Synthetic Customers: PwC and Bain on a New E-commerce Tool

The e-commerce industry is forming a new approach to studying customers — synthetic customers (synthetic customers, or synthetic buyers): artificial intelligence models that imitate the probable reaction of real customers to price, advertising, interface, and the product itself, as discussed by analysts from Bain and Market Logic.

How this differs from classical analytics

Classical e-commerce analytics — what the buyer searched for, which product cards they clicked on, where they abandoned their cart, how they reacted to discounts, and which products they bought together — has long been the foundation of online retail, but it has a fundamental limitation: it almost always explains what has already happened. A new price, advertising creative, product card, or launch of a new category are usually tested "in battle" — when the budget is spent, the product is launched, and the team is only beginning to understand how customers react to it. As noted at Market Logic, in many companies customers change faster than product roadmaps and results of classical research are updated.

What are synthetic customers

Synthetic customers are AI models that imitate the preferences, doubts, and attitudes toward price, advertising, interface, or product inherent in real customers. Such models allow you to test scenarios in advance: how the audience will perceive a new product, whether a specific promotion will inspire trust, at what step a user might abandon the purchase, and why a certain segment does not see value in the offering. PwC calls this approach an "invisible focus group" (invisible focus group).

• The term is used by analysts at Bain (synthetic customers/synthetic buyers) and PwC ("invisible focus group")
• Market Logic notes: in many companies, customers change faster than product roadmaps and results of classical research are updated
• Synthetic customers test hypotheses before launch — based on price, advertising, product card, new category
• The approach is described as a shift from the model "launched — looked at the result" to the model "first checked the reaction — then made a decision"

Where the risks and limitations are

The approach itself does not eliminate the need for real research — it complements it at the stage when a full test on a live audience is not yet possible or is too expensive. The synthetic customer imitates the probable reaction based on accumulated data about real customer behavior, so the quality of the forecast directly depends on how representative and fresh this data is. This is why in materials from Bain and Market Logic, the approach is described as a way to test hypotheses faster and cheaper at early stages, not as a complete replacement for traditional surveys, usability tests, and pilot launches.

A similar term — "synthetic users" — is used by NN Group, which specializes in UX research, showing that the idea of modeling user reactions through AI is being discussed in several adjacent disciplines at once, not just in e-commerce.

What this means

Synthetic customers are not just a new type of analytics, but a new way to deal with uncertainty in a market that changes faster than classical research can keep up with. For e-commerce, this means the ability to test strategy before market entry — not only through surveys and pilots, but also through simulation of customer behavior based on data — rather than only post-factum analyzing the results of sales that have already occurred.

Frequently asked questions

What are synthetic customers?

These are artificial intelligence models that imitate the probable reaction of real customers — their preferences, doubts, attitudes toward price, advertising, interface, or product — in order to test a hypothesis before market entry.

Who introduced the term "invisible focus group"?

Analysts at PwC call the synthetic customer approach an "invisible focus group" (invisible focus group) in their material on the application of AI in retail.