The big data strategy using social media


Published on

The concept of Big Data emphasizes the use of the complete data set to analyze process and predict various phenomena in the business world. This document describes the business uses of Big Data and outlines a Strategy for implementing Big Data analytics for Social Media

Published in: Technology, Business
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

The big data strategy using social media

  1. 1. The Big Data Strategy using Social Media
  2. 2. Table of Contents 1. BIG DATA...................................................................................................................................... 2 1.1 Impact of Big Data ...................................................................................................................... 2 1.2 Why a company should implement Big Data? ............................................................................ 2 2. OBJECTIVE ................................................................................................................................... 3 3. BIG DATA ANALYTICS FOR SOCIAL MEDIA ........................................................................ 3 3.1 Barclays and Adidas - Customer sentiment analysis .................................................................. 4 3.2 ING Direct – Using customer feedback for customized product offerings ................................. 6 3.3 DreamWorks’s - Tracking PR events and promotional campaigns ............................................ 7 3.4 TD Bank – Social media for rapid customer Service .................................................................. 8 4. IMPLEMENTATION FRAMEWORK .......................................................................................... 9 4.1 Information flow ....................................................................................................................... 10 4.2 Operating model........................................................................................................................ 10 4.3 Big Data implementation process ............................................................................................. 11 4.4 Future Scalability ...................................................................................................................... 13 5. IMPLEMENTATION PROCESS................................................................................................. 13 5.1 Process Map for social media systems ...................................................................................... 14 5.2 Implementing Social Media systems using specialized products ............................................. 14 5.3 Implementing Big Data using Hadoop...................................................................................... 17 5.4 Implementing Big Data analytics systems using packaged products ........................................ 19 6. RISKS AND MITIGATIONS ...................................................................................................... 20 8.1 Security Risks ................................................................................................................................. 20 8.2 Analytics’ cohesion with business goals ......................................................................................... 21 8.3 Managing complexity and Storage capacity ................................................................................... 21 7. SUMMARY / CONCLUSION – .................................................................................................. 21 7.1 Focusing on Big Data opportunities.......................................................................................... 21 7.2 Recommending a Strategy for implementing Big Data systems: ............................................. 22 Appendices............................................................................................................................................ 24 References ............................................................................................................................................. 32
  3. 3. 1. BIG DATA 1.1 Impact of Big Data It is not an understatement to say that data has helped many organizations make rational decisions and take calculated risks in the recent past. Data has helped managers to focus their efforts on specific portfolio or products or customer group to make organizations profitable. For a long while, data used for these analyses were usually samples from a larger dataset analyzed by statistical methods to paint a picture of the entire group. Now, the concept of Big Data emphasizes the use of the complete data set to analyze process and predict various phenomena in the business world. It is believed that these decisions are more accurate if appropriate system and analysis tools are employed. According to IBM, we create about 2.5 quintillion bytes of data every day and 90% of the entire amount of data available in the world has been generated in the past two years. The three V’s namely Volume, Velocity and Variety are believed to be key attributes for a Big Data system to produce trustworthy results. It essentially means that the Big Data systems should have capacity to process high volumes of variety of data both unstructured and structured data at a very high speed to produce desired results. 1.2 Why a company should implement Big Data? In today’s scenario, companies are threatened by low margins, uncertain economic outlook, changing trends, and new entrants. It has become very important for companies to utilize the data they have and study their customer behavior effectively and efficiently in order to retain their competitive position in the marketplace. Appendix 1 analyzes the factors affecting a firm. Today, customers care more about convenience than service provider. Furthermore, customers are willing to provide more information if firms can then provide better personalized services to them. The leading firms should offer customers “the right offer at the right time” by leveraging customer data gathered from stores, websites, social media and other sources, thus creating one integrated multi-channel experience.
  4. 4. 2. OBJECTIVE Big Data analytics can significantly improve a firm’s ability to improve customer segmentation, provide personalized services, build brand-awareness trough social media and provide instant offers facilitated by real-time analytics. To do this, a firm needs to utilize the data that it already possesses and merge this data with newer data available from external parties or social media platforms. With deep analysis of data, the firm can develop a segmentation strategy that would identify each group of customers on relevant attributes and create effective loyalty programs that would incentivize its customers to stay with the firm and even recommend it to others. Various businesses have already successfully implemented these programs leading to higher profitability and a sustainable long-term advantage. These business cases will be analyzed to understand how their chosen strategies can be applied to any firm to drive business value. The document will focus on how a firm should build its data capability to take an advantage of opportunities provided by Big Data systems. 3. BIG DATA ANALYTICS FOR SOCIAL MEDIA Social media is the engine that has transformed the web from being a one-way, information tool to a two-way collaboration mechanism. In the world of social media, customer preferences for products or services are influenced by ideas, perspectives, insights and experiences provided by other users. This is achieved through peer reviews, referrals, blogs, tagging, social networks, online forums and other forms of user-generated content (Oracle, 2009). Social media Big Data analytics provides a measurable means of gathering, processing, analyzing and delivering business intelligence from social media channels. The benefits of having a social media analytics program include micro-target marketing, brand protection, customer engagement and loyalty, and promotion feedback (Todd Nash, 2013). Social media has emerged as an influencer of brand awareness and loyalty, as well as a powerful catalyst for community building, although with new compliance implications. Firms can leverage social media as an analytics engine. Using social media Big Data to power analytics applications, firms can better understand customer preferences and align communications, products, sales strategies, distribution channels, and customer service strategies to facilitate better individual customer experiences.
  5. 5. Utilizing social networks and link-analysis techniques can also assist in the discovery of relationships between accounts, customers, households, groups, rings, and institutions, and lead to more in-depth customer knowledge. By leveraging advanced analytics, firms can also develop more sophisticated models to understand the stages of customers’ lifecycles, providing differentiated customer experiences that are relevant to them (Deloitte, 2011). The Big Data analytics of social media information can help in understanding sentiment drivers, identifying characteristics for better segmentation, measuring the organization’s share of voice and brand reputation compared with the competition, determining the effectiveness of marketing touches and messages in buying behavior, using predictive analytics on social media to discover patterns and anticipate customers’ problems with products or services (TDWI, David Stodder, 2012). In 2013, the social media landscape has evolved far beyond the traditional channels to include countless data resources, including but not limited to:  Facebook, Twitter, LinkedIn, Google+, etc.  Review sites, like Angie’s List, Yelp, Urbanspoon, TripAdvisor, etc.  Blogs and news sites that include/encourage comments  Video and photo sharing sites, like YouTube, Flickr, etc.  Search engines, such as Google, Bing, Yahoo and others (Todd Nash, 2013). The following major categories illustrate how various companies use social media Big Data analytics for specific purposes. 3.1 Barclays and Adidas - Customer sentiment analysis Sentiment Analysis: Used in conjunction with Hadoop, advanced text analytics tools analyze the unstructured text of social media and social networking posts, including Tweets and Facebook posts, to determine user sentiment related to particular companies, brands or products. Analysis can focus on macro-level sentiment down to individual user sentiment (Jeff Kelly, 2012). For instance, Adidas capitalized on social media when it introduced its latest running innovation, a shoe called Energy Boost. Lia Vakoutis, head of digital strategy at Adidas America, says the strategy paid off. "We saw a dramatic increase in positive sentiment
  6. 6. around Adidas following the launch of the Adidas Energy Boost running shoe," she told All Analytics. Vakoutis said the company uses a variety of tools to monitor social media sentiment, including Salesforce Radian6, Sysomos, and Crimson Hexagon. Combined, she says, they provide "the most holistic view of Adidas sentiment on the web." (Noreen Seebacher, 2013). Similarly, in Feb 2012, Barclays launched a mobile banking application called PingIt. In the days following the launch, Barclays made significant changes to the application as a result of real-time social media analysis. A sentiment analysis was carried out for understanding the sentiments for this newly launched product. Although the application was very well received, a small proportion of mentions were negative. Barclays was able to drill into this data to see what was causing the negative mentions and found out quickly that many users were unhappy that the application didn’t work for under 18’s. It wasn’t only teenagers that were unhappy, but also parents who couldn’t transfer money to them. This could easily create a PR disaster, but the data allowed Barclays to act quickly. Within a week, 16 and 17 year-olds were given access to the application, showing that Barclays were responsive to customer feedback. It wasn’t only the negative comments that Barclays bank looked into. The positive mentions also revealed some surprises that they were able to act on. For example, there were a lot of positive comments about being able to check your bank balance from the app. This was only intended to be a side feature, but proved to be extremely popular. As a result of this feedback, Barclays developed new apps specifically for this purpose. (Oursocialtimes, 2012) The sentiment analysis can be used by a firm in understanding its current customer satisfaction levels. It can understand what kinds of its products are generating a large number of negative sentiments or people in what regions are most unhappy with the firm’s services. Similarly, it can look out for competitor sentiments and upcoming trends in the industry. The firm can also look at the above example of Barclays bank and utilize the social media Big Data for understanding the customer sentiments for its newly launched products. This way, it’ll be able to tweak its product offering to better suit the customer needs. In terms of data analysis, the firm should gather data from various social networking sites such as Twitter and Facebook (including their own Facebook and Twitter pages), Blogs and forums, especially industry specific, consumer complaint websites, log details of consumers etc. This data can be collectively analyzed with the customer feedback (or customer
  7. 7. complaints) data generated from different sources such as online support, voice support etc. and consumer research data generated from various surveys. 3.2 ING Direct – Using customer feedback for customized product offerings ING Direct is a different kind of bank as it doesn’t have bricks and mortar branches. Social media was an important focus for the bank and it has been in the forefront of social media usage in the financial services industry. ING Direct’s biggest social media challenge was to be seen as “more than just a bank”. ING initiated a new product – THRiVE Chequing – an online no fee daily chequing account that actually pays interest. ING engaged over 22,000 of their clients in product’s preview, with their feedback directly influencing the final offering. In addition to the bank’s website, they gathered customer insights through Facebook and Twitter. They believe in asking direct feedback, which is a great, proactive way to get or keep the conversation going. This method created many valuable suggestions for the THRiVE Chequing, including increasing the number of free cheques and increasing number of bill payees. They continued to ask feedback from its customers to drive its promotions and better understand client’s needs. The THRiVE chequing account product was a major success for ING DIRECT. In a very short time, the campaign attracted over 40,000 active THRiVE-ers. The campaign had over 5 million impressions on social media sites. Blog posts covering the THRiVE chequing public launch were read 53,000 times and #THRiVETASTIC was mentioned online to an audience of over 3.6 million users. (Salesforce, 2011) This method will be beneficial for a firm for developing its new products. While developing a particular product, the firm can get feedback from its customers on their needs and requirements related to those products. This way, the firm will be able to create customerfocused products which will have a high chance of success. Additionally, as the customers are already made aware of this kind of product, it’ll a lot easier for the firm to market its products to its target customers. The crowdsourcing feature of social networks acts as a powerful tool to target product development at the financial lifecycle and future needs of its customer segments. (Kishen Kumar, 2013) For data analysis in this case, the firm should seek feedback from its customers on developing similar kind of products and gather data from various customer-touch points such
  8. 8. as personal emails, personal customer visits, personal feedback calls to customers, online feedback from social networking sites such as Twitter and Facebook (including its own Facebook and Twitter pages), and blogs and forums (company sponsored or external). This data can be collectively analyzed with the information on products that the current customers use and their spending patterns. 3.3 DreamWorks’s - Tracking PR events and promotional campaigns DreamWorks was trying to understand if it could determine how movies would open based on the social buzz. It found out that the ability to understand public sentiment in real time was very predictive of how a movie would open and what advertising worked. For example, it tracked the DreamWorks’ movie Puss in Boots, which had a slower following in the weeks leading up to its release. After analyzing initial response, it discovered that before the movie’s release, the Twitter conversation about the film was sparse and surprisingly negative (Jonathan Taplin, 2012). In response, the studio created a new TV ad campaign that was well received. When it introduced this new big TV ad, it observed that within two days, Puss in Boots became the most talked about movie. It analyzed social media posts from Twitter, Facebook and other social media related to this movie release and immediately observed that the ad campaign had worked. The movie was a hit, and the Twitter mention volumes and positive sentiment increased significantly (IBM, 2013). A firm can effectively utilize social media analytics information to monitor the impact of its advertising campaigns and can get feedback on its promotion efforts. Additionally, it can alter its promotional campaigns based on the response received from its customers. This way the firm will be able to generate better return on its advertising and promotions efforts. Finally, it can segment its customers at a micro level and target its online marketing efforts according to specific customer needs. In terms of data analysis, the firm should collect data about the level of activity observed for a particular advertisement or a promotional campaign. This information can be collected from different social networking sites such as Twitter and Facebook (including their own Facebook and Twitter pages), blogs and forums etc. and the level of activity should be measured against the keywords related to that particular advertisement or promotional campaign. Another major source can be the information about number of clicks for the paid
  9. 9. searches and online ads and the related traffic generated for its related products. This all data can be analyzed together to find out the measure of success for that particular advertisement or promotional campaign. 3.4 TD Bank – Social media for rapid customer Service Customers have started putting their complaints on forums, blogs, Facebook, Twitter and other social media. By listening carefully to these communities, customer concerns can be easily identified and a better service can be provided. TD Bank understands that customers place a great deal of trust in their bank and they expect it to be as accessible, helpful and responsive to their needs as possible. To TD that means being there for customers where they feel most comfortable, whether it’s in the branch, on the phone or on social media channels. TD has built social media teams in Canada and the US to provide customer service (or “Social Service” as they call it). These teams are located in major call centers and are focused on delivering customer service via Twitter, and engaging customers with help and advice on blogs and on TD Money Lounge on Facebook. The teams use social monitoring tools to analyze the Big Data in social media to track mentions, find relevant conversations and manage the team’s workflow. Using social media data analytics, the bank has been able to track and identify and provide help with thousands of customer inquiries on a range of topics, from service issues to banking hours. For example, during Hurricane Irene, which shut down much of the east coast for several days, TD was able to update affected customers with information about branch and ATM availability (Salesforce, 2013). TD Bank has been successfully using social media to assist their customers, share information and make connections. “People are very candid on social media, and it gives us the chance to get feedback on our branch hours and services or help a customer resolve an issue. We find our customers are happy to know that we are listening and that we are here to help” says Wendy Arnott, VP of Social Media & Digital Communication for TD Bank. TD Bank does a fantastic job of delivering friendly and efficient customer service via Twitter and engaging customers with help and advice on blogs and on their TD Money Lounge on Facebook (Julie Meredith, 2012). This is an excellent example of how TD, an industry leader in customer satisfaction is using social media for enhanced customer service. A firm can leverage the social media channels
  10. 10. such as Twitter and blogs to understand customer needs, monitor customer complaints and provide quick resolution to their queries. The firm can track the results of its social media customer service in terms of reduction of customer complaint calls it receives. This way, a firm can reduce customer service costs and improve customer satisfaction. For data analytics in this case, the firm should get data from 3 different sources. The first source of information is about online customer feedback or complaints mentioned on various social media websites and forums. This includes online blogs wherever there is a mention of this firm, the Twitter feeds or tweets that include any related hash tags, Facebook posts or comments mentioning about the firm or any of its products and customer complaint forums where there is any complaint or issue against this firm. The second source of information is about the customer complaints information from its phone service, online feedback (through its website) and email feedback. The third source of information is about the details of its customers related to products they are using, services they are offered and the volume of their activities. These 3 sources should be collectively analyzed to find out the real issues concerning its customers and act accordingly to resolve those issues. Appendix 2 gives the details of data requirements and sources for Big Data analytics business cases. Appendix 3 shows the resource architecture for social media from where a firm can collect the Big Data information on social media. 4. IMPLEMENTATION FRAMEWORK In the following section we will discuss potential implementation process of the Big Data systems at a firm with a particular focus on use of social media tools. According to Deill and Ross (2008), before implementing new systems, a company should understand what is not working with the current systems and how the new system will help achieve company’s objectives. It has to be clear to management and data analytics employees how information flow will happen under the new system, and how conclusions derived from data will be incorporated in strategic decision making. Authors suggest that a company should create an integrated IT strategy focused on business processes as opposed to data management. In this case it means that it has to be clear how new data analytics systems will contribute to a firm’s business.
  11. 11. 4.1 Information flow In order to illustrate the information flow and how it will affect decision-making, we have created a high-level information flow diagram. To provide reliable, updated and integrated information, all data should go through one central data analytics group as represented in the middle of the diagram. The information gathered from previously analyzed data would be transferred to the relevant departments. For instance, a firm’s data analytics group could help marketing department to segment customers and suggest customized offers. Additionally, some data can be distributed to external agencies either to sell data or create shared services as long as the firm can guarantee sufficient safety of their customer privacy. Although data from various departments would flow in Central analytics group, the actual meaningful information would come from this central analytics group as they would have advanced tools to analyze the various types of data supplied. Marketing External agencies Management Central analytics group Commerce Operations Adopted from “Hub and spoke” model for web analytics team deployment within the organization, (Peterson, 2008) 4.2 Operating model As part of creating IT strategy, a firm would have to decide on its operating model. The operating
  12. 12. model is selected based on how integrated and standardized are the business processes at a large-sized firm. As can be seen in this table, different systems provide different benefits. For instance, systems that are integrated, allow gaining efficiencies from sharing information across different branches whereas standardized systems allow implementing changes quickly. For Big Data, ability to integrate different systems play a crucial role since data pulled from various sources allow for more representative and reliable analyses. Standardization of Big data processes would allow the firm to transfer knowledge gained from data analytics rapidly, saving time and allowing engaging in more real-time offerings. Thus, we would advise firms to choose highly integrated and standardized operating model, also called unification model. This model would provide the firm with nation-wide data access from all its subsidiaries and foster standardized processes across all of its units. 4.3 Big Data implementation process According to Microsoft researchers (Fisher et al, 2012) 5 step Big Data pipeline process outlined in Figure 1, those processes also reflect the main challenges associated with usage of Big Data. The first step is acquiring data. Understanding which data is required is a big challenge for many companies. Data can be generated internally, but also acquired from public databases, social media or bought from private companies such as Microsoft’s Azure Marketplace and Infochips. Sometimes linking data from various sources and formats can be technically difficult although data itself is available. Thus, as highlighted in previous analyses concerning operating model, it is important to have systems that can be easily integrated. Figure1. The Big Data pipeline The second step is selecting the architecture based on cost and performance. Since the local programs cannot perform extensive Big Data analyses, an appropriate platform has to be chosen. Several options are provided through cloud computing. platform It used is likely that will look
  13. 13. substantially different from local programs, thus analysts will have to get acquainted to the new platform. When selecting the platform, both costs and design should be considered. Often costs are based on how extensive are the analyses, clients paying more for more computation and when larger systems are bought. Unfortunately, it is hard to estimate the costs or duration of computing, as more real time data can be continuously added to systems and impose non-linear costs in terms of overhead, storage and other aspects. Usually estimates are made by iteration and re-running the analyses to eventually achieve time-cost balance point. The third step involves shaping the data to the architecture. Throughout this process analyst has to ensure that the data uploaded is compatible with how the computation will be structured, distributed and portioned. Cloud-computing systems use data storage in a different way than desktop machines. For instance, there are cloud-based data-systems such as Amazon’s RDS or Microsoft’s SQL Azure, distributed file systems such as Hadoop and more novel data structures such as Azure’s queues and blobs. One of the challenges with using these systems is moving data back and forth from the cloud to the local machines. Furthermore, large files need to be organized, partitioned and prepared before uploading them on the cloud. Furthermore, once data is uploaded, it also needs to be cleaned, which is a difficult process require expertise from multiple people. The fourth step is focused on writing the code. Basically in this stage analysts decide what type of analyses will be performed with the data. It could be C£, Microsoft’s SCOPE, but also such languages as R, Python or PIG over Hadoop. High-level languages have to be able to support parallelism in order to break down and manipulate different analyses. These high level languages allow analysts to abstract away from considering where the data is processed, and focus more on the nature of computation. However, a common challenge is lack of transparency typical to the Big Data analyses. When analyses are being performed in parallel, detecting trough system failure is more complicated as true symptoms if failure can be masked by other problems. The last step is concerned with debugging and iteration. To test if the system is running smoothly, analysts will look for potential bugs. However, debugging in a cloud-based environment can be much more complicated since a single crush might be distributed across multiple virtual machines with trace files also distributed on a variety of machines. Furthermore, if a virtual machine fails jobs are moved to different machines hiding errors and
  14. 14. reducing transparency. Another problem associated with Big Data analytics is difficulty to modify some parameters, decreasing analysts’ ability to adopt iterative approach. It can take several hours for analysts before trying different parameters due to extensive amount of data that need to be analyzed beforehand. Finally, ability to visualize and see the context is critical when working with large data sets. It enables to see correlations between variables and identify patterns in data. Working in analytics cloud environment has many challenges. To drive the change, it will require the firm to carefully assess different systems and how can they be integrated. Furthermore, firms should analyze security issues related to using cloud-based systems. Additionally, significant improvements can be expected within the next few years that would allow to better break down and analyses various types of data allowing partial computations and more rapid iterations. 4.4 Future Scalability Future scalability is an important factor to be considered before selecting the technology for Big Data. With numerous options available in the market, it is important to evaluate the scalability of these systems for future trends so that the system doesn’t become obsolete sooner. Adopting tools or software successfully implemented in other businesses is a riskreducing strategy but it is important to understand that those companies based their implementation decisions on the options that were available at an earlier point in time. Another important consideration that the organization has to make concerns the capacity of the disks and the number of disks (or servers) to procure. Capacity is important when data storage is the predominant use of the system. But, if the requirement is to constantly access the data, more disks (or servers) are required to reduce the retrieval or processing time. So it is important to find the balance between the storage capacity and the number of disks (or servers). 5. IMPLEMENTATION PROCESS Considering the implementation framework given above, a firm will have to select the most appropriate tools for the company’s Big Data strategy and its goals. We have outlined a process map for implementing systems and listed several relevant alternatives for social media analytics tools in the following paragraphs.
  15. 15. 5.1 Process Map for social media systems To derive benefits of data in social media, a firm should consider the following set of actions: 1) Setting Objectives – Link the data being gathered and/or analyzed directly to the business goals to be achieved. Typical objectives include understanding customer sentiments, feedback on marketing or promotions, reducing customer service costs, getting feedback on products and services and improving public opinion of a particular product or business division. 2) Defining KPIs - After identifying the business goals, key performance indicators (KPIs) for objectively evaluating the data should be defined. For example, customer engagement might be measured by the numbers of followers for a Twitter account and numbers of retweets and mentions of a company's name. It can also be in terms of cost of reduction in customer service calls due to enhanced customer service on social media. 3) Identifying social media monitoring tools - There are a number of types of software tools for identifying and analyzing unstructured data found in tweets, blogs, forums and Facebook posts. (Margaret Rouse, 2012) 4) Test your hypotheses. After gathering the data, filter it and look at it from multiple perspectives (such as over different time frames) to test your hypotheses. 5) Draw insights. Finally, the data should help the firm arrive at well-informed assumptions and insights, which can then guide its actions in social media or in other customer-facing channels. (James A. Martin, 2013) 6) Engage and Act – After analyzing the data and drawing insights from it, engage your audience. Figure out what they're looking for, so be sure to act upon the data once it has been analyzed. (Ben Parr, 2009). The strategy for implementing Social Media Big Data analytics is shown in Appendix 8. 5.2 Implementing Social Media systems using specialized products The next section gives an overview of the tools available for social media monitoring. A comparison of various Social Media monitoring tools is shown in Appendix 4, where these
  16. 16. tools are compared on the basis of several factors. A detailed description of top 5 Social Media monitoring tools is provided below: 1) Radian6 – This product helps brands to listen more intelligently to their consumers, competitors and influencers and provides detailed, real-time insights. Beyond its monitoring dashboard, which tracks mentions on more than 100 million social media sites, an engagement console is also available that allows a company to coordinate its internal responses to external activity by immediately updating company’s blog, Twitter and Facebook accounts all in one spot. Everything is fully automated (J.D. Lasica & Kim Bale, 2011). The Salesforce Radian6 product doesn’t process any information from the firm’s internal systems such as emails, social media systems, sharepoint and other internal systems. But its open API has allowed many systems to integrate with Radian6. With features developed on their platform API and Social Metrics Framework for integrating third-party data, Radian6 now supports the integration of social customer relationship management (CRM – only Salesforce), web analytics, and other enterprise systems. Cost: The dashboard starts at $1000/month (could range higher depending on mentions, (Zach Ellis, 2013)) and includes the following features within the basic package: Product Package Users Name Radian6 – Features Additional Price Clients features Basic Salesforce Marketing cloud 1000 users  Social listening – Web 20,000 mentions Analytics  Up to five social presences  30 days historical data  100 topic profiles  Training and best practices (Salesforce, 2013) $1,000 TD, Red Cross, per Adobe, month Cirque du Soleil, H&R AAA, Block, March of Dimes, Microsoft, Pepsi, Southwest Airlines 2) Sysomos – its Heartbeat is a real-time monitoring and measurement tool that provides constantly updated snapshots of social media conversations delivered using a variety of userfriendly graphics. Heartbeat organizes conversations, manages workflow, facilitates collaboration and provides ways to engage with key influencers. Sysomos also offers a Media Analysis Platform.
  17. 17. Cost: Entry-level price of $500/month. Clients: IBM, HSBC, Roche, Ketchum, Sony Ericsson, Philips, ConAgra, Edelman, Shell Oil, Nokia, Sapient, Citi, Interbrand. (J.D. Lasica & Kim Bale, 2011) 3) Lithium - Lithium monitors the search-specific mentions and sentiments in social media outlets and outputs them into easy-to-read graphs and numbers resembling the stock market. Lithium will aggregate information from a variety of platforms including blog posts and comments, Twitter, Facebook, Flickr and many others, and it’ll assess emotions surrounding the brand’s pre-, mid- and post campaign so a company can adjust its strategies accordingly. Cost: Base plan of $249/month for five users and five searches. Clients: Best Buy BT, Barnes & Noble, FICO, Disney Online, Stubhub, Motorola, Coca Cola, Focus Features, Netflix. (J.D. Lasica & Kim Bale, 2011) 4) Collective Intellect - Using a combination of self-serve client dashboards and human analysis, Collective Intellect offers a robust monitoring and measurement tool suited to midsize to large companies with its Social CRM Insights platform. It applies spam management techniques and text analysis to clean data sets, delivering customers rich intelligence. Collective Intellect blends heavy-hitting technology and algorithms to search, collect, filter, cleanse, analyze and produce robust reports. Collective Intellect uses impressive real-time social analytics for powerful monitoring and sentiment accuracy (Toptenreviews, 2013). Cost: Pricing starts at $300/month and scales based on specific client needs, according to published reports. Clients: General Mills, NBC Universal, Pepsi, Walmart, Unilever, Advertising Age, CBS, Dole, MTV Networks, MillerCoors, Paramount, Verizon Wireless, Viacom, Hasbro, Siemens. (J.D. Lasica & Kim Bale, 2011) 5) Alterian SM2: This tool tracks mentions on blogs, forums, social networks like Facebook, microblogs like Twitter, wikis, video and photo sharing sites, Craigslist and ePinions. SM2 monitors the daily volume, demographics, location, tone and emotion of conversations surrounding a brand and aggregates results into positive and negative categories for quick review by anyone on staff. Cost: Pricing is based on volume of results and ranges from $500/month to $15,000/month. “Freemium” trial plan allows for five keyword or phrase searches and a total of 1,000 results. Alterian also provides additional custom solutions. Clients: Rosetta, MDAnderson, Pursuit, YouCast.
  18. 18. Other specialized vendor services for social media monitoring includes Brandwatch, Beevolve, UberVU, Viralheat, Trendrr, Attensity360, Simplify360 etc. (J.D. Lasica & Kim Bale, 2011) Appendix 4 shows the comparison of the tools in a tabular format. Recommendations: Depending on the goals established by a firm, the company can choose between various options. As can be seen, TD already uses Salesforce tool Radian6 that allows receiving insights about costumers, competition and influencers as well as coordinating internal responses to external activity via all accounts. In contrast, HSBC along with IBM and Shell Oil use Sysomos that provide real-time monitoring and measuring tool with user-friendly graphics. These two tools have also been ranked as the top two in a comparison of social media tools for 2013, as can be seen in appendix 4 due to their vast amount of features. We would advise the firm to use Appendix 4 to evaluate features that would be required for achieving its goals. However, we would advise either Radian6 or Sysomos as they have all the necessary features if required. 5.3 Implementing Big Data using Hadoop Many of the previous cases have shown that applying such systems as Hadoop can bring great benefits to the company. Apache Hadoop is a high scale, open-source distributed computing platform that includes the Hadoop Distributed File system and an implementation of MapReduce(Lamont,2012). For instance, using Hadoop allowed Sears (Henschen, 2012) to reduce campaigns for its loyalty club from six weeks to weekly analyses. This was achieved because Sears moved from its mainframe Teradata and SAS servers to Hadoop’ s cloud environment. Furthermore, it allowed Sears to perform more granular targeting, which in some cases included even individual customers. Previous models used 10 % of available data whereas analyses performed by Hadoop used 100% of data provided by Sears. Hadoop’s strengths come from its ability to divide workloads across many servers and perform analyses simultaneously. According to Shelley, the CTO of Sears, Hadoop also enables the company to create significant cost savings, since mainframe computers would cost between $3000 to $7000 whereas Hadoop’ s costs are small fraction of that. Another upside of Hadoop is its ability to store data in a raw format. If a company wants to perform
  19. 19. analyses with a different model five years from now, it has all the data available in the right format. However, the downside of Hadoop is that this platform is relatively immature and there is a lack of Hadoop talent. For instance, Sears had to learn everything about this platform by trial and error with a limited help from external consultants. Furthermore, for Sears it takes 90 minutes to extract the data from mainframe servers to Hadoop and bring results back to servers. This is a cost Sears has to pay for using legacy systems while simultaneously operation o Hadoop. Recommendations: Hadoop has proven to give relevant insights for many companies including Sears by allowing it to reduce time required to perform analyses, enabling more real-time offers and personalized services for many different segments. However, a firm should also consider downsides of Hadoop such as scarcity of talent and expertise as well as the time required to transfer information forward and backward from Hadoop systems. Furthermore, the firm should also perform risk analyses to assess safety issues related to transferring data. Additionally, Hadoop can be combined with many different applications, and serve as bases for more advanced tools. One of those tools created by IBM will be discussed in the next paragraph. Infosphere BigInsights by IBM Infosphere BigInsights by IBM is based on Apache Hadoop, and by combining power of Apache Hadoop with its own innovations, IBM provides companies with insights from new and emerging type of data that previously were not possible to analyze (IBM, 2013). BigInsights provides tools on advanced analytics, performance optimization, enterprise integration, visualization and others. Furthermore, application connectors make BigInsights data accessible to any Java Database Connectivity compatible data store, including Cognos Business Intelligence. Additionally, IBM’s own unique innovations include “sophisticated text analytics module, IBM Big Sheets for data exploration and a variety of performance, reliability, security and administrative features" (IBM, 2013). With Infosphere BigInsights IBM has integrated individual Hadoop components and their own added features into one single product to simplify development, implementation and management for enterprises. This allows companies to both optimize their day-to-day operations and gain micro-level understanding of “customer attitudes, trends and relationships” “sophisticated text analytics
  20. 20. module, IBM Big Sheets for data exploration and a variety of performance, reliability, security and administrative features" (IBM, 2013). Recommendations: Using IBM’s Infosphere BigInsights would allow a firm to leverage knowledge possessed by IBM and avoid drawbacks of not having sufficient expertise in Hadoop systems. Furthermore, Infospheres BigInsights have added security and reliability features to their tools, ensuring more safety for data gathered by the firm. This tool is based on open source platform, but is user focused and simplifies and accelerates implementation process of Big Data processes in the company. Considering all these aspects, it would be a valuable alternative for using and understanding Apache Hadoop in its raw form. Furthermore, Infosphere BigInsights is particularly suitable for analyzing customer segments based on Big Data. However, for smaller scale analyses such tools as SPSS Advanced Statistics or Intelligent Miner data mining suite can also be used. 5.4 Implementing Big Data analytics systems using packaged products 1) IBM SPSS modeler – is a data mining and modeling tool that helps the clients see the trends and patterns in their data. Clients can easily build predictive models quickly without any programming. SPSS modeler helps you make effective decisions by analyzing structured data, utilize advanced linguistics technologies and process large unstructured text data. It also includes social network analysis depicting social behavior of individual or groups, and identifying social leaders influencing behavior of others. IBM SPSS modeler performs automated modeling by estimating and comparing number of different modeling methods in order of their effectiveness generating results in very interactive and visual format. Appendix 6 shows the sample architecture for real-time implementation. 2) IBM Analytical Decision Management – helps the clients make best business decisions by optimizing and automating high-volume decisions and solid data analysis. It helps in applying predictions within real-world constraints to reach optimal decisions using analysis of structured and unstructured data. The client can adapt recommendations through feedback mechanism. For example, customer service agent can access marketing offers tailored to specific customers in real time thereby improving customer attainment, growth and retention (IBM, 2012)
  21. 21. 3) Cloudera Enterprise – gives 360 degree customer view by combining information stored on different systems such as CRM, financial, point of sale, marketing, customer support etc. It is a combination of Cloudera’s open source Hadoop stack (CDH), powerful management platform (Cloudera manager), and Cloudera’s expert technical support. Financial institutions can create central data hubs that combine large diverse data and after in depth- analysis can provide personalized recommendations to its customers by uniquely targeted offers, crossselling and up-selling products (Cloudera, 2012). Appendix 7 explains the sample for integration of transaction data and interaction data. Recommendations: In this case a firm might consider using any or all three tools as each of them has different advantages. Cloudera provides a comprehensive analytical view of the data that any big firm might need but it is important to note that it integrates different technologies appended to existing systems which may increase complexity and integration issues in future. But its use of open source systems and ease of integration with them may bring down the cost considerably. Implementing the IBM tools such as SPSS modeler and Analytical decision management tool would make seamless integration among them and also provide the features other systems do. Also support will be available from the vendor as compared to possible scarcity of resources for Hadoop and related systems. 6. RISKS AND MITIGATIONS Though Big Data has many advantages, implementing Big Data system in an organization has some inherent risks associated with it and sincere mitigation efforts are required to reap the benefits of the system. Below are some of the risks and corresponding mitigation techniques that are to be employed to maximize the effectiveness of the Big Data systems. 8.1 Security Risks Organizations that deal with sensitive customer data like financial institutions may face a huge security risk by implementing Big Data, if proper control measures are not employed. Security breached and loss of information also makes organizations face law suits apart from losing customer satisfaction. To avoid security breaches, IT has to use additional security products built to specifically apart from using usual security procedures like restricted access, encryption of confidential data.
  22. 22. 8.2 Analytics’ cohesion with business goals It is important to set the goals and specific objectives for the Big Data system so that it meets the business goals. Many organizations lose direction as what they want to accomplish through the Big Data system and cannot obtain the expected benefits. Also they should be completely aware of data sources they are going to use and the integration between them. To reap maximum benefit it is important to develop an implementation road map and clear process to classify and utilize data. 8.3 Managing complexity and Storage capacity An inherent issue that Big Data has is the complexity that builds up with more and more data the system handles and the effective maintenance of the stored data. With time the complexity will nothing but increase and it will require storage capacity and investment to expand the capacity. To make effective usage of the Big Data system, the firm has to allocate budget for maintenance, and periodic upgradation of technology, security features and hardware. 7. SUMMARY / CONCLUSION – 7.1 Focusing on Big Data opportunities After looking into various cases above of how Big Data is utilized by various firms to derive benefits for different purposes, it becomes apparent that a firm can considerably benefit from the Big Data analytics. Following is a summary of the different cases that a firm can consider to implement Big Data in Social Media space: BIG DATA ANALYTICS FOR SOCIAL MEDIA  Customer sentiment analyses  Using customer feedback for customized product offerings  Tracking PR events and promotional campaigns  Social media for rapid customer Service
  23. 23. 7.2 Recommending a Strategy for implementing Big Data systems: For implementing Big Data analytics, there are several alternatives that a firm can consider. According to a research done by a non-profit organization AIIM (Association for Information and Image Management), many firms that are planning to implement Big Data systems are tempted to press ahead with in-house developments using open-source components (Such as Hadoop) as it might give them early-mover competitive advantage. A lot of vendors are moving quickly to provide packaged product sets, and this is driving a need for standardized connectors to provide unified data access to as many different databases as possible. However, usability outside of the technical department is important, and for Big Data, assurance of robust security is essential. (AIIM, 2012). As observed from the below survey, a majority of firms prefer a combination of options for implementing Big Data systems. Multi-Criteria Decision Analysis (MCDA): We carried out a Multi-Criteria Decision Analysis for evaluating the above mentioned alternatives. In total 4 alternatives were evaluated for implementation of Big Data Practice within a low-risk oriented firm – Option 1 is in-house development using open source tools such as Hadoop, Option 2 is going with Specialized products (Such as Radian6 mentioned above), Option 3 consists of “Packaged product sets” (e.g. Packaged products from Oracle or IBM) and Option 4 is about having a mix of Specialized products and Packaged products. The model consists of two components. “Evaluating Strategy Value” and “Ability To Implement”. The “Evaluating Strategy Value” component consists of rating the options on
  24. 24. several Strategy factors - Quick Impact, Data Security, Low Risk, Integration and Features. While the “Ability to Implement” was evaluated on factors such as Funding, Skills, Ease of use, Support and Flexibility. Appendix 9 gives the details of analysis. Strategy Value Matrix 245 Opt4 230 Opt2 Opt3 Opt1 215 S t r a t e g y V a l u e Options 200 0 100 200 Ability to Implement 300 Recommendation: From the above Strategy Value Matrix, it can be observed that Option 4 (or Mix of multiple models) looks as the best suitable option for implementing Big Data systems for a firm. This option is followed by Option 2 which is about implementing “Specialized Products” and Option 3 which is about going with “Packaged Product Sets”.
  25. 25. Appendices Appendix 1: Porter 5 forces for analyzing factors affecting a firm Threat of new entrants – Few barriers to entry and high profit margins may attract new entrants Threat of substitutes – New trends or products that might substitute existing products Power of customers – More choices and consumer driven markets Firm Competitive rivalry – Price war and similar products Power of suppliers – change in supplier prices and availability
  26. 26. Appendix 2: Data requirements and sources for Big Data analytics business cases Data Sources Data requirements for Big Data analytics Internal External Availabl e Firm Products Online forums, blogs, consumer complaint websites × × × × × × Customer transactio ns Online social networking sites (Twitter, Facebook) × Cases Customer demograp hics Set of services provided to a customer × × × Customer support voice, online Paid searches, online ads Barclays and Adidas Social Media ING Direct × × × DreamWorks TD Bank × × × × ×
  27. 27. Appendix 3: Social Media Resource architecture Sample Retrieved from
  28. 28. Appendix 4: Comparison of Social Media Tools
  29. 29. Appendix 5: CRM Software comparison CRM software System Strengths Weaknesses Overall Sales force On Contact Sage ACT -More tools than other CRM Best for viewing social Has more detailed functions softwares apart from networking sites & company than any other CRM function Salesforce websites Only available as an online Weaker marketing and sales package functions No unique features Avidian Prophet A lot of assistance Not intuitive AIM CRM Track routes and redirect customers to right departments Only webhosted The best overall alternative Good for social networking, Great in sales and marketing Easy to use, many detailed Provides a lot of assistance, for available CRM software but not as good for other assistance, but not as good options lacks some features packages functions for other functions CRM software reviews. Retrieved from: Appendix 6: Real-time data architecture
  30. 30. Appendix 7: Big Data is confluence of transaction data, interaction data and Big Data processing Retrieved from:
  31. 31. Appendix 8: Social Media Strategy – As social media strategy evolves, valued customer relationships grow stronger Social Media Strategy: Retrieved from,%202011.pdf
  32. 32. Appendix 9: Multi-Criteria Decision Analysis Model (MCDA) EVALUATING STRATEGY VALUE CRITERIA Weight Option 1 – In-house Developme nt Quick Impact Data Security Low Risk Integration Features 7 7 7 4 5 5 9 9 6 6 Option 2 Specialized products (SaaS) Option2 Weighted Rating Option 3 Packaged product sets 35 63 63 24 30 30 Option1 Weighted Rating 9 7 7 6 9 63 49 49 24 45 7 8 7 8 7 215 230 Option3 Weighte d Rating 49 56 49 32 35 Option 4 - Mix of options 2 &3 Option4 Weighte d Rating 9 8 7 7 8 63 56 49 28 40 221 236 EVALUATING ABILITY TO IMPLEMENT CRITE RIA Optio n 1 Ratin g Option 1 Weighti ng criteria Option 1 Weight ed rating Optio n 2 Ratin g Option 2 Weighti ng criteria Option 2 Weight ed rating Optio n 3 Ratin g Option 3 Weighti ng criteria Option 3 Weight ed rating Opti on 4 Rati ng Option 4 Weightin g criteria Option 4 Weight ed rating Funding 6 8 48 8 7 56 6 8 48 8 7 56 Skills Ease of use 6 8 48 7 4 28 7 4 28 8 4 32 6 4 24 8 6 48 6 5 30 8 6 48 Support Flexibili ty 7 5 35 9 6 54 8 7 56 9 6 54 7 5 35 8 7 56 7 6 42 9 7 63 30 190 30 242 30 204 30 253 Results Ability to implement Strategy value Opt1 190 215 Opt2 242 230 Opt3 204 221 Opt4 253 236
  33. 33. References Hsieh C., 2009, Casino CRM: Issues and some implementation strategies, Communications of the IIMA. Sauber W.M. , 2009, Using customer analytics to improve customer retention, Aspen publisher Inc. Oracle. (2009). Building a bank’s brand equity through Social media. Retrieved from Todd Nash, (2013), Exploring Big Data in Small Steps, Starting with Social Media Analytics. Retrieved from Deloitte. (2011). Analytics in banking - Taking a fresh look at your challenges. Retrieved from TDWI, David Stodder, (2012). - TDWI best practices Report - Customer Analytics in the Age of Social Media. Retrieved from Bates, J. (2012). Banking on a future in big data - how DBS bank is driving customer engagements with decision analytics. Retrieved from Business making progress. Baumgartner, T. (n.d.). Sales growth Five proven strategies from the world's sales leaders. Cloudera. (2012). Why are financial services firms adopting cloudera's big data solutions? CRM software reviews (2013). Top ten reviews .Retrieved from Deloitte. (2011). Rethinking retail banking growth. Harris, D. (2012). How Intuit uses big data to 'delight' you. Retrieved from IBM. (2012). Fiserv, IBM software smarter computing. Retrieved from IBM. (2012). IBM SPSS modeler help. Retrieved from IBM. (2012). SPSS software. Retrieved from IBM. (n.d.). Business Analytics in action. Retrieved from
  34. 34. Shalom, N. (2012). Realtime Analytics for Big Data: A Facebook Case Study. Retrieved 2013, from Nati Shalom's Blog: What is big data? (n.d.). Retrieved from IBM: Weil, A. (2013) Loyalty programs, HomeCare Magazine 33 Young, E. &. (2012). Global consumer banking survey. Young, E. &. (2012). The customer takes control . B66.ashx Lamont, J. (2012). Big data has big implications for knowledge management. KMworld, April. Peterson, E. (2008) Special Issue Papers Competing on web analytics. Journal of Direct, Data and Digital Marketing practice. Henschen, D. (2012). Big data, big questions., Nov. 5. Jeff Kelly, (2012). Big Data: Hadoop, Business Analytics and Beyond. Retrieved from,_Business_Analytics_and_Beyond Noreen Seebacher, (2013). Adidas Strikes Gold Mining Social Media. Retrieved from Oursocialtimes, (2012). How to use social media monitoring for a product launch. Retrieved from Salesforce, (2011). How to create a social media strategy for the financial services industry. Retrieved from Jonathan Taplin, (2012). Social sentiment analysis changes the game for Hollywood. Retrieved from IBM, (2013). Case Study: University of Southern California Annenberg Innovation Lab. Retrieved from Salesforce, (2013). CASE STUDY: TD BANK TD Bank uses social media to help make their customers even more comfortable. Retrieved from Julie Meredith, (2012). 4 Ways Banks Can Connect with Gen-Y on Social Media. Retrieved from
  35. 35. Margaret Rouse, (2012). Social media analytics. Retrieved from J.D. Lasica & Kim Bale, (2011). Top 20 social media monitoring vendors for business. Retrieved from Salesforce, (2013). TURN CONNECTIONS INTO CUSTOMERS FOR LIFE™. Retrieved from Zach Ellis, 2013, e-mail, 15 March, Toptenreviews, (2013). Collective Intellect. Retrieved from Kishen Kumar (2013). The real Big Data opportunity for banks lies in unstructured data. Retrieved from Ben Parr (2009). HOW TO: Track Social Media Analytics. Retrieved from James A. Martin, (2013), 5 Tips for Mashing up Big Data, Social Media. Retrieved from AIIM, (2012). Big Data - extracting value from your digital landfills. Retrieved from: Crunch, T. (2013). Salesforce Radian6 . Retrieved from IBM. (2012). Optimal segmentation approach and application. Retrieved 2013, from : IBM. (2013). Evaluate: IBM InfoSphere BigInsights . Retrieved 2013, from Kang, E. (2013). Live Pearson receives CUSTOMER Magazine's 2012 Product of the Year Award for LiveEngage. Retrieved from