The Pepsi Challenge - Could They Use ‘Virtual Store’ Simulations to Predict Shopper Behaviour?

by Michael Letchford on April 28, 2009

Although Matt Kinsman’s article doesn’t record the beginning of Fifth Dimension’s Virtual Store technology developments as suggested by the writer (that’s covered in this article), it’s the first information ever published in the media about our efforts in this field.

Using computer generated imagery and virtual environments to realise design concepts had been part of the company’s core skill set for many years, but landing a series of projects with the US arms of both Procter and Gamble and PepsiCo in late 1999 and in 2000 respectively put us on a new path.

Studying Customers’ Behaviour in a Virtual Store

Generating realistic and interactive retail environments that made it possible to research consumers’ reactions to new ideas was still in its infancy in those days. We had learned that consumers often claim to behave in planned ways when shopping, but it was obvious to us, and well known in research circles, that that’s far from the truth. In fact, consumers react in all sorts of ways when faced with varying stimuli and so we wanted to create virtual store environments where we could explore this behaviour under more controlled conditions than in a real store.

Clients were also aware of the many potential advantages this approach would give them. Not least the abiity to test new ideas in private without alerting their competition to their latest initiatives before they were ready for market. So it was that in late 2000 we began discussions with PepsiCo on a range of projects, some of which are described below as illlustrations of what was possible in those days.

Virtual Fast Food

The first of these projects, without giving the game away, was PepsiCo’s idea of helping drive-through food service providers increase their throughput of customers by using concurrent queues in managed traffic flows coupled with innovative ways of helping customers choose their meals from innovative menu systems, ordering their preferences more efficiently and collecting and paying for their meals by various innovative system options.

We were asked to create a virtual drive through ‘fast food’ outlet so PepsiCo could test the various options and see which was the most efficient and acceptable combination to the respondents. The goal being to increase the number of customers who could be served in peak periods without compromising quality of service.

After several months of data capture, preparatory computer modelling and software development work, by January 2001, we were ready for the concept to be tested at multiple locations in-field with selected respondents. The final results were, of course, PepsiCo’s business, but the test was very successful in determining the best option. It is important to note that a virtual approach was taken because, as you might guess, it would have been almost impossible and, in any event, far too expensive to test these ideas in the real world.


The video shows one of the options being tested in the drive through scenario.

A 30,000 Item Virtual Store

A more ambitious project was then built for PepsiCo in early 2002. This was a full 14 aisle, 30,000 item, Virtual Grocery Store. The project was built in what had become our formal virtual consumer research environment, Discovery. Discovery was engineered by Fifth Dimension, to make it possible for many different kinds of consumer research to be conducted in a virtual store environment with maximum flexibility, at modest cost, in-field, using ordinary shoppers just as they entered their store of choice.

The grocery store test was designed to help PepsiCo determine the best possible merchandising strategy for the ‘in-between’ zone in the carbonated soft drinks aisle. The so-called ‘in-between’ zone was actually the product display between the space dedicated to Coca Cola products and those from PepsCo. It was felt that it should be a full basket shopping trip, which meant that respondents would be shopping in the virtual store for up to forty minutes, something that had never been attempted before.

diet-pepsi-twist-12-packDuring the early months of 2002 we imaged countless products, modelled many thousands of items, built the store and all its aisles and created what we believe was the first complete Virtual Store, of that size, of its time. Much of the product displayed in the store was derived directly from the relevant category planograms.

Since the initiative was to be tested in ten centres around the US, regional pricing was very important for the validity of the study. Information Resources Inc was asked to provide the regional pricing data for all shoppable products. It took three weeks of data analyst’s time to clean and restructure that pricing data alone. By May we were ready for the field trials.

consumer-research-pepsi-virtual-grocery-store

Once again the Virtual Store approach made a study possible where, in the real world, it would have been extremely difficult to persuade a retailer to disprupt ten operational stores to acommodate the changes necessary to accuratetly test customer reaction to the various merchadising strategies explored by PepsiCo in this study.

Virtual Convenience Store

In late 2001, PepsiCo were also interested in the opportunities available for increasing the sales of carbonated soft drinks in convenience stores. They were pursuing an initiative related to the use of drink dispensing fountains among other things. To help test these ideas and their potential implementations, we were asked to build them a Virtual Convenience Store. The study went into field tests in July 2002.

This project gave us the opportunity to do a great deal of advanced analysis. It was the first time we fully explored customer flow patterns, sequential item purchase analysis, customer dwell time at the point of selection and the role of point of sale signage. A notable innovation was the testing of various point of sale signage variations, displayed in multiple locations, which were viewed by respondents as they filled their virtual car’s fuel tanks with Virtual Gas!

The video snippet shows the initial drive in to the store forecourt and the customer’s arrival at the door.

A copy of Matt Kinsman’s article’s text is available here for your convenience:-

Better Than the Real Thing - Text

Alternatively, if you would like to download a pdf of the original PROMO magazine article, click on the link below:-

Better Than the Real Thing

Leave a Comment