Marketing Research Blog

What is the right mix of our products to maximize reach for a new retail partner?

You could possibly look at scanner data to see which products generate the most sales, but that doesn’t necessarily mean they have the greatest combined reach. To do that you likely need to do a TURF (Total Unduplicated Reach and Frequency) analysis using a MaxDiff (Maximum Differential) exercise. Better still, you could do a choice-modeled shopping exercise to (hopefully) help the retail partner understand how much lift they would get from including more of your products and fewer of your competitors.

Do I need to do market segmentation to understand who my brand’s best customers are?

Maybe? Do you have a segmentation scheme that helps you understand your category and your customers? If not, segmentation research could shed light on different groups of consumers and who you should be targeting and how. A good consumer segmentation will shed light on the emotional and rational drivers of purchase, which will impact your approach on how to engage with the segments.

If yes, are you using it? Consumer segments should have profiles that describe them demographically as well as attitudinally and/or behaviorally. Laying brand usage on top of consumer segments will indicate who your brand’s best customers are. Finally, you may just need to do some basic consumer profiling to understand who your most frequent customers are, and why they prefer your brand over competitors.

How can I understand what consumers are really saying about our products?

This could be a loaded question, but the gist of it is this. How do you integrate qualitative richness with quantitative robustness? Fortunately, there are several ways to address it.

• First, you can always do multiple phases of research that include both qual and quant studies. The order will depend on whether you need to explore before you validate or if you want to drill down into specific areas.
• Second, you can use a hybrid methodology that includes aspects of both qual and quant in one study. These are done with guided platforms that use AI moderators to ask questions and probe answers from dozens to hundreds of participants at once.
• Third, although it’s not generally recommended due to the time and expense required, you could just do a whole lot of qualitative interviews or focus groups.
• Fourth, some quantitative survey platforms are now integrating limited AI moderation of specific open-ended questions to allow for probing on typed responses.

How do I optimize my bundles of products or services to maximize revenue?

Whenever I hear a question that indicates we are changing more than one variable at a time (products, prices, brands, etc.), I primarily think about choice modeling. It’s the gold standard for managing multiple variables at the same time. If the number of variables and levels (2 brands x 2 price points x 2 SKUs, etc.) isn’t great then we could consider other methodologies like MaxDiff. Normally, though, we want to include competitors in the mix and doing that greatly increases the scope of the research, which takes us back to choice modeling. Choice modeling also works well in testing different bundles of products, including levels of bundles like product tiers that are built upon one another. It is a very powerful tool and highly customizable. That’s why it’s the gold standard.

How do I know which ads, messages, or concepts will yield the greatest ROI?

The good news is there are many viable approaches. Typically, I would lean toward monadic testing where each person reviews/rates only one execution. This is the cleanest approach since participants aren’t seeing multiple concepts, ads, or whatever. There isn’t any exposure or order bias based on seeing other versions. But monadic costs more because you must have balanced samples for each execution you are testing. Sequential monadic is where each person rates/reviews all executions one at a time in random order. It greatly reduces the number of participants you must recruit and pay, but it increases the survey length, which can be problematic and cause respondent fatigue (not to mention exposure and order bias). Proto-monadic is a compromise between the other two methods; participants thoroughly evaluate one execution but are then shown all executions at the end and asked to select their most preferred and indicate why. You still need large sample sizes as with monadic testing, but it provides you with direct comparisons between executions which can be helpful in certain situations.