The Order Shelf report is live: we now measure who gets the AI order

1 October 2026 · Vladimir Fedorov, co-founder and CTO, ShelfSignal

ShelfSignal started as an AI visibility report: ask buyer questions, measure which brands are named, and show the pages assistants used to build the answer. That is still useful. But visibility is not the whole buying journey. When a shopper says “buy me magnesium”, the assistant has to choose one brand, one seller, one channel and one product. That moment is the Order Shelf, and we now support it as a separate report

The new report answers a sharper question: if an AI agent is asked to buy in your category today, who gets the order?

What the Order Shelf measures

Each audit runs six grounded purchase briefs across five assistants, with two runs per assistant. The briefs include real constraints a shopper would give: budget, delivery, pickup, certification, channel preference and product attributes. Every run stops before payment, no accounts are created, no carts are created, and no real orders are placed

The output is not just a ranked list. We record the brand chosen, the seller chosen, the channel used, the reason the assistant gave, and the decisive thing that made competing offers lose. In the magnesium sample, assistants split orders across brand sites, Amazon, iHerb and pharmacy chains, which means a brand can be well represented in one channel and invisible in another

Why this is different from visibility

A visibility report tells you whether you appear in the answer. The Order Shelf tells you whether the assistant would actually buy from you. Those can diverge. A brand can be mentioned often and still lose the order because the assistant cannot verify a certificate, cannot find delivery windows, cannot compare price per serving, or prefers a seller with clearer checkout data

That is why the report includes implementation, not just measurement. Each audit ends with six ranked moves: for example a certificate page, shipping data, dose and attribute cleanup, price and count changes, reorder path, or channel coverage. Each move has a scope, a timeline and a fixed price, so the report can become a work plan instead of a dashboard

What we support now

We support the Order Shelf report for any category where a shopper can ask an assistant to buy. Supplements are the first public example because the buying criteria are visible: dose, form, testing, allergens, price and delivery. The same report structure works for music software, local retail, beauty, pet products, sports nutrition, home goods and other categories where assistants compare offers before choosing

Each audit covers one market, either a city or a country. Additional markets can be added when channel availability, delivery, pickup or seller preference changes the answer. That matters because an assistant buying in New York may not choose the same seller it chooses in London or Berlin

How to use it

If you are not sure whether assistants know your brand at all, start with the AI Shelf Index. If assistants already mention you, or if the commercial question is “why do they buy from someone else?”, start with the Order Shelf audit. The audit fee is credited against a fix programme that starts within 30 days, and every fix includes a re-run of the same briefs so the before/after change is visible

The short version: visibility gets you into the answer. The Order Shelf shows whether you win the purchase. ShelfSignal now measures both, and we can help fix the storefront and listings that decide the order

Method: 6 purchase briefs × 5 assistants × 2 runs = 60 purchase decisions per audit. Runs stop before payment and record the assistant, prompt, market, selected brand, selected seller, channel, cited rationale and lost-order reason