CI professionals know that having a detailed understanding of their future competitive landscape is essential to build a winning strategy. In January 2017, a survey asked market and competitive insights leaders ‘What does best in class market forecasting look like?’. This highlighted four particular challenges:
1. How do I inject data analytics into the forecasting process?
2. How can I be confident that a ‘trend’ isn’t a one off, but here to stay?
3. How do I understand new markets where we have limited knowledge?
4. How do I deal with change & unknowns (new entrants)?
This webinar will answer these questions and share forecasting best practice with the SCIP UK community. During this webinar we will show you how to forecast complex market dynamics by providing tools and techniques that you can start using in your business straight away. This will cover scenario planning, research methods, decomposition forecasts, diffusion modelling and market mapping. The session will also provide an opportunity for you to ask questions about the challenges that your team faces.
As a result of this webinar, you should be able to review your current forecasting activities and implement best practice improvements. This will ensure that you are able to identify the right markets and the right products to invest in, support increased sales volumes, reduce costs and achieve better company-wide strategic alignment.
Speaker Profile
Jonathan Davenport, Head of Market Analysis, Milner Strategic Marketing
Jonathan leads the Market Analysis team at Milner. Jonathan’s expertise lies in building bespoke market forecast models which are used to understand market dynamics, including customer buying behaviour and competitor strategies. He also specialises in strategy development, using his extensive knowledge of tools and theory to develop robust market-driven strategies. He has a strong business to business sales and marketing background, developed over 14 years working across the energy, telecoms and bio-tech industries.
Jonathan is a Chartered Marketer and Member of the Chartered Institute of Marketing. He holds the Professional Postgraduate Diploma from Cambridge Marketing College and won the prestigious Elijah Hepworth Memorial Prize when he graduated with a BSc in Technology and Management.
Thank you Andrew , I’m very grateful to my SCIP colleagues for organising this event and providing me with an opportunity to share some Best in Class Competitive Intelligence – today, looking at market Forecasting.
During this webinar, I will briefly introduce myself to you and explain why market forecasts are so valuable to investors, leadership teams, product managers and insight professionals.
In fact, in a recent survey, market forecasting was highlighted as one of the key areas that professionals like you wanted to know more about
The respondents of the survey identified 4 key ‘stay awake issues’ relating to forecasting the future of markets
The main focus of our time today will be spent addressing these, where I will provide you with the tools that you can use to improve your forecasting in practice.
So we’ll explore:
Data analysis
Trend identification
Understanding new market opportunities
Dealing with change & unknowns such as new entrants
As we address each topic, I’ll share different tools and techniques that you can adopt within your teams in order to ensure your market forecasting is best in class
Andrew invited me to speak today because the company I work for; Milner Strategic Marketing has a lot of experience of building bespoke market forecast models for companies in the technology space
As the Head of market Analysis at Milner, developing market forecasts is an area where we have built up considerable expertise – in fact during the past 9 years with the company, I personally have built over 50 different models for technology businesses
Many of these are incredibly complex models – where we decompose the market into granular parts to provide a documented and defensible rationale for the forecast we create
I have worked with companies such as Arqiva and BT in the UK and internationally with business like HTC to help them understand how their competitive landscapes would change in the future, exploring technologies such as:
Microwave backhaul
IP Address Management
Next Generation Access super fast broadband
And SP and tablet adoption rates
We have also been lucky enough to share some of our expertise with others. Recently, I have
Presented at the 34th International Symposium on Forecasting in Rotterdam
Gave a talk last year to the SCIP UK Chapter in London
And we’ve Had articles published on forecasting the future for profit in SCIP’s Competitive Intelligence magazine and Cambridge Marketing Review
I know that each of you will be at different stages in your market forecasting journey: I’d be delighted if you need any help or advice my contact details are at the end of the presentation deck – please don’t hesitate to reach out for an informal conversation
What is a market forecast model? A forecast model calculates the market size by aggregating the purchasing behaviour of target customers in a given time frame. This can be measured by the number of units bought (unit sales) and the total money spent (market value). I’ll come onto how to do this later.
The model itself allows:
The industry attractiveness to be assessed and understood
Competitor strategy to be understood (which geographies is a particular vendor strong in, which products are they selling into certain markets and what pricing strategy will they adopt)
Consumer behaviour to the tracked (how many buyers are there, how many purchases will be made, what the churn cycle looks like, the make up of different sectors and segments and how this varies by geography)
The competitive strength of your businesses to be measured
Armed with an understanding of the wider macro environment, the internal strengths and weaknesses of your company can be assessed by looking at market share
Importantly, this doesn’t just tell us about what the competitive landscape looks like now, but provides a forecast about what it will look like in the future
Having one data source to understand the market to support these decisions, aligns the whole company. This means each interaction between each department can start with “what are we going to do?” instead of “what is going on?” The model allows departmental figures to be checked against the model for feasibility, as the model is based on the company’s own strategy and fed into by all departments.
Here are three application examples to show you what I mean:
Example 1: The Board may use this to support an investment decision
A detailed understanding of the size of the market and its growth rate, coupled with insights about the strength of the competition (market share), should underpin any investment decision. This is because size and growth are key measures of an industry’s attractiveness. Evaluating the market will de-risk the investment process, as likely returns can be calculated based on a quantified market understanding. We have found this to be relevant for both internal investment and private investors or grant bodies.
Example 2: Marketing and Product Development may use the model to develop the product roadmap
A market forecast model demonstrates how the buying behaviour of different market elements such as customer sectors, geographies, pricing segments and sales channels will change over time. These market insights can be used to develop rigorous and robust product roadmap plans. For instance if a company wanted to assess the PC market, it could develop a market model that forecasts demand across different elements; product forms (Laptops, Hybrids or Desktop PCs), screen sizes and memory requirements, over a five year period. This data would help the company manage its product portfolio and plan its product development to ensure it had the right products available for the changing needs of customers, at the right time.
Example 3: Sales may use the model to identify new revenue opportunities
The in-depth breakdown of changing customer demand across different market elements (such as sectors, geographies and pricing segments) provides an invaluable tool to identify new sources of revenue. Once the market size has been derived, the company’s market share can be calculated across a range of elements. This identifies areas of the market where the company is strong and likewise areas where there is room for development. It will also identify segments of the market that are unattractive now, but which will grow over the forecast period, presenting an attractive opportunity in the future. A long-term strategy and shorter-term sales plan can be developed based on this information, which will allow internal resources to be assigned to opportunities with the greatest potential returns.
It is because of the strategic importance of the applications we’ve been thinking about that in January 2017, a survey of market and competitive insights leaders found that Best Practice Forecasting was a topic that there was significant interest in.
But there was an acknowledgement too that people lacked experience in this area. The survey showed that there were four particular challenges that people wanted an explanation of:
How do I inject data analytics into the forecasting process?
How can I be confident that a ‘trend’ isn’t a one off, but here to stay?
How do I understand new markets where we have limited knowledge?
How do I deal with change & unknowns (new entrants)?
I am going to spend the reminder of our time today taking each of these challenges in turn and sharing with you some tools and techniques that you can use to address these ‘stay awake issues’
Data analytics – analysing the data you have available to deepen the accuracy of your forecasts
Repository for market intelligence – take the qualitative and quantitative data and feed this into the market forecast. We find this is particularly important because markets are just too complicated to understand the impact of all of the separate interactions we hold in our minds.
This is where decomposition forecasting, which breaks the market down into its constituent parts, becomes important
Data analytics should be used to understand the historic shape of these constituent parts, before forecasting forwards
Each constituent part has an individual forecasts made, before being sense-checked and triangulated at an aggregate level. We’ve found that its possible to have two things that make sense in isolation, but when you try to put them into a model where you can only model one thing at a time, you realise that they both can’t be true.
Data analytics should be used in decomposition forecasting to assess the historic trends in each component part of the market. The value of data analytic capabilities is that it allows you to understand and forecast the market at a much more granular level than would otherwise be possible.
Decomposition forecasting therefore provides 3 key benefits:
Greater accuracyIt is easier to be specific about smaller areas than about the total market
Greater transparencyExplicit assumptions behind each component forecast can be revealed
Greater consistency Decomposition forecasting helps companies stay focused on the most relevant factors of the market and maintain consistency in the forecasting approach
Let me show you what this looks like in practice
I wanted to start by showing you what a market forecast looks like. Rather than taking you into the complexity of the Excel, I’ve drawn a diagram or schematic to explain how we calculate the size of the market opportunity and how this demand can be broken down (or decomposed) into a number of different parts of structural elements
In this example, you’ll see that the company had 7 target markets that it wanted to understand in detail (and which it thought would account for the majority of customer demand worldwide), it then identified 5 target sectors (plus an other), had three core product groups which customers bought and a further four more specific product areas. So for example, if the company was selling laptops, tablets and smartphones as its three product types, it would be able to calculate the market demand for its products within a specific country, say Germany, addressing a specific sector or segment, say healthcare or a particular product type, for example a tablet and drilling down to market demand for a specific product for example 10” tablets.
Plus, once you’ve built the entire market model in this way, you can analyse customer demand across any one of the structural elements. So, using a model like I’ve described here, a company could:
consider whether customer demand from one product type such as smartphones would cannibalise sales from another product e.g. Laptops (by analysing the product type data)
Or they could explore the changing consumer demand, for example the increasing demand for larger screened handsets by analysing the product structural element
Likewise the importance of different geographies and sectors could also be assessed
You’ll see that the model contains 5 years of historic data starting in 2012 and going up to 2016 and then a 5 year forecast. So these models do not just provide a snapshot in time, but allow historic trends to be analysed and scenarios about the future to be explored. Some of you will want to understand seasonal demand, so your models may also have monthly or quarterly data to provide a more detailed picture of the competitive landscape.
Once you’ve understood the unit sales within a market, you can overlaying market pricing assumptions to calculate the market size by value.
It’s worth noting, that the more structural elements you add to the model, the more complex the model becomes, plus you also need to consider where you will collect the historic market data. This is important, because as we will see, it is from the trends in this historic data that will allow us to forecast future market behaviour.
However the deeper your understanding of the underlying market opportunity, the better decisions you should be able to make regarding investments, product roadmap developments or sales targeting etc.
This data analytics allows you to see the trend in each individual element of the market and use this to improve your forecasting accuracy. For example, the model above has 576 separate elements to analyse and forecast. Data analytics means you can isolate the individual trends in each geography, Segment, product type and product and use this to forecast forwards at a granular level.
By understanding the trends in each of these granular breakdowns, you can thoroughly understand the market drivers and the rationale behind the total forecast. This is invaluable when talking to other stakeholders in your organisation.
Desk research should be used as a starting point. Publicly available information should be systematically gathered and assessed, which will yield insights into:
Market dynamics
Competitive landscape
Macro-economic outlook
Regulatory framework
Innovation and rate of change
Channels to market
To truly understand a new market, I would recommend primary research as this helps you understand the specific trends in the market and customer and competitor behaviour within it.
Primary research is invaluable to understand the intricacies. As well as filling in gaps in assumptions that you have not been able to answer with desk research, primary research gives a much richer picture of how the market works in practice. This covers aspects such as:
Market size by segment
Customer requirements and behaviour
Technology attitude
Attitudes to vendors
Competitor behaviour
The strength of primary research is the qualitative insights it will give you about customer and competitor behaviour. Interviewees or respondents can include industry experts, customers, internal staff for customer insights or channel partners, suppliers, customers, internal staff or Freedom of Information requests for competitor insights. Industry bodies and analysts can also hold deep knowledge about the market.
The 2 key methods are structured interviews and online surveys.
Online surveys can be a very useful way of extracting both quantitative and qualitative information about consumer and competitor behaviour
Structured interviews consist of explorative questioning in a conversation about a topic and yield rich, fully-explained insights
This gives us information about what the market looks like now. We then need to use this to help our forecasting activity
Customer buying behaviour
This tells us a lot about the consumers that are buying the product – you may be familiar with Everett Rogers’ work (1962) on diffusion theory where he describes consumers of products as either Innovators, Early Adaptors, Earl Majority, Late Majority and Laggards.
Setting strategy
Geoffrey Moore (1991) built upon this work to describe the different types of strategy needed at each stage of the technology adoption life cycle
New Adopters – Diffusion of Innovation:
According to Diffusion of Innovation, the adoption curve can be broken into five sections of adopters with different characteristics and behaviours, affecting the market dynamics. The curve predicts the proportion of the target population that will adopt the new technology over time.
Market Churn:
Technology has a predictable life cycle. Milner analyses the churn (replacement) rate, which is affected by product durability and technology upgrade, to forecast the long-term sales demand for replacement products.
Total Demand:The total demand forecast takes into account both sales of new and replacement products over time.
It should be taken into consideration that one market might be growing and maturing in different regions simultaneously. The correct forecasting methodologies need to be adopted depending on the stage of development of the market. Furthermore, not all markets leave the growth stage – some fail to reach maturity stage and move straight to decline
Here we have a chart which shows the adoption (and subsequent pentation) of 16 different products by US households between 1900 and 2005
It is clear to see that many of these products show an S or logistic curve
This allows us to forecast consumer adoption behaviour. But forecasting is an art as well as a science – we see for example that the curves can be disrupted by external influences – these need to be taken into account
We see that penetration (definition … the proportion of customers that own a product i.e. have bought it at least once) of car (autos) actually declined because of the great depression followed by WWII when metal and fuels were needed for the war effort and sale of new cars was temporarily banned
If we differentiate this curve, it tells us about the new adopters that buy a good or service for the first time – turning the S shaped adoption curve into a Gaussian or bell curve distribution
Data analytics give us deep quantitative insight into trends in the market, but not qualitative insight into why they are happening. It tells us ‘what’, but not ‘why’.
There is a five step process to analyse data points and look for trends
There is a science to trend identification, but there is also an art where we need to rely on intuition – so we need to utilise both the left and right hand sides of our brains
When using scenario planning to evaluate trends, it can be helpful to think about them in terms of hard trends and soft trends. This affects the likelihood of a trend continuing.
A hard trend is a forecast that is based on measurable, predictable and tangible factors, for example technological progression or demographic change. There is a high degree of certainty that a hard trend will occur.
A Soft Trend is a projection based on statistics that have the appearance of being tangible, fully predictable facts. It’s something that might happen: a future maybe. Soft Trends can be changed, which means they provide a powerful vehicle to influence the future and can be capitalized on.
Understanding the difference between Hard and Soft Trends allows us to know which parts of the future we can be right about. Hard Trends give us the ability to see disruptions before they happen and the insight we need to create strategies based on a new level of certainty. Hard Trends also provide a way to accurately predict consumer behavior changes based on game-changing technology shifts. Soft Trends can be changed and therefore influenced, producing another way to influence the future.
A simple example of a demographic Hard Trend is the retirement of Baby Boomers. A Soft Trend relating to this would be which companies will implement a system to collect knowledge and wisdom from their Baby Boomers and implement a knowledge-sharing network before they retire.
A simple example of a regulatory Hard Trend would be a U.S. law that was passed in 2013 that allows U.S. chicken producers to ship chicken to China for processing and then back to the U.S. for retail sales with no labeling requirements. A related Soft Trend would be trying to identify how many U.S. chicken manufacturers will process chicken in China for sale in the U.S. Another related Soft Trend would be the amount of U.S. consumer backlash that might occur.
A soft trend is a projection based on statistics and is something that may happen, for example a particular company being first to market or a certain country adopting a technology first.
This trend classification process gives forecasters a good grasp of where technology-driven change is likely to lead.
With anchor metrics, we can see if changes in the market for our product are due to changes in the underlying market drivers. I will take you through a worked example.
Anchor Metric:
A unit which allows us to scale the size of the Segment and understand the market trends e.g. car production.
Production Output Multiplier:
The amount of product used in a single anchor metric point e.g. number of widgets per car.
Total Segment Demand for one time period:
The Segment’s anchor metric is multiplied by the appropriate production output multiplier.
Total Segment Demand over time:
The Segment demand is anchored to changes in Segment size, by the Segment anchor metric. This allows changes in demand over time to be calculated.
Market mapping, also called perceptual mapping, is a great tool for identifying where competitors in the market are positioning themselves. Axes should be chosen which are the key aspects that matter to customers. Clear correlations or clusters are then usually apparent. This means you can identify who a new entrant is likely to take share from, and how big a threat they are to your business, based on where they are positioning themselves.
Information for mapping the position of the competitors on the axes can be drawn from a range of sources. Some, such as pricing information, may be public or well known; some may be more sensitive, and can be gained by talking to industry experts or your own employees who are ex-employees of the competitor in question.
On the market map above, each competitor is represented by their logo. We can see that in this market customers are only willing to pay a higher price for a well-known brand. Therefore, the new entrant shown in purple is unlikely to be successful.
Scenario planning is a well-established technique through which a range of possible futures can be explored. It provides a structured process for consciously thinking about the future environment. The process of scenario planning helps you to understand the critical variables in the market and what might happen as a result.
The 4 steps of scenario planning are:
Identify critical variables
Analyse potential scenarios
Describe each scenario
Forecast the impact of scenarios
Understanding, discussing and agreeing the impact of each possible change in advance will mean you are pre-prepared when these types of changes happen and can anticipate the consequences before they occur.
I hope this webinar has been helpful. Further information on best practice in forecasting can be found in our published white paper called ‘Forecasting the Future for Profit’, which I would be very happy to provide you with a copy of.