant to the research question and to collect data regarding the
subject area. The subject areas are listed below:
How will the questionnaire be administered? Are the participants known? Does
it affect the research question?
What is the purpose and framework for analysis? Reason for doing the
questionnaire. What other studies are there similar?
How will people respond to the questionnaire? How will people be persuaded
to answer the questions?
When will it be administered? What data analytics will be done?
A questionnaire can be delivered in a many number of ways and it can be specific to
what kind of research is being done. The cheapest way to collect data is through a web
form that has simple questions which are explicit. The data is then quantified and
analysed producing evidence to support a theory, other possible ways are: going out on
the high street and asking, another is postal and many more. Another tangent would be
the audience, who is going to fill out the questionnaire? The data collected could be
affected from a many number of variables like job status, gender, age and therefore it
is worth considering who will give the most appropriate answers for the research
For a questionnaire to be successful it is required to understand what data is required
for the research question to be relevant. If the data collected is irrelevant to answering
the question then time has been wasted and the results have been nullified. To answer
this question, a thorough understanding has to be researched to review if any previous
research questions have already been answered.
Once a questionnaire is deemed reasonable to be considered a decision will have to be
made on how to get persuade participants to fill out the questionnaire. Simply sending
out a web form or a letter does not guarantee that it will be filled out and returned.
Sometimes a cash prize can be given as reward or their names could be put into a draw
to entice people. This is not so much of a problem if the target audience is within an
organisation and management can be used to enforce the importance of the
questionnaire. Some of the reasons people do not bother to fill out questionnaire is
that they can be too personal and if this situation arises it is worth considering an
anonymous questionnaire unless it is important to the research question.
Another important consideration when creating a questionnaire is the timing of when it
should be administered. Sometimes a questionnaire can arise from a situation and the
questions could be directed towards the change or incident that has resulted. If a
questionnaire is done before an event occurs the results of the questionnaire will be
different if done during or after, therefore it is important that timing is considered. It
will also affect how the data gathered can be described and this can be done by
categorising the data appropriately.
The final consideration for a questionnaire is data analysis and representation. There
are many data analysis tools available which can represent the data in different
formats. To determine which tools should be used two factors can help: the amount of
data used and budget available. If there is a massive amount of data collected then it
would be possible to use a data mining tool which will provide very detailed statistics
very fast, these are also the most expensive. However if only a small sample is collected
it can be done using simple tools like spread sheets and OLAP techniques depending on
how the data is stored. However this is also assuming that two or more variables are
connected in some way otherwise these tools will be useless if the questionnaires are
not constructed appropriately from the beginning. (Horn, 2009)
2.3.3 Issues within data collection
126.96.36.199 Quantitative research
There is a failure to distinguish a person’s interactions with the natural world.
Quantitative research does not take into account an individual’s perspective of the
world and how they interact with it. It takes a snapshot of the individual’s time to
answer the questions which could affect the answers given. This means a questionnaire
could be answered differently from the same person due to the timing of the
questionnaire. The data collected from this will not reflect the views on which the
individual may have, which could lead to un-necessary bias to certain answers.
Many participants do not understand the context of a question.
When answering questions it is not uncommon that the participant will not understand
what they researcher is asking because of the lack of context. This could therefore end
up with unintended answers but even if the intended answers are given the data could
be counted as invalid because of the participants’ lack of understanding. A solution to
the answer could be to limit the responses that are given but this would stem the
variety of data collected. Using this method the questionnaire will be controlled but it
does not fix the problem it simply steps around the issue.
There is a disconnection between the research and the real world.
Quantitative research requires control over its data collection techniques so that the
data can be considered as valid. The control over these techniques can distance itself
from how the actual world works and in itself make the data collected invalid. It also
relies on the sense that the participant knows what is relevant to real world life which
would require extensive knowledge. For example a participant may answer a question
based upon their work habits instead of their actual habits.
Analysis of variables can be individual from a person’s perspective.
When the data has been collected and an attempt is made at analysing the results, a
relationship has to be connected between two variables. However, when a researcher
is analysing the data it is not clear whether the relationship has been created artificially
or naturally. This means that the relationship could be individual from the researchers’
perspective and not relevant to actual practical situations.
(Bryman, Alan and Bell, Emma, p167-8, 2011)
188.8.131.52 Qualitative research
It is too subjective.
Qualitative research can be influenced in a number of ways and the relationship that is
gained from talking to participants can cause the data gathered to be affected. Usually
when qualitative research is considered it consists of an interview with open questions
that gradually get narrower. The main criticism for qualitative research is that it’s not
clear on why the researcher has chosen to narrow in on a subject.
Difficult to replicate results
It is difficult to replicate the results gained from qualitative data collection because of
the unstructured approach that is used. It makes the study impossible to re-create
because the investigator is as much a part of the data collection as the participant
being questioned. The factors that are likely to affect the data collection could be
personality, age, gender and because of the unstructured nature of the data collection
the interpretation of the answers is a big factor. Therefore it is difficult to ‘restudy’ and
apply the exact same research method again.
Qualitative data collection results are often generalised and applied to a wider context
where they should not be used. It can sometimes be applied but it is only valid if the
research is similar to a specific case. However it has to be adapted to every case in
order to be successful. This does not always happen and generalisation happens a lot
based on qualitative research. To be able to generalise statistics can provide a more
definitive answer because of its wide variety of data sources.
Lack of transparency
Qualitative research is sometimes difficult to understand how the investigator actually
did some parts or how they arrived at the study’s conclusions. An example would be
not explaining why a certain person was chosen for observation or an interview. This
information would be important if someone wanted to adapt the research for a
particular case study. In order to apply the same methods and understand why the
conclusions are as they stand. (Bryman, Alan and Bell, Emma, p408-9, 2011)
2.3.4 Choice of data collection method and justification
After studying the different types of data collection methods it can be determined that
some would be incorrect for the research question. For example, considering the time
and resources that would be involved, observations and experiments cannot be done
because for this research question it would require participation of many SME’s
particularly from different industries and even then it would have to be generalised. As
opposed to this secondary data collection will be used in the form of a literature review
which will involve previous case studies and action research. However to get the most
useful data in answering this research question, interviews and questionnaires will have
to be used. This is because they will be providing the project with enough detail to gain
a conclusion from.
This project will be using the ‘mixed method’ of data collection because it will be using
both qualitative and quantitative data. Questionnaires will provide the evidence to
back up any provisional recommendations that are made. On the other hand an
interview will provide in depth detail that is required to alter any misconceptions that
could have been made from a questionnaire.
To gather data from the CBBIA’s it will be relevant to use a questionnaire to compare
the features that each CBBIA shares. The data collection will have to be from the same
perspective and this will provide integrity to the data. Therefore the questions will have
to be planned carefully as to not favour any particular CBBIA. It will be important to
maintain no bias towards any particular CBBIA for the data to be valid.
An interview will be created to provide any recommendations that can be sought out
through the questionnaire. The participant will have to be someone with particular
knowledge in CBBIA’s or Business Intelligence tools in general so that assumptions can
be clarified. This will provide more detail on provisionally recommendations made
through the questionnaire and give quality details that can be easily missed through
Proceeding with the mixed data collection method will help prevent the issues with
data collection that have been identified. As a result the project will benefit from all of
the advantages from both qualitative and quantitative methods. However without
careful planning and consideration of the issues all of them will be relevant for both
data collection methodologies and this will provide only invalid data and poor results.
3. Research Implementation
This section will describe how primary data will be collected and the design of
collection methods. It was determined that a mixed method of data collection will be
the most appropriate for this investigation. This will allow for the comparison of
CBBIA’s and results from interviews to be analysed together, allowing for a better
recommendation to be made. This section is important so that the results can be
replicated if further research is carried out. The interviews will allow the same
questions to be asked but it will determine who is being interview as to what answers
are given. The questionnaire because it is a comparison type data collection technique,
can be replicated with many CBBIA’s.
3.1 Aims of the Interview design
The interview design will be a semi structured approach and will allow for maximum
freedom when answering. However a certain number of questions will help achieve
similar results which will allow for a better comparison. The purpose of the interview
will be to gather data about how popular the cloud is amongst SME’s, whether a BI tool
is already in use (if so how), what are the most important factors for BI’s to have within
The answers will be asked so that as much data can be gathered without being
constrained to a particular area, the idea being everything they say could be of
importance. It will start with what their business does and how it can relate to BI. Then
the final question will be a direct question about what is their opinion of the cloud and
using a BI tool on the cloud that is not owned by them.
3.2 Aims of the questionnairecomparison chart data collection
The Questionnaire will be directed to produce specific results in order to compare
variables within a chart for data collection. This will help with analysing the results and
finding any patterns that may exist. It may show that certain CBBIA’s have a specific set
of features which appeal to certain types of BI. This would therefore show the
differences between them and what industries may be more interested in CBBIA’s.
3.3 Justification of research implementation aims
The reason for the research to include both methods of data collection is so that the
project will be more aligned to what is happening within industry. The whole idea of
the project is to provide recommendations on CBBIA’s and justify the choices that have
been made. The interviews will give a background into how BI’s are currently
implemented and for what type of organisation. From this it would be possible to
match up the findings from the questionnaire to the current BI use. Using both
techniques allow more in depth look to how CBBIA’s can be used and adapted.
From the interviews it will also be possible to gather a perspective of what level of
confidence is there within CBBIA’s. This could be interesting to find whether there is
longevity within the new concept and for further research.
4. Analysis and Results
This section is to analyse the data collected to better understand what the results
mean. The interviews will be analysed per subject and the answers will be related into
how they are useful for the project objectives. The questionnaire will be justified and
summarised to describe the different CBBIA’s features and capabilities. This will then
be synthesised into useful knowledge combining both results and how they can suggest
4.1 Analysis Method for questionnaire
The questionnaire is constructed in a way so that results can be easily compared and
differences can be highlighted. Each variable will provide data that will can potentially
define a CBBIA and set them apart from the others. Each variable has been adapted
from Thompsons’ (2004) list of priorities that SME’s will look for and added to for the
purpose of the investigation.
Functionality – This refers to what features have been made available to use by the
service provider. It is where the more established CBBIA‘s will allocate their resources
to make it unique. If a feature is important enough to an organisation it could make
more attractive to buy. However claiming to have a lot of functionality does not say
how well it does the task. For Example a CBBIA may claim to use dashboards but there
are only a select few of defaults to choose from.
End user ease of use – This variable is to measure what features will help an end user to
use the CBBIA. It will of course depend on what level of computer literacy the end user
is at and therefore it will only look at what features have been implemented to assist as
much as possible. The type of features that will be covered here will be interactivity
Integration to existing applications – This variable will be how compatible the CBBIA
tool is with existing systems. Without being able to test the CBBIA‘s on actual systems it
will only be able to see what platforms they support and what programming languages
can be used.
Price – The pricing if possible to obtain will be considered per licence fee.
Performance – This is an area that should be considered but due to the fact of it being
important to SME’s to get the best performance for their money. However using cloud
computing the performance of the CBBIA will be based upon the service providers own
computing power, this means the only factors concerning performance will the size of
the data store and bandwidth issues, which is the same for all cloud based applications.
Ease of use for application builders – This variable is considering what third party
applications can be made for the CBBIA platform. If many programming languages can
be considered the more flexible the platform is. However it is worthwhile to note that
CBBIA’s are designed for end users and to make BI available to managers and not just
programmers, or Database admins.
Mobile enabled – This is to see whether CBBIA’s are in general mobile enabled which
means that the BI can be used on the go. This can be useful to show BI with data and
what it can do when presenting for example.
ETL capabilities – This variable will identify the facilities that are available for the
Extract Transform Load process. If a CBBIA has its own ETL tool it can be useful to
implement data into preloaded designs and make the data more manageable.
Data source capability – This variable is to discuss what type of data source is
supported by the CBBIA. For example it could support SQL server and Oracle database
but no others.
4.1.1 Interpretation of the questionnaire
Using the framework above it will be able to compare results more easily however
because the data collection will be done through investigating the CBBIA’s first hand it
will only show one point of view. If this project could be done again it would be
interesting to test the CBBIA’s at a few organisations using a dataset on their systems.
This would give feedback on how the CBBIA could be used for an SME directly and the
functionality that would be relevant for each particular organisation from certain
CBBIA’s. However on this project it is only possible to give an overview over the
possible suggestions for SME’s.
It would also be possible that using the framework will favour certain CBBIA’s if they do
not have certain functionality. For example if a CBBIA does not have the functionality to
do ETL or a mobile app. This could give at first glance an advantage to the haves as
opposed to the have not’s.
A possible solution to this problem would be to give the questionnaire to a few other
subjects and let them answer the questions themselves. However this could lead to
other bias effects and persuasions, possibly due to their background and previous
experiences in the social world.
4.2 Analysis Method for interview
The interviews will be mainly unstructured with some questions to guide and prompt
the subject to enable them to cover areas that have not previously been identified. The
purpose of the interviews is to gather information about how they use BI within their
organisation and whether they think that the cloud is a platform that could be used in
the future within their organisation.
4.2.1 Interview with Alan France (Appendix 2)
Alan France is an Operations director for a company called Idhammer Systems Limited
(http://www.idhammarsystems.com) the company specialises in identifying inefficient
manufacturing systems and creating solutions. His company works in a similar way that
a BI tool should work, identifying how systems should work and then finding out how it
is actually working. His experience does not reflect BI directly but he knows what SME’s
will be looking for when approaching software development and whether the cloud is a
suitable investment for a small business. He could help to identify what SME’s are more
likely to choose from when looking at Thompson’s (2004) priorities list.
4.2.2 Interview with Max Willis (Appendix 3)
Max Willis is an optimisation manager for Arla foods plc and specialises in supply chain
optimisation (http://www.arlafoodsuk.com ). The Enterprise Resource Planning (ERP)
system they use for the maintenance is SAP so there is a chance that the interview will
be bias towards SAP systems. To help alleviate the bias the questions for Max will have
to be less specific about what BI he uses but the application from which BI is used.
Max’s role within the organisation will enable me to find out what he does, but it will
not allow me to go in depth with how BI is directly used within Arla. Mainly because he
is not high enough in the organisational structure, ideally the person to interview
would be the man who picked the BI tools. However nothing will be discredited from
what Max has got to say because it will give me insight into how a big organisation can
Interpretation of the interviews
The results from the interviews will be filtered so that only the relevant information will
be used for the investigation. The relevant information will be determined on the
quality of the interview, if it is all relevant then it can all be used, if not it will be filtered
out. The only problem with this method is that relevant information could be different
between individuals and therefore it has to be taken into consideration that the
information gathered could be adapted to the project. This can give the interviews a
different perspective because they can be used in different contexts.
If this project could be repeated, more interviews with SME employees should be done
to gather results of what are the most common uses of BI within SME’s if they use
4.3 Data collection analysis
From the interviews and questionnaire being completed it is possible to analyse what
has been collected and how it can be used. The subjects of the interviews are directed
to find how BI can be used within industry and the questionnaire/survey should show
what each CBBIA can provide for Industry. The analysis will have to take into account
that it will only be from one perspective and if the project is repeated it could get
4.3.1 Quantitative Analysis
Each of the CBBIA’s tested had very similar features and this has become a common
theme because they don’t want to lose business because of some unique features. The
three applications that have been tested were chosen because of their flexibility and
popularity. They do not have a target audience in specific and can appeal to
organisations of any size this is because they use the cloud and can scale accordingly
with data sizes. A larger organisation may not be as interested in hosting their data on
the cloud just yet but with some of the CBBIA’s it is possible to use your own data
source anyhow. This is an example of the flexibility that has been enabled to entice
new business. The price could not be established for all CBBIA’s and therefore it cannot
be compared fairly which means the analysis will be based upon functionality and use.
184.108.40.206 BIRT onDemand 360 (Appendix 4)
BIRT onDemand 360 is a product which has been developed from previous BIRT
products which revolve around reporting and analytics. BIRT onDemand is the next
development that incorporates all of the other products and puts the service on the
cloud. It is an open source application which means that add on’s and templates can be
made and downloaded from a marketplace. It uses the Eclipse framework to allow
developers to create dashboards and reports; they are then uploaded to the cloud
space and used.
Functionality – Once BIRT OnDemand has been purchased it is immediately available
because they have bought the usage of Amazon cloud space which is scalable. This
means that when you upload reports and dashboards they will be stored on Amazon
servers not BIRT’s. If a business would like to upload their data to the cloud they can
hold it on Amazon’s Relational data store which would make connecting to reports and
dashboards easier. It is possible for a business to set up job scheduling which can
synchronise personal databases with the cloud’s this will keep the data up to date if it is
required. However if simple OLAP analytics are to be done it is possible to just upload a
cube and provide displays for only that data.
BIRT onDemand also has applications on mobile devices written their native
programming language for example they are available on Android, IOS and Blackberry.
A connection has to be established with the cloud but once enabled all of the same
reports can be accessed and created from the mobile device. This means that BI
analytics can be used on the go, depending on internet connection.
Once all it has all been set up it is possible to save some of the views of dashboards and
reports to PowerPoint, pdf, excel or word compatible files. This can allow snapshots of
the data analytics and can be useful to show the performance of the business to
anyone who cannot access the system.
End user ease-of-use – The user experience is like most BI tools where if certain areas of
results are selected it will give more details about the results. It depends what data is
already available to be used for example geographic location. If data about a
customers’ location is held then it would be possible to map them to see where most
customers are from. It would then allow the user to click into what country, county, city
town, street all depending on how granular the user requires.
Integration to existing applications - BIRT onDemand has programming roots and
because of this it makes it extremely flexible, it can be adapted with Java, c#, C++,
Visual Basic and many others. Therefore if a small business which has specific built
systems BIRT can interpret the data easier from the same method being used currently.
There are also some applications on the market place which have been built to extract
data from CRM’s and other data collection packages. It is also worth noting that BIRT
does not support SAP integration so this would have to be adapted from the data
Ease of use for the application builder – BIRT onDemand is built on to the eclipse
toolset which gives an application builder a wide range of functionality that they can
implement. It can understand a many different programming languages which allows
for easier integration and there is a community around BIRT because of its open source
nature which means problem solving can be shared. There is an in depth help feature
that encourages more programming to be done upon the onDemand platform. Using
the onDemand platform allows Mashboards to be created which are unique to BIRT.
However they are the same as a Dashboard but with different unrelated data being
able to be shown.
Mobile enabled – It has enabled mobile use through having apps on all major
marketplaces for example, Google Play, Apple’s Appstore and Blackberry’s Appworld. It
is written in native mobile app language (unique for each one) and can use the
onDemand features to create reports and view dashboards.
ETL capabilities – The Extract Transform and Load (ETL) capabilities are somewhat
limited for BIRT onDemand because it cannot be done on the cloud. The ETL process
will have to be done before hand using a database expert to sort the data out
appropriately. BIRT however offers its own software to do this but it will have to take
time in order for someone to learn how to use it. The software is a drag and drop
interface which gives the user plenty of options but it will not be a quick process.
Data source capability – It will be able to synchronise the cloud databases with the
original data source if required but there is no comprehensive list of what data types it
can support. Presumably it would read any Relational Database Management System
(RDBMS) that support JDBC 2.1 and SQL -92 but this is not written down anywhere.
Social Media Interaction – This is a new concept to BI tools and has only recently been
adopted by some BI tools. However BIRT will have social media functionality built in the
near future but currently does not support it.
220.127.116.11 Jaspersoft Live (Appendix 5)
Jaspersoft is a company that aim to improve old world analytical systems where it
required large custom built systems to handle. To do this it is utilising new technology
that includes the cloud and mobile BI. To use Jaspersoft products before the cloud a
server had to be used to hold the data and all of the dashboards/reports. They utilise
the cloud in the same way but the server will be on the cloud instead. The way in which
Jaspersoft is different is that it communicates directly with the cloud storage supplier
to provide the analytics. This means that the cloud environment that Jaspersoft use for
BI is completely separate to the data. It does work in a similar way to BIRT however
Jaspersoft are more of a solution provider and have sponsored cloud storage providers
called RightScale, Vertica and Telend.
Functionality - Jaspersoft Live is available immediately however it is the data which will
need to be cleansed and uploaded onto the cloud which will take the time up before it
is used. It is one of the advantages of cloud computing, the services are either on or off.
Like BIRT it is open source software which means that dashboards can be created and
uploaded onto the Jaspersoft platform. The unique selling point for Jaspersoft Live is
that it supports multiple tenancy. This is where a collective group of organisations that
share resources can use the same system but one business will not be able to see the
others data. Considering that all of the data would be on the cloud it would save money
on the fact not every business would need an account, it could be purchased as a
On the reporting side of Jaspersoft Live is the reporting functionality, because it is open
source it can reports can be created separately from Jaspersoft, however it is possible
to create ad hoc reports using the system. For a user’s perspective it is good news
because it would make simple reports easy to create or change to how is required. It is
also possible to set up a schedule of reports for example setting up a months end
report, or a monthly performance report.
Another good reason for the application being open source is that it allows a
marketplace and community to be created. Jaspersoft have a marketplace where
integration tools can be downloaded.
End user ease-of-use – Jaspersoft Live allows the user to create ad hoc reports and view
dashboards that have been created for them. The interactive dashboards allow the
user to investigate the figures further showing what data has been used to create the
overall figures. For an end user it is very simple to use and for a power user it is easy to
Integration to existing applications – It is possible to integrate Jaspersoft into existing
systems because of its open source nature. There is a marketplace that can assist in
capturing the data and uploading it to the cloud. There is also SAP integration which is
something that BIRT did not have. Therefore if a company has a SAP system and no BI
system in place then Jaspersoft can be considered.
Ease of use for application builders – For an application builder Jaspersoft Live will
provide tools to be able to create dashboards for the data. However there is also a
function to create complex reports and dashboards on the platform itself, providing the
connections have already been established. It would make sense for the application
builder to use eclipse to design these dashboards but to make a general dashboard it
would be quicker and easier to use the platform. Jaspersoft have really created an
application that can be easy to set up.
Mobile enabled – Jaspersoft Live is available to download for the iPhone but has not
been adapted to the iPad or been developed for Android devices. This can be limiting
for on the go BI but it still save money through not having to own a server to maintain.
ETL capabilities – Jaspersoft have made their own ETL application which can set the
data to how it is required for their BI tools. This can still be used for their cloud system
before uploading the data to the cloud server. This can be useful and save a lot of time
when cleansing the data numerous times. Jaspersoft have also made native
connectivity to some ERP’s like SAP and Salesforce.com which means that if those
systems have been used, the Jaspersoft software will recognise the data format quickly
Data source compatibilities – Jaspersoft have published a dataset support list which
gives an overview of all the supported formats required for Jaspersoft Live. The list
details that it supports any JDBC 2.1 or SQL -92 relational database which is important
when using cloud servers and understanding which ones can be used.
Social Media Interaction – There is no social media interaction at the moment but it
could happen in the future because it seems to be the direction that BI is headed. This
means sharing the dashboard experience with staff through a social method.
18.104.22.168 MicroStrategy Cloud Enterprise (Appendix 6)
Microstrategy is a company that has an extensive portfolio of clients which has
partners that include IBM, Oracle, Sun, HP and Teradata. Their main product is the
MicroStrategy BI Platform which is an integrated approach to BI and has a number of
products that can process the different areas. Their cloud venture is the standard
version of MicroStrategy BI platform on the cloud, which provides many automated
processes and requires just a data source to use the platform. The platform can directly
link with Facebook and seems to be an adventurous development because it is
combining Facebook with business.
Functionality – Microstrategy Cloud Enterprise is the only CBBIA that identifies 5 BI
styles that the platform provides. The 5 areas are: Enterprise Reporting, Cube analysis,
Ad Hoc Query and Analysis, Statistical Analysis and Data Mining and finally Report
Delivery and alerting. These 5 areas provide very powerful analysis tools that require
minimal effort to use. MicroStrategy’s systems have more integrated features but have
more control over their systems than other CBBIA’s. The application is not open source
which means that it loses the flexibility from a developer’s point of view. This does
mean that the application has to provide more to convince potential customers. If
MicroStrategy’s cloud BI platform is used it would be because the client would like a
system that works and will not have to require add ons to collect data from their other
To use MicroStrategy’s cloud platform once bought it will take 48 hours to setup. This
means that it is not as quick as the other CBBIA’s to set up but it will provide an
integrated BI service which concentrates on the strength of its analytics and the
diversity of displays.
There are two ways in which to connect to data when using MicroStrategy’s cloud BI
platform and they are: connecting to an existing database or by uploading your data to
a cloud server and using that, however it has to be compatible with the MicroStrategy
platform. The platform also incorporates ETL which means that it is possible to do
simple ETL after uploading the databases; however it would be better to provide ETL
End user ease-of-use – MicroStrategy cloud is designed for the ease of use for end
users, therefore to use the platform it does not require much technical knowhow to
use. It would still require some learning to use all of the functionality and make the
best use out of the dashboards. For example it has interactivity to investigate further
into situations and forecasting to predict what will happen if a certain variable changes.
Integration to existing applications – It is possible to integrate existing databases with
the MicroStrategy platform and these databases can come from a number of sources. It
would be possible to integrate MicroStrategy directly with SAP systems which can in
turn update the cloud database. For a non-SAP system it is possible to just schedule a
job manually from existing servers.
Ease of use for application builders – For an application builder it might not be suitable
because it is a contained system. This means that it cannot be altered on the cloud;
however it would be possible to create dashboards and reports using their platform. It
is designed this way so that it does not require a technically adept person to use it.
Mobile enabled – Microstrategy cloud enterprise has apps on the Appstore, Play and
Appworld. These apps allow the user to view dashboards and reports that have been
made but not create new ones. It even downloads the cubes for offline analysis which
is something that the other CBBIA’s do not currently do, however it does take it away
from the power of the cloud.
ETL capabilities – As previously stated the ETL can be done on a basic level through the
cloud platform to help with specific analytics.
Data source compatibilities – There is a long list of compatible data source’s available
for MicroStrategy cloud enterprise some of the most notable sources are Teradata,
Salesforce.com, Microsoft SQL, IBM DB2 and Oracle. There are however many more
and it can even read text files.
Social Media Interaction – MicroStrategy have really spent time making use of social
media to share information and dashboards. Their cloud personal service even allows a
user to be able to login through their Facebook profile.
22.214.171.124 Quantitative summary
From investigating the different CBBIA’s it seems like the MicroStrategy cloud platform
is the most robust and can deal with most database systems. This means that it could
appeal to more established organisations that have big ERP systems like SAP
implemented. The fact MicroStrategy have made a platform on the cloud gives the
same opportunity of BI technology to SME’s who do not have ERP systems
implemented. Therefore an SME is more likely to use their BI platform because of their
reputation which equates to less risk and more chance of a successful implementation.
There is a difference between MicroStrategy’s strategy compared to BIRT and
Jaspersoft. They are very similar and very lightweight BI platforms which are designed
to be customised to a system in a specific context. They can achieve more than they
claim but allow companies to figure out what they can use with their own systems. This
helps companies get what they want from BI without having to commit to buying all
systems that might not be used.
The difference between BIRT and Jaspersoft is that Jaspersoft can offer Multi Tenancy
which shows that a small lightweight CBBIA can support BI for business groups and not
just single organisations. The idea that both platforms have a marketplace to download
add-ons is something that can really be useful for companies. It could lead to a
company using this service because of a specific add on to their system. Through the
BIRT marketplace a Facebook sharing app has been created which shows that if there is
demand for a particular add on It can be made, this is good news for SME’s looking to
utilise cloud based business intelligence.
4.3.2 Qualitative Analysis
Each interview done was useful to learn what some people in industry understand what
they can use BI for and whether it is being used for SME’s. The reason cloud based BI’s
might become more successful is that they can give business intelligence to smaller
organisations that cannot afford large ERP systems. Alan France deals with the
efficiency of machinery and uses “Business Productivity Analysis” as is defined by
Loshin (2003) to show where a company may be losing money and prove a solution to
the problem. Whereas Max Willis identifies that Arla Foods use “Supply Chain
Analytics” (Loshin, 2003) and “Sales channel Analysis” (Loshin, 2003).
126.96.36.199 Interview with Max Willis
From the interview with Max Willis it was interesting to gain knowledge about what
Arla use Business Intelligence for. He claims that there is a department within Arla that
specialise in gathering information about their strengths and competitors weaknesses.
Therefore they use BI to gain a Market Advantage which is a similar type of BI to
Loshin’s (2003) sales channel analysis. Max also specifies that he personally uses SAP to
collect data to monitor the maintenance of machinery in the dairy’s. His job is to
optimise the process and increase productivity where everything is measured in pence
per litre. This could be classified as supply chain analytics (Loshin 2003) attempting to
enhance productivity. He also mentioned that he uses an add on to SAP to improve the
scheduling of tasks. It uses the data from SAP to calculate the best possible schedule
for planned production.
How is it relevant
This interview was useful to know how Arla use BI in certain areas but not in others, it is
possible that high up the chain of command for example a CEO would use some overall
analytical tool. However Max did not know, it was useful to know that Arla put effort
into gaining a competitive advantage from the systems which an SME could not do
without employing a lot of resources. It is interesting to know that a company like Arla
have third party add ons that use SAP data already. It could mean that CBBIA’s can be
used for larger organisations that do not yet know how they can use their data.
188.8.131.52 Interview with Alan France
From the interview with Alan France it was possible to gain his perspective of what
SME’s might think when approaching implementation of a CBBIA. He uses a method
called Overall Equipment Effectiveness (OEE) to find where loses are being made within
a manufacturing plant. This requires some software that can calculate the problematic
areas with certain machinery. This is similar to Loshin’s (2003) Business Productivity
Analysis which defines that a BI tool can be used to improve the business by some
means. If a BI tool was implemented to do the same thing it would require data
collection from the machine and many different areas, but it would mean that
consistent reports could be made every week to monitor progress. Alan also identified
that SME’s number one fear when implementing any project is the fear of failiure. An
SME will not be willing to take a big risk which also means that they will not be willing
to spend a large amount of money. Alan also stated that there are a lot of companies
that specialise in extracting data for and organisation. He believes the reason for this is
that when implementing a system extraction of data is not always considered. When
asked about cloud computing he says that while at the moment he does not use any
cloud data storage, the company does have servers and data spread around with some
of its client, as well as some externally hosted data. It would make sense in order to
keep control of their data, cloud storage could be employed.
How is it relevant
The interview was useful to see what is most important for an SME when implementing
a project and also how his business could also make use out of business intelligence. It
is also interesting how he mentioned that some companies do not consider data
extraction when making their systems because of cost. This means that CBBIA’s could
be highly useful for SME’s that are in this situation.
4.4 Analysis Justification
The analysis that has been done has given some very interesting insight into how
CBBIA’s can be used and how SME’s might be interested in using CBBIA’s. However the
project has not followed the plan because initially it was to follow the pragmatist
approach where bias would be considered and therefore a number of viewpoints
should be considered. A couple of interviews have given two points of view which may
not be enough to fully cover the subject but it would have been interesting to interview
a few different areas. The interview questions were however open ended to give
collect as much data as possible to show their point of view clearly. It is important to
understand the position of the social actor and what factors have affected their view.
The questionnaire was important to show the differences between CBBIA’s and was
designed to give an overview of what their unique features are if any. From the analysis
it was difficult to answer some of the questions because they are not clear from a
user’s perspective on what has to be done to use the CBBIA’s. Therefore to a non-
technical point of view it would be a bit daunting to approach any CBBIA because of the
way they are presented. MicroStrategy is the most friendly to approach because it
offers a free personal cloud CBBIA to try their systems using your own dataset.
It is also fair to say that there is bias within the interviews because Max Willis uses SAP
and that will mean his answers will reflect that, Alan France come from an organisation
that has adapted systems created in c# and SQL Server which gives him the opposing
view the Max in that respect. Also the questionnaire data was collected from one
perspective which does not allow for falsification of results. However the analysis will
still be valid but it is something that has to be considered when approaching any
5.1 Understanding CBBIA suitability for SME’s
From investigating what CBBIA’s are capable of, it has been possible to suggest whether
they are suitable for SME’s. From understanding what types of BI analytics there is,
Loshin’s (2003) definitions are suitable and give more details about the functions that
BI can perform. However it is evident from the data collected about the CBBIA’s that
they can all be used in the same way but they require different methods of achieving
the same goal. This could mean that Loshin’s description of BI is outdated and that all
BI tools can be used to give the same results.
One of the differences between the CBBIA’s is the ETL method, which can either be
provided by themselves or done in-house. This is one of the ways that separates large
organisations from SME’s, as they would already have an ETL method built through
their ERP provider. The tools that are provided for ETL will be unique to the CBBIA
provider and may have to be done before uploading onto cloud servers. This will
probably be the longest process of setting up a CBBIA; however it can be used
One of the most important features of a CBBIA is ease of use from and end users
perspective. The interface has to be designed from a business user’s perspective
otherwise the application would be useless and easily forgotten about. Turban (2011)
defines BI as “BI’s major objective is to enable interactive access to data, to enable
manipulation of data, and to give business managers and analysts the ability to conduct
appropriate analysis”. All the CBBIA’s provide an easy use experience but it is down to
the designers interpretation of requires as to how easy the interface will be able to
manipulate. In the interview with Alan France he claims that most IT projects are
deemed failures because the client does not know what they want in the first place.
Therefore it would make sense to give the client the ability to make the decisions on
the dashboards and reports. The way BIRT and Jaspersoft applications are designed
suggest that without having some technical ability they cannot be changed so much
and that there will always be a need for a technical dashboard designer. MicroStrategy