What are the pros and cons of virtual shelf display conjoint?

Virtual shelf display conjoint promises an opportunity to make the survey more realistic for respondents, but what are the downsides and is it actually a better tool for decision-making?

The Conjointly platform natively offers the “normal” choice-based conjoint, however our team have had experience with conjoint studies in a virtual shelf. Naturally, virtual shelf displays look more realistic, but we have not seen clear evidence in practice or in academic literature to say that they are better at predicting market outcomes than a “normal” conjoint.

What are the differences in setting up?

  • Potential number of SKUs:

    • In a normal conjoint, we would put up to 5 SKUs per screen.

    • In a shelf display, we can increase the number of SKUs to 10 or even 30, but would still want to show several shelves to make the shelf more realistic for respondents.

  • Time to prepare:

    • It is fair to say that a normal conjoint can take a couple of hours to prepare, or up to a day if multiple images are involved.

    • A shelf display conjoint takes at a minimum several days.

  • Only desktop devices are suitable for shelf-display conjoint studies

If there are more SKUs present in shelf display conjoint, wouldn’t it be more difficult for the model to estimate the importance scores for attributes and levels?

Broadly speaking, these are the factors to consider:

  • Choice modelling (conjoint analysis, both standard and virtual shelf) typically makes the assumption of independence of irrelevant as alternatives, which means that the number of inferior SKUs offered in a set with a superior SKU will not affect people’s choice: whether there are 4 or 11 inferior SKUs in addition to the superior SKU – it suggests either approach is fine.

  • An exercise of 12 SKUs is more realistic in terms of the number of SKUs offered (typically you have closer to a dozen SKUs rather than 5) – it suggests the virtual shelf display is better.

  • Because for every choice set we record one choice, having a larger set of offerings has the benefit of better discerning the top option – it suggests the virtual shelf display is better.

  • For the same reason, a larger choice set means that we fail to recognise close seconds (which can be the winning option if the top option is excluded) – it suggests the standard conjoint is better.

  • Many respondents will not have the attentiveness to review all 12 options, making the results noisier – it suggests the standard conjoint is better.

Thus, there are both pros and cons in this consideration.

When is it better to do a shelf display conjoint?

  • When you are assessing allocation of products on shelves (and especially if you are making decisions in relation to paying for shelf space in supermarkets).

  • When you are assessing the overall impression of the brand on the shelf.

When not to do a shelf display conjoint?

  • When time and costs are limited.

  • When your product are complex or unfamiliar to respondents (because they will not have patience to explore every product by reading its description).