A growing share of purchase research now happens inside a chat window rather than a search results page.
The 2026 South African Customer Experience Report, produced by Rogerwilco, ovatoyou and Julia Ahlfeldt CX Consulting, found that consumer use of AI search tools such as ChatGPT, Gemini and Perplexity nearly doubled in a single year, moving from 12% in 2025 to 23% in 2026.
On the business side, only 27% of organisations report having a dedicated strategy for managing how they appear in AI-generated answers, with a further 27% still exploring the idea. For an ecommerce business, this gap between where customers are starting their research and where the business has invested its visibility efforts is worth taking seriously.
What changes when the search bar becomes a chat window
Traditional search returns a list of links and leaves the comparing, filtering and deciding to the shopper. An AI assistant does that work for them. Ask it to recommend a good pair of running shoes under a certain price, or to compare two blenders, and it synthesises an answer from whatever it can find, often without the shopper ever visiting the retailer’s site to get there.
Julia Ahlfeldt makes this distinction directly in the report: “AI isn’t just another Google. It’s where customers make up their minds. Nobody used Google to tutor them or talk a decision through.” A search engine surfaces options. An AI assistant narrows them down, and in doing so, it becomes the layer through which a growing number of customers first encounter a brand, or fail to.
This has a practical consequence for product pages. If an AI tool is drawing on your site to answer a customer’s question, it needs content structured in a way that can be read, extracted and trusted. A product description written purely to persuade a human browsing the page, full of adjectives and short on specifics, gives a language model very little to work with.
Clear specifications, honest comparisons, plainly stated pricing and shipping information, and answers to the questions customers actually ask, all give an AI assistant material it can accurately summarise. The businesses whose product content reads like a well-organised reference document, rather than a marketing pitch, are the ones more likely to be the source an AI tool pulls from.
Reviews carry more weight than most businesses assume
The report also found that customer reviews on a company’s own website are the single most trusted information source for consumers researching a purchase, cited by 58%, yet only 16% of business respondents ranked reviews as an influential factor for their customers.
That gap matters here too, because AI tools draw heavily on review content and independent platforms when forming a recommendation. A product with thin or outdated reviews gives an AI assistant less to cite, and less reason to recommend it over a competitor with a fuller, more current review base.
Encouraging genuine reviews and keeping review content visible and current is no longer only a trust signal for human shoppers. It’s raw material for the systems now doing a portion of the recommending on their behalf.
The businesses getting ahead are asking a different question
Tanita van der Merwe of Altron Group, commenting elsewhere in the report on AI adoption more broadly, frames the shift this way: “Every business asks ‘What can AI do?’ But the ones pulling ahead are asking ‘What will we catch first? What will we fix faster? What do our customers need before they even know they need it?’”
Applied to product discovery, this means moving past the question of whether to optimise for AI tools and toward a more specific one: what would a customer need to know about this product for an AI assistant to recommend it accurately and favourably. That often means going further than a standard product description, addressing common comparisons, use cases, and the practical questions a shopper would otherwise have to ask a salesperson.
A visibility gap that compounds over time
The businesses without a generative-engine-optimisation strategy aren’t necessarily doing anything wrong today. The risk is more gradual. As AI-assisted research becomes a bigger share of how people shop, the retailers whose content is structured for machine readability accumulate a visibility advantage that’s difficult to close later, in much the same way early search-engine-optimised sites built a lasting edge over competitors who treated SEO as optional.
The report frames this shift as part of a larger pattern: customer experience increasingly begins before a customer ever reaches a company’s website, in the search engines, review platforms and AI tools where research now happens first. For ecommerce specifically, that means the product page is no longer only a page. It’s a source document, and the businesses that treat it as one will be the ones an AI assistant reaches for when a customer asks it to decide.