CPGmatters: Heinz Optimizes Assortments Using Virtual Shopping Platform
By John Karolefski
Through a special arrangement, what follows is an excerpt of a current article from the monthly e-zine, CPGmatters, presented here for discussion.
Heinz wanted to simplify its planogram in stores. Three kinds of packaging – traditional bottle, the Upside Down with Stay Clean Cap, and the Fridge Door Fit in multiple package sizes – often confused the consumer and resulted in increased production costs for Heinz. For solutions, the company turned to Decision Insight’s virtual shopping platform called SimuShop that enables product manufacturers to test options on the shelf.
Heinz worked with Decision Insight to find out what would happen if certain SKUs were removed from the product assortment.
Here’s how the program works: A panel of qualified consumers was selected and a scenario was set up for a shopping trip. A combination of text and video cues set the context for an online shopping experience. Video takes shoppers from the parking lot into the grocery store and into the aisles. They can click on a product for more information such as size and price and to make a purchase. Consumers could buy or not buy any product, or even walk away without any purchase (this determines the “walk-away” rate).
“Virtual shopping is very powerful,” said Barbarita Marbelt, Heinz’s consumer and customer insights manager, speaking recently at the Shopper Insights Conference in Chicago. “You are observing shopper behavior rather than asking what they would do. Asking questions is not enough. We went with virtual shopping to see differences in behavior because what people say is not what they do. That makes the methodology more reliable as well as faster and more economical than in-market testing.”
Previous research indicated that households with larger size bottles of ketchup use it for more occasions. That is key for a category with a penetration rate of 96 percent. Rather than focusing on attracting new users to grow the category, the obvious strategy for Heinz would be to get current consumers to use ketchup more often. Such thinking led to a focus on keeping the larger Fridge Fit pack in assortments.
Online testing of this strategy resulted in two conclusions:
Overall sales went down when the Upside-Down bottle was removed. There was a significant loss in buyers, volume and dollar sales.
Removing the smaller size bottles posed risks. While most buyers remained with Heinz, a small minority switched to a national competitor.
In the follow-up study, findings indicated Heinz could reduce pack types from three to two by keeping the Upside Down bottle. Plus adding a smaller sized Upside Down bottle kept more customers.
Virtual shopping also indicated that reducing the set of products made it easier for the consumer to shop the category and increased overall category sales that benefited the retailers. Along the way, Heinz was able to reduce production costs while increasing brand sales as the category grows.
The online research was later confirmed with in-store testing at Wal-Mart and a grocery outlet. “The fact that it validated the online behavior of shoppers with similar results tells me that if I have to do something quick and economical, I’m going with virtual shopping,” said Ms. Marbelt.
Discussion Question: What do you think are the advantages and disadvantages of using online virtual shopping in determining planograms? How would you rate it versus other methods of gauging actual consumer purchasing behavior at the store level?
Typically at Decision Insight, we see very strong correlations between the virtual shopping sales numbers and actual sales. Our correlations are above .9, indicating a strong degree of accuracy in our estimates. So, as Liz Crawford comments above, “virtual worlds mimic real world behaviors.”
As for qualified respondents, our clients define what counts as a qualified consumer, usually based on how often they shop the category or purchase a product in the real world. More often than not, our respondents are highly familiar with the category we ask them to shop.
I’m still hesitant to give up on the old method of testing the real thing in a few stores and seeing what happens.
Virtual shopping might work well with qualified and loyal consumers to test their propensity to switch among the various ketchup products offered by their favorite brand, but switching to a different brand of something you put in your mouth can be an emotional decision.
I’m still a fan of trying a new theory virtually to ensure it will fit and meet basic merchandising rules – then move to a live test to valid your plans.
There is mounting evidence that virtual worlds mimic real world behaviors. The question is, who qualifies as a user and does the sample represent real consumers?
This seems to be the wave of the future for research (sociologists and psychologists are using these kinds of mechanisms now). Refinement of the methodology seems to be the main issue for accuracy. In my mind, the key is to validate the findings with real feedback to prove the results, at least initially.
Virtual shopping is a different exercise from real shopping and — as such — ought to yield different results. How different only time will tell.
Interscope has had terrific experiences with the SimuShop methodology. This Decision Insights’ methodology provides a valuable blend of high technology and practicality that we’ve recommended to quite a few clients. In each case the shopper insights were clear and compelling to direct which aisle layout or communication materials showed the most promise. We then married this insight with retail analytics to understand the size of the prize for retailers and brands.
In-store research is still the gold-standard. However, virtual is catching up and and provides a much more budget and time friendly alternative. Just as online research is routine now for consumer learning, online shopper research will be increasingly accepted across industries.
Before I comment, I would like to know what a ‘qualified consumer’ is! I haven’t seen much about virtual planograms but I still use pilot store testing. I think it’s important to actually see the customer’s reaction and action in the aisle. Not sure you can get that from a virtual model although it does sound neat.
My company, has been doing Virtual Reality shopping for 15 years – we’re pretty confident about our ability to predict, with good accuracy, what will happen when you change prices, promotions, packaging, shelf assortments, shelf layouts, or add new products into the mix. There are several proprietary keys to doing VR research well – our clients know these because we don’t let them violate these principles or the results will become unprojectable.
For those who doubt, fear not – we have been doing validations for years on end. But keep in mind, not everything is amenable to VR testing. We do not do in-store interactive media, for example, because we do not think you can capture the impact correctly. Likewise, we don’t do products where touching/feeling/smelling is relevant (perfume and clothing, obviously, but also categories like shampoo and deodorant in the Middle East).
As with any research technology, it’s a buyer beware market. And remember, just because it’s online does not make it virtual reality.
Great market research is often quite simple and inexpensive, compared to marketing failures. You can use virtual reality or paper questionnaires or focus groups, etc. Most value creation: the research topic. Least value creation: the specific technique. If you’re asking the right question, you’ll get value. If you’re asking about trivia, you’ll get nothing with any impact.
Virtual simulations have tremendous value in doing “what if” investigations, checking one format against another, or evaluating a new idea compared to the status quo. Results can drive great insight. However, before rolling out a new plan based upon the simulation alone, doing a test in the market (a store, a region, a group of stores, etc.) would be a good idea. While the simulation provides tremendously useful information, the consumers in the panel may or may not represent all your most valuable consumers around the country.
Whenever you try to measure a consumer’s response, you risk the pitfall of the results not being projectable. Most simply, the subject who participates is not your generic shopper. There is always something different about anyone who wants to actively participate in behavior measurement methods.
However, the example of optimizing the Heinz product offering is perfect for the use of this type of tool. While it would be foolish to project absolute sales of the products being measured, it should be very projectable to measure the difference in the two product offerings.
Further, if the number of variable (products) being compared at any one time were limited in number, this tool could work very well for retailers’ plan-o-grams as well. If I were a retailer, I think the first option I would test, by category, would be the second level of private label products noted in yesterday’s questions.
In making decisions about planograms, many variables will influence shopper choices. Brand marketers consider shelf sets and package placement carefully. Online virtual shopping testing can screen out a number of possibilities, then assist in refining the test hypothesis more efficiently. Several brand marketers have reported good results with this system, and are working to make it more realistic for shopper panelists. There seems to be agreement that this technology is predictive, very useful in designing planograms, but must be verified with actual in-store comparisons.
Like many of the other respondents I would want to know more about what “qualifies” a customer. I can see an advantage to doing this type of study prior to doing a live test market and going through the expense of start-up for a new product or line extension. With current economic pressures on manufacturers, changing or streamlining SKUs is becoming increasingly important in the profitability of a company.
Virtual Shopping Studies could provide some cost effectiveness to streamline or adding SKUs in a logical progression of steps. But a test market would still be necessary to provide real world scenarios that involve stock levels and shelf height placement of the product and the customer’s reaction to other real life environment situations.