Thursday, June 2, 2016
2:00 - 2:20 PM
Ted Talks - Duluth/MacKenzie

Down with top two box scores

Predictive models are one of the top emerging methods in market research. But how many of us know what predictive models are? How many are using them to their full potential?

Concept testing traditionally involves exposing consumers to a single idea and asking a series of standardized questions (e.g. purchase intent, liking, value for money, uniqueness). The reporting involves lots of top two box scores. If you are luck (or unlucky), you get comparisons vs. a normative database. There are three fundamental issues with this kind of research:

  1. Monadic concept tests are a completely different cognitive task vs. shopping, which is focused on trade-offs between competing alternatives.
  2. Because monadic tests don’t understand trade-offs, they cannot meaningfully project the effect of innovation on the broader category. To what extent does the innovation cannibalize the existing business, steal share or grow the category?
  3. There is a huge chasm between the data that results from this kind of research and the data that businesses actually need – projections of units, dollars, profit, source of volume and share.

Let’s kill traditional concept tests and replace them with predictive models that:

  1. Treat innovations as trade-offs vs. existing offerings;
  2. Allow us to understand interactions, cannibalization and incrementality; and,
  3. Generate business projections, not research data.

Interest Statement
With the growth of computing and analytical power, it is possible to now test innovations in a fundamentally different way. We can create experimental markets where we manipulate all of the levers that can be managed in the real market: package design, package size, SKU mix, brand mix, size mix, regular price, depth of discounting, frequency of discounting, distribution, planogram design, etc. We can translate all of those variables into projections of units, dollars, profit, cannibalization and share.

This kind of methodology changes the role of research in the business dialogue. We move from …

…“do consumers like the new package more than the previous package?” to “what is the dollar volume effect of the new package?”

…“how does the new product idea perform vs. norms?” to “does the new product generate sufficient trial interest vs. existing products to justify the investment?”

… “what would happen if we went on sale more often but at a less deep discount?” to “if we went on sale two more weeks of the year, but at $0.50 off instead of $1.00 off, how would that effect unit volumes, dollar volumes and profit?”

People will want to attend this session so that they are equipped to participate in this new dialogue.

As a result of this session, participants will understand:

Participants will be exposed to a series of case studies in which predictive modeling was used to translate product innovation, package size, product mix, pricing strategy and claim mix into
Michael Edwards
Co-Founder, Dig Insights

Michael is one of the four founders of Dig Insights. He has worked on both the client (Kraft Foods) and supplier (Environics, ACNielsen and Maritz) sides of the research industry for 20 years. This gives him perspective into how research can effectively support smart, effective decisions. Michael's greatest areas of focus are portfolio management, pricing, product attribute optimization and understanding the effect of design on brand/product performance.

 


Parul Verma
Consumer Insights Manager, Coca-Cola Ltd.

Parul Verma has 10+ years of experience in market research and is currently a consumer insights manager at Coca-Cola Canada. She has also worked on the supplier side at Ipsos, Northstar Research Partners. In her current role, Parul influences growth strategy for the brands she supports and has a perspective on how actionable research recommendations can influence business decisions.