Fluens AI
GEO for ecommerce, DTC and CPG

Product research moved into the assistant

Shoppers ask for a recommendation and get two or three brands named, with the reasoning already done for them. Almost none of that reasoning comes from your product page.

What the sources look like
2.8%
of cited sources were the brand's own pages
59%
were third-party: roundups, comparisons, review coverage
31%
of the time a brand was named, its own page was cited too
Shero Commerce, across 1,851 sources cited for purchasing questions in 60 product categories on Google AI Mode, ChatGPT and Perplexity.
The shape of the problem

Third-party coverage decides whether you are a candidate

Your own pages decide what gets said about you once you already are. Both matter — in that order.
01

Retailer and marketplace listings

Heavily cited and frequently out of date. Old pack sizes, retired variants and superseded formulations get repeated as current, because nothing on those pages says otherwise.

02

Roundups and 'best of' coverage

A single well-placed roundup can decide a whole category's answers. If you are not in the ones being cited, no amount of work on your own site compensates.

03

Review platforms and forums

The association between an active review profile and being named is strong across the research, though most of it is vendor-funded and shows correlation rather than cause.

What we do

Work on the sources, in order of leverage

01

Find which sources decide your categories

We run your buying questions repeatedly across assistants and record every citation. In most categories a handful of pages are doing nearly all the work.

02

Correct the listings that are wrong

Retailer pages, marketplace entries and aggregator listings carrying outdated product facts. These are usually the quickest wins available.

03

Get into the coverage that gets cited

Roundups, comparisons and category guides, placed where the answers in your category actually draw from rather than wherever will take a placement.

04

Track it monthly, per product line

Visibility varies hugely between SKUs and categories. A brand-level number hides the product that is invisible.

Common questions

Does my product page matter at all?
Yes, but later in the sequence than you would expect. In the Shero Commerce study, when an assistant named a specific brand it cited that brand's own page 31% of the time. Your site gets read — it mostly gets read after something else has already put you in the running.
Is in-chat checkout something I need to prepare for?
Not urgently. OpenAI's Instant Checkout reached roughly 30 Shopify merchants, had accuracy problems, and by March 2026 OpenAI had moved checkout back out of the chat. Etsy, one of its largest partners, described ChatGPT as a valuable discovery channel while purchases through it stayed low. Discovery moved into the assistant; buying stayed on your site.
We sell through retailers as well as direct. Does that help or hurt?
Usually it helps, because retailer product pages and category listings are heavily cited. It hurts when those listings carry an old formulation, the wrong pack size or a discontinued variant, because the assistant will repeat it confidently.
How does this work for a brand with no reviews yet?
Review volume matters, and the association is strong, but the causal evidence is thinner than vendors imply. We would start by finding which sources the answers in your category actually cite, because in some categories a single roundup or retailer page is doing most of the work.
Schedule your free auditDo reviews actually help? The research →