Back to the drawing board

Formulating, testing and launching a new product is never easy. Dunnhumby’s Martin Hayward asks whether the market research business can add value at every stage of the development process

“Innovation is the central issue in economic prosperity” said Michael Porter, and it’s hard not to agree with him. The trouble is that it’s just so difficult to get it right.

Conventional wisdom tends to suggest that only one in ten new product launches eventually goes on to long-term success. Looking at some slightly harder facts, in a recent 18-month period over 6,500 new supplier products have been launched in Tesco stores alone, and that excludes seasonal lines and minor product tweaks. Of these 6,500, nearly 2,000 have already been delisted within a year and only 400 have reached peak weekly sales of £60,000 or more.

Obviously it is impossible to get it right every time, and trial and error, or ‘Test and Learn’ as we call it at dunnhumby, is an important learning cycle, but the current success rates are so bad, and so expensive to sustain, there is a constant appetite for better tools to inform the process and improve the odds and return on investment.

However, the rather good news is that it does look as if the odds are about to turn and that we are on the cusp of a new era of marketing insight that will provide better, more accurate and more timely information to inform all stages of the product innovation process.

At the heart of this new era lies a number of quite dramatic changes in what it is possible to find out, how quickly that information can be provided and how information can be joined up.

We know more than we did
It used to be hard to find out what people did. We had to ask people to remember what they’d bought. We had to ask them to remember what they watched and listened to. And because this was such an expensive exercise, we had to rely on small samples of them to be representative of the population as a whole. For many macro marketing issues this was, and still is, fine. Big behaviours tend to move slowly and a decent sample can be predictive of the population as a whole.

But, for most brands and products, and especially new products, the target audience is a fraction of the whole. Very few brands ( 10.5%) enjoy a penetration of over 5% over a six monthly period, rendering smaller samples inaccurate in understanding the nuances of the majority of consumer behaviour.

The arrival over the last few years of new sources of behavioural insight derived from shopper loyalty schemes has solved many of these problems. Samples of literally millions of shoppers’ behaviours allow even the smallest brands to be analysed in extreme close-up every day, creating a fact based context for subsequent marketing decision making.

Looking forward, this step change in insight granularity will accelerate rapidly into other areas of research over the next few years. Media consumption data is already beginning to shift in a similar way from survey-based approximations to real-time analysis of actual behaviours as media delivery becomes more direct and therefore measurable. The digitisation of financial transactions, communications, location and so on will continue to provide a richer, fact-based insight into actual consumer behaviour.

Additionally this real behavioural data makes it much easier to recruit relevant respondents for concept testing or other exploratory research. Rather than trying to find regular users of particular products by random sampling, companies such as dunnhumby can offer direct research access to accurately qualified respondents who have given their permission to be contacted for research purposes. Recruitment to research can even be triggered at the time and point of purchase through the giving out of telephone, or internet-based research invitations on till receipts.

We not only know more, we know it faster
In addition to a clearer picture of current behaviours and trends, we can also rely on much faster insight. This is particularly important in the early stages of a new product launch. It used to take months to get an accurate sales read of a new product, due to its low penetration. Today we can see not only how many of the product have been sold but, more usefully, who’s buying them, what else they buy and – most important of all – whether they buy it again. A panel of say 13,000,000 shoppers in the UK should be able to provide detailed insight into even the most niche of products within days. Large scale, customer-level data is revolutionising the speed with which we can assess the likely success of new product launch.

The sooner one is able to see who’s buying or not buying a new product, the easier it is to flex the launch plan to better target communications and offers to the right consumers.

Dunnhumby’s analyses of many thousands of new product launches demonstrate that the single best predictor of a new product ‘Star’ is, with reassuring logic, the repeat buy rate achieved. This can only be measured quickly and accurately by large-scale, customer-based panels.

Connecting the dots
The final new and much needed development in research capabilities lies in an increasing ability to ‘connect the dots’ between different types of data and across time to help with the development and launch of new products.

Traditionally we have had to attempt to reconcile different data sources used throughout the development process. Inevitably this creates discontinuities and errors as the different samples and results from a qualitative exploration exercise are then used to inform a concept test which is then applied to a test market measured with retail audit data.

We are now approaching a new era of connectedness as true single-source opportunities emerge to create a seamless end-to-end solution for new product development. This will improve both the accuracy and speed of outputs.

Basing the development process on the vast behavioural resource of multi-million, household retailer loyalty panels, we are beginning to see the benefits at every stage of the product development process.

Explore phase
• Analysis of millions of shoppers’ behaviours provides the broadest and deepest fact based landscape within which to identify gaps in the market and trends in purchasing behaviours.

Identify phase 
• Quantifying potential opportunities requires granular understanding of household purchasing behaviours

Test phase 
• Actionable concept testing is reliant on the quality of respondents recruited. Again, behavioural panels provide opportunities to expose concepts to highly targeted individuals. If the concept is particularly appropriate for customers who, say, buy lemons, the Daily Mail on Wednesdays and organic chocolate bars, then it can be exposed to these shoppers. 
• The potential for closer collaboration with retailers through their panels creates better opportunities for in situ market testing.

Launch phase 
• Closer collaboration with retailers again pays dividends for optimising launch success. Faster market feedback through rapid reporting of behavioural data allows the launch to be amended and optimised in real time.
• Increasingly sophisticated usage of retail media allows messages to be targeted and measured accurately both in-home via mailings/email; and in-store, via TV, posters, floor stickers.

Overall we can look forward to a new era of enhanced efficiency within the new product development process. Better understanding, faster feedback and joined-up development, launch planning and analysis should enable us all to shorten the odds of success.

Marketing can be defined as making sure that what is produced is matched to what consumers value. Until now this has been more of an aspiration than a reality because of the difficulties of actually identifying, with enough clarity and quickly enough, what it really is that consumers value.

Even if we did manage to paint a clear picture of their needs, it was yet another hurdle to get awareness and distribution of the product targeted sufficiently well to ensure success.

Today, as I have highlighted above, there are exciting developments that overcome the challenges of accuracy, speed and implementation to give new products the best possible chance of success.

Of course, the odds will continue to be against the success of any new products. Eight of today’s top ten confectionary brands were in the top ten in 1959, underlying the inherent stability of many categories over time. However, nothing ventured, nothing gained remains as true as ever.

Ironically, in many ways, we could perhaps expect to see more new product launches in the future, rather than less because the cost of development can be more efficiently managed and with faster feedback loops, mistakes can be identified more easily and removed.

July | 2008