Retail Packaging Optimisation using Discrete Choice Modelling
First published July 2015
A packaging experiment for the organic chololate wholesaler, “Chocolate & Love” showed that even small layout changes have an impact on choice.
Results show that the best layout had twice the preference of the worst. Actual results are reproduced below with the kind permission of Richard O’Connor of Chocolate & Love.
Chocolate & Love prides itself on producing superior chocolate. As a new entrant growing into a crowded market, they faced the classic packaging problem – what to put on the box that will give it the best chance of being sold?
A working packaging was in place, but there was still uncertainty around:
- The best placement of the brand.
- The importance of the origin.
- Whether “Fair Trade” should be included on the packaging.
- How the claim of “organic” influences choice.
- Whether the product be “80% Panama” or “Panama 80%”
The True Size of the Problem
Consider for a second how you yourself would solve this problem. Looking through some of the combinations below, there are over 60 possible combinations, its not easy to see which is the best option.
In fact looking at each one in turn and trying to pick the best one is an impossible task. It is also an unrealistic exercise to imagine how a customer, who only ever sees one version, would choose.
Polling all combinations in turn would be possible, if not exceedingly expensive, and the best overall package could be found. However, it would not explain how those choices were made.
The combinations of layouts were provided to SurveyEngine which then generated a short online experiment. Data were collected from respondents and the data was modelled to understand how choices were made.
The model below showed that people are choosing primarily on the “Organic” claim and a logical layout. The claim of “Swiss” origin was either irrelevant or, at worst, negatively distracting.