Visual vignettes and virtual shopping

When I began teaching about food, I pointed out to my students that market researchers had much better information than social scientists, including studies that looked at how people moved through a store, what eye level was the most compelling for purchases, and whether it mattered if items were up front or back on the shelves. But the virtual shopping experience described by Valla Roth and Matt Draper of MarketTools reminds me that old fashioned social science research techniques can often provide better data than in-store surveys and tests. The system allows potential customers to walk through a supermarket on their home computers. The pictures of the aisles and the movement through the store are facilitated by 3D graphics, allowing the researcher to assess how and when a shopper might pick up a 3 for 1 deal or ignore the end-of-aisle displays.

Long before we had the computer graphics to make it virtual or sexy, I learned a similar technique demonstrated by the late Dr. Peter S. Rossi, head of the Social and Demographic Research Institute at the University of Massachusetts Amherst. Rossi pioneered a new method of research called the vignette technique (or in its less glamorous name, factorial survey), in which different descriptions of a situation are given to respondents, who are then asked a series of questions to elicit individual responses to each description. The vignette technique is great because it allows the researcher to measure and analyze fairly complex scenarios in a way that surveys and linear question-and-answer formats do not. Rossi and his wife Alice (an equally famous sociologist) used it to measure how people perceived their relationships and obligations to family, but vignettes have also been used in studies of AIDs/HIV, religious beliefs, and the delivery of social services to different populations. For all this, consumer market research has shown surprisingly little use of the technique, opting more for focus groups, surveys, and in-store tests.

In a way, MarketTools’ new approach is an “upgraded” vignette method. Giving the vignettes a 3D interface and having the consumer act out inside of a virtual environment combines the vignettes with a much-needed visual element. Since people are better at relating and responding to stories with a images,  we hope the collected responses will be better too. And imagine the next step: perhaps a Wii-like interface where you actually “walk” through the store and interact with objects.  It would be like The Sims for market research (but of course, in The Sims pizza never goes bad and groceries can be delivered at any hour of the day...)

Here are some limits and suggestions, though: No matter how sophisticated our graphics get, they are still a mere representation of reality. People’s main complaint about shopping on line is that the product image on the website does not match up to what arrives in their home. And while the grocery store layout is ubiquitous to most people, there is a certain fudge factor to getting it as realistic as possible, and comparable to what people are used to in the real world. In the end, real interactive video might prove more useful than graphic representations. Also, t one thing this method alone won't mimic the different types of shopping trip that people take -- the data won’t tell you if people on a milk run or a weekly shopping trip are most likely to make point-of-sale purchases. Tying the tech back to storylines and vignettes would help: give people a scenario before setting them lose in the virtual shopping world (“It’s Wednesday, only your teenage daughter will be home for dinner, and you have to be home in half an hour.”).

It’s almost impossible to follow customers around and see how they react to deals without being invasive. As much as I like talking to my informal research group in the supermarket, I don’t learn much about their point-of-sale behavior. But I am much more attuned to their overall patterns and cycles, their brand preferences, and the frequencies at which they buy certain products. I'd love to see how they make decisions in real time, even in an imperfect imaginary realm, so if that's the current state-of-the-art, I'll still take it.

Tags: shopper marketing, market research, shopping

What's in a shopping cart? Only a slice of the pie

Thursday I made a stop at the supermarket during the middle of the day, middle of the week – no market research on senior citizens this time, just a quick trip for things we needed to make it until Saturday. At the checkout I ended up with: navel oranges, apples, cilantro, lettuce, cukes, tofu, eggs, cheese, two kinds of bread, organic peanut butter, specialty crackers, macaroni and cheese, yogurt, and six different kinds of chocolate bars. Only about four things were on my grocery list. Some were impulse buys – but not perhaps what you'd think. All that chocolate was on the list. Some of the produce was not.

But what can you tell about my consuming and shopping future from just this one cart? (Here’s one hint: almost everything, including the crackers, are brands I buy all the time. The only variation was the chocolate – I buy it every week, but always try new kinds . Today they actually had new varieties on sale!). How would you characterize my purchases? What can they tell you about brand loyalty (some), consistency (a lot), and general purchasing trends (produce as impulse buy!!) ? Looking at shopping carts is a tried and true research method, and is always a staple of larger shopper marketing programs. It’s certainly better than consumer surveys. CPG News recently reported (as noted in this RetailWire BrainTrust post) that:

Catalina Marketing has spent two years examining 250 million shopping baskets weekly from 130 million separate shopper identifications. The goal is to probe the gap between what shoppers say they buy in surveys and what they actually purchase.

I probably wouldn’t mention those chocolate bars in a survey. But the shopping cart in isolation – one day, one cart – is as problematic as the shopping survey. One of the things I’ve been studying for years is everyday meals – how frequently people eat at home, what they cook, what they get as take out or prepared meals, who they eat with. The most important thing I’ve learned is that people live their lives according to patterns, but one slice won’t give you even the slightest idea what any given day might look like. For example, the Sunday dinner is different from but dependent on its contrast with the weekday supper, the holiday meal captures the extremes, and there are other punctuation marks all along the way.

Shopping carts are going to reflect patterning, too. Mine was a Wednesday cart on a week when someone had done the week’s shopping on the previous weekend. But we eat a lot of produce and always run out midweek. I’d also exhausted my chocolate supply, which sometimes (but not always) happens too. But if you looked at our cart over a month, you would get a very different picture – where do we run to for a loaf of bread or eggs? How much do we shop at little local markets or at big supersavers like Costco? What happens on a week when we do get down to the Saturday morning farm and specialty markets (the Strip District, for those who know Pittsburgh). How much did we eat out this week? Who is doing the cooking and who is being fed (the mac and cheese was for my daughter’s friend, who never eats anything else I make)?

Certainly many market researchers worry about whether people will tell “truth,” but truth is not in a snapshot. Rather it's in multiple images, over time and place (the whole pie!). More importantly, shopping carts are the repositories for later use. If we don’t find out how people are making use of the products, we don’t understand their purchases. Shopping cart data may be rich, but it's ultimately less useful if we don't map patterns of food use along with food purchases.

Still, watching shopping carts is particularly important right now as two consumer trends are butting up right against each other: one, the growing concern with organic, natural, and “green” products (which still tend to be a bit more expensive in most markets) and two, the rising cost of food in general, which tends to push people more towards bargains, coupons, and other sale items.

Mapping patterns seems like a lot of work, but without it, you’ve got tons of data and no anchoring scheme. The distinction between shopper and consumer is too simplistic to capture what’s going on (on Wed I was both and neither!!!). It also needs to be matched against what’s on sale, what’s out of stock, and what’s new in the store (seasonally, for example).

Shopping cart data needs to be put into a meaningful framework in order to make sense. If the data comes without questions and information about how the products are about to be used, all we know is what people have bought. There’s no insight into why.

But they've probably figured out that I buy a lot of chocolate...

Tags: shopping, shopper marketing, market research