New research published this week finds that AI assistants are increasingly shaping product discovery, but reviews still seal the deal. Modern Retail's coverage of the findings describes a two-stage journey: generative surfaces introduce shoppers to brands and products they would not have encountered through search, while review content determines whether that introduction converts. The implication for brands is that visibility and credibility are now separate investments, and optimizing only one of them leaves the funnel broken at a specific, measurable point.
What the New Research Found
The research setup
The study, reported by Modern Retail's AI strategies desk, examined how shoppers use AI assistants across the path from need recognition to purchase. Its central finding is a division of labor between surfaces: generative answers excel at surfacing options and framing criteria, while review platforms supply the proof layer that shoppers consult before committing. The two surfaces are not competing so much as operating in sequence, with different content winning at each stage.
Key numbers
The headline conclusion, as the research coverage puts it, is that AI boosts discovery but reviews still seal the deal. Assistants raise the consideration set; review depth closes the decision, a finding quantified in behavioral terms rather than opinion. The research tracks how often an assistant-introduced brand survives the verification stage, making it one of the more actionable datasets published on generative commerce this season.
Discovery Has Moved Upstream
Discovery moving upstream changes what "being found" means. In the search era, shoppers arrived with a category in mind and compared options; in the assistant era, the category frame itself is often set by the conversation. A shopper asking for a gift idea, a solution to a problem or a comparison may receive brand names before they have typed a single query into a search box. That upstream position is valuable precisely because it shapes the criteria by which everything downstream is judged.
This upstream shift also explains the investment pattern across retail. Target's holiday push pairs affordability and exclusives with an in-app assistant, and the scale of Amazon's seasonal hiring plans shows how deeply operational models are being rebuilt around new demand surfaces. When the introduction happens inside a conversation, the content that wins is the content the assistant can parse, quote and trust, which raises the premium on structured, factual brand information over lifestyle marketing copy.
Reviews Still Decide the Winner
Review depth signals
The proof layer has its own economics. Review depth, meaning the volume and specificity of verified feedback, functions as the tiebreaker among the options an assistant surfaces. Shoppers reading reviews are not looking for stars alone but for evidence about fit, durability and use cases, the same signals that determine whether a recommendation survives contact with reality. Brands that treat reviews as a reputation chore rather than a content asset systematically underinvest in the decisive stage.
Answer-level evidence
The research also notes that assistants themselves increasingly cite review content when constructing answers, which couples the two layers tighter than the sequence model suggests. Modern Retail's analysis describes answers that name options and immediately reference what other buyers report, meaning review presence now influences both the human decision and the machine recommendation. A brand strong on structured facts but weak on verified feedback is optimizing half the funnel.
The Gap Between Visibility and Conversion
Putting the findings together produces a diagnostic that brands can run on themselves. If AI answers never mention the brand, the problem is upstream: product information is not structured, quotable or present in the sources assistants trust. If answers mention the brand but shoppers do not convert, the problem is downstream: review depth, specificity and recency are insufficient to close the decision. The two failure modes look identical in a revenue report and require opposite interventions.
The gap is widest for mid-size brands. Large brands already maintain both layers through budget and staff; small brands can win on authenticity and niche specificity. Mid-size brands often have enough presence to appear in answers but not enough review density to survive comparison, a position that generates awareness without revenue. Marketer optimism for the season suggests budgets are available, and the research implies where they should go first.
How Brands Should Rebalance
Rebalancing starts with measurement. Brands need separate scorecards for assistant visibility and review strength, because a single blended metric hides the funnel break. The practical checks are straightforward: query assistants with the questions real shoppers ask, note which brands appear and with what evidence, then audit review volume, recency and specificity against the same consideration set. Run quarterly, this becomes an early warning system for both layers.
The investment logic follows from the sequence. If the brand is invisible upstream, prioritize structured product information, authoritative mentions and factual consistency across sources, because none of the downstream work matters without the introduction. If the brand appears but does not convert, prioritize verified review generation and the content that gives reviewers something specific to say. The research finding that reviews seal the deal is a budgeting instruction as much as an insight: fund the layer that is actually losing the sale.
