This document discusses a method for optimizing product concepts through dynamic market research. It involves starting with many possible combinations of product features and levels, evaluating concepts in steps to identify preferred levels, and narrowing down the possibilities until reaching an optimal concept. By removing worse performing levels at each step based on regression analysis, the method learns more about preferred concepts without respondents needing to evaluate all possible combinations. This allows developing a winning product concept based on evaluations from a sufficient number of respondents, while avoiding the "Frankenstein effect" of an incoherent overall concept.
2. When thinking of
developing an
innovative product,
many features can be
varied to build an
optimal concept.
For a phone, you can
think of coverage,
camera type,
waterproof option,
storage…
3. A typical market research solution is to combine
different levels of each product feature to create
concepts, to have them evaluated by respondents
and to estimate the value of each level.
SD
5 MP
SD
4 MP
SD
5 MP
4. When it comes to accumulating the different
product features and levels, the amount of
combinations quickly adds up.
This means that each individual concept might be
seen by very few (or no) respondents.
5. In some areas, that’s not a problem : for example
in telecom. Putting the best levels of all product
features together, will give you the best phone.
6. But when it comes to optimizing FMCG
products, the overall concept needs to
make sense and drive a coherent
message which fits the brand image.
7. Treating your concept as a
sum of perfect levels is a
risk, and could lead to a
“Frankenstein effect” as
nobody - or very few
respondents - really
evaluated the final concept.
8. This situation arrives very quickly as
already 4 features with 4 levels leads
to 256 possible combinations…
9. NO LIMIT
As we are in the early
stage of the innovation
process, this is the time
where creative ideas
should have their place.
No constraints should be
made in the amount of
levels and features to try.
10. In a nutshell, our dynamic optimization method
is about considering all the possibilities (all
features with all different levels), while finding
the best concept that has been evaluated by a
sufficient amount of respondents.
11. Our solution is to break down the optimization process into steps.
Each step consists of creating a few concepts, having them evaluated
by respondents, running a regression to identify the worse levels and
removing them from the set up.
By doing so, we learn more an more at
each step on the preferred concepts.
12. In the very last step, all final (10)
concepts are being evaluated by a
sufficient set of (60) respondents.
This allows a very good read on the
winning concept.
10
60
13. So far for the theory.
In practice, we program all this stepwise approach in advance. We
determine sample size, levels to be
removed and experimental designs.
So once we go into field, our
optimization algorithm works on his
own until it gives us the solution.
14. Let’s go through a case study
regarding an anti-parasite
product packaging.
15. The goal was to optimize their
current packaging.
16. Design, color, cat picture, claim, logos… Many pack
features could vary, leading to 108,000 possible concepts.
17. How likely would you be to
purchase this product?
Respondents
were facing a
5 minute
evaluation
survey with 10
screens with a
different pack
to evaluate.
Extremely
unlikely
Neither
Likely nor
unlikely
Extremely
likely
18. In the end, with 400 respondents, the possibilities were narrowed
down in a step-wise process to an optimal concept which proved
to be significantly better than the current packaging.
19.
20. At the end of this optimization process you will
have a clear recommendation on the best
concept (or set of winning concepts), with
performance indicators for all features & levels.
Additionally, qualitative insights on why and
how can be provided for the winning concepts.
21. Respondents can click on concept elements and provide
(open end) feedback on what they like most and least.
+
-
22. This dynamic optimization approach has multiple
applications, anything combining text and visuals.
PRODUCT
CONCEPT
BROCHURE
ADVERTISEMENT