Skip to main content
When a shopper asks an AI assistant for a product recommendation, the answer often arrives as a shelf — a short, ranked set of specific products, with names, merchants, and prices. That shelf is the modern equivalent of an end-cap in a store, and being absent from it is invisible in a way no brand-level metric captures. AI Shelf shows you those shelves: which products appear for your category’s prompts, which are yours, which belong to competitors, and where you rank among them. It is a different question from the rest of the dashboard. Everything else measures whether AI talks about your brand. AI Shelf measures whether AI recommends your products.

​
The five numbers

Share of Shelf is the headline, but it is a product of the others, and that is what makes them worth reading separately.

​
How to read it

The three that matter most form a funnel, and each failure has a different fix. Shopping-Mode Rate — does a shelf appear at all? If this is low, your tracked prompts are not the kind that make AI produce a product list. That is not necessarily a problem: some categories are researched conversationally rather than shopped. But if buyers in your category do ask shopping questions, low shopping-mode rate means your prompt library is not covering how they ask. Presence Rate — when a shelf appears, are you on it? This is the hard gate. A shelf you are absent from cannot be improved by ranking better. Absence usually traces to the product not being discoverable to the AI in the first place — missing from the retail and review sources these answers draw on. Top-Pick Rate and Avg Position — when you are on it, where? Only meaningful once presence is healthy. If you appear on most shelves but sit near the bottom, the problem is comparative: something about how your product is described, priced, or reviewed puts it behind the alternatives.
Read Presence Rate before Share of Shelf. A rising Share of Shelf driven purely by competitors dropping off is not the same as one driven by you appearing more often, and only the funnel tells you which happened.

​
Exploring the shelf

Below the metrics, the explorer breaks the shelf apart. You can filter by AI platform, and by who the product belongs to:
  • Own — your products
  • Tracked competitors — brands in your configured competitor set
  • Untracked competitors — everyone else appearing on your shelves
That last group is the one worth watching. Untracked products are brands AI is recommending in your category that you have not told Cited about — which is often how you discover a competitor you did not know you had, or a retailer’s own private label quietly taking shelf space. Drilling further shows the individual product appearances: title, position, merchant, price, platform, and the topic that produced the shelf.

​
Why a product was surfaced

Where the underlying AI answer explains its reasoning, Cited captures it — the attributes and phrasing that came with the recommendation. This is the most directly actionable part of the surface. Knowing you rank fourth is a score; knowing the three products above you were all described in terms of a specific attribute is a content and merchandising instruction.

​
What it covers

AI Shelf reads the AI platforms that actually produce product shelves. Not every tracked platform does — a shelf requires the answer to render a structured product set, which is a behaviour specific to shopping-oriented surfaces. Product-to-brand attribution is deterministic rather than inferred: a product is matched to your brand or a competitor by rule, not by a model’s guess. Anything that cannot be matched confidently is reported as untracked rather than assigned.
Prices shown are the prices the AI quoted, captured as they appeared in the answer. They are not validated against your catalogue and can be stale or wrong — which is itself worth knowing, since that is the price a shopper was shown.