AI assistants are now fielding wholesale and trade questions

A pattern consolidating through September 2026 is that AI shopping assistants are being asked to answer wholesale and trade questions — minimum order quantities, tier pricing, lead times, duty terms — and that they handle them with the same fact-extraction discipline they apply to consumer product pages. The consequence is that a supplier whose trade terms exist only inside a gated PDF or a request-a-quote form is structurally invisible to the channel.

The reason is mechanical rather than political. An assistant producing a shortlist needs values it can state: a number, a currency, a range. "Contact us for pricing" supplies none of them, so the supplier cannot be placed. Assistants do not submit forms on a buyer's behalf, and they do not infer a price from a retail page.

What buyers are actually asking has become fairly consistent: can a mixed order combine SKUs to reach the minimum, is the quoted price DDP or ex-works, does the lead time refer to dispatch or delivery, and is a reseller certificate required. Each is a single stated value. Suppliers who publish those values in page text are being included in shortlists and compared on terms; suppliers who gate them are being skipped without the merchant ever seeing the omission in an analytics report.

The shift also changes the shape of the question. Wholesale enquiries to assistants arrive with more constraints per turn than consumer queries — quantity, destination, Incoterm, unit price ceiling — which raises the number of facts a trade page must state before it can survive the filter. A page that answers four of six constraints is not half-visible; it is excluded, because the assistant cannot confirm the unstated ones.

There is a real tension here worth naming. Trade pricing is commercially sensitive, and publishing it has costs. The practical middle ground that appears across merchant data is bands rather than exact tiers, published on a page that also states the account requirement — a stated range still lets an assistant position the supplier, while a locked page lets it position a competitor. The other recurring failure is not missing data but conflicting data: a trade page promising five-day dispatch beside a shipping policy listing seven to ten days gives the assistant a contradiction it cannot resolve, and unresolvable contradictions tend to resolve to exclusion.

This is the same "publish the fact, once, in text" discipline we apply to consumer product pages, extended to the trade side of the business. An inconsistent MOQ statement — quoted differently on the trade page, the product page, and a marketplace listing — is not three facts, it is one unreliable fact. Visora's scan reports where your trade and product pages disagree on pricing, order minimums, and lead times, which is exactly the kind of disagreement that reads to an assistant as a reason to skip. Running a scan after any terms change shows which pages have already drifted. See [the audit page](/audit) for what the consistency checks cover.

Wholesale was gated because gatekeeping worked. In an assistant-mediated channel, the gate is the exclusion.