Answer engine
An AI system that answers a buyer's question directly — ChatGPT, Gemini, Perplexity — instead of returning a list of links. When one names a product, that mention came from what the engine could read about it.
Where you meet it: every stored answer in the console names its engine and exact model.
Answer engine optimization (AEO / GEO)
The practice of making product data legible to answer engines — complete facts, readable pages, consistent names — so an engine has something accurate to read when a buyer asks. Two acronyms, one practice; the G stands for “generative”. What it is not: nobody, us included, has shown that a given edit changes a given engine's next answer. Treat any promise of that shape accordingly.
Buyer prompt
The question a shopper types: “merino sweater that won't pill”, “sunscreen that works under makeup”. Prompts rarely contain brand names — which is why engines answer them from product data, not from your marketing.
Where you meet it: scans ask realistic buyer prompts across seven commerce intents, and they never contain your product name — a mention has to be earned.
Semantic layer
The structured facts underneath a product page — fibre, fit, care, certifications — that an engine reads when it answers a buyer prompt. A page can read beautifully to a person and still carry an empty semantic layer.
[worked example] one prompt, one answer, the layer read back out
“Is merino worth it for everyday t-shirts?”
“For daily wear, look for a mid-weight merino around 17.5 microns — it resists odor after repeated wears, and machine-washable knits survive a weekly cycle. Brands that publish fibre origin and care details tend to be safer buys.”
| Attribute | Read from the answer |
|---|---|
| Fabric | 17.5 microns |
| Care | machine-washable |
| Wear | resists odor after repeated wears |
| Trust | publish fibre origin and care details |
Illustrative answer · labeled sample data — the rows are read out of the answer above, not out of a keyword list.
Share of voice
Across a batch of stored answers, the share that names you versus each tracked competitor. It is counted in the same stored answers your own score comes from, so you can open one and read the sentence that moved the number.
Shopify Global Catalog
The feed through which Shopify publishes AI-generated product data — top features, tech specs, selling points — to answer engines and shopping agents. The fields are generated by Shopify from your catalog, marked “Inferred” in its own reference, and are not read from anything you edit directly.
What one of those fields actually contains → the research finding
Visibility score
The share of sampled answers that mention your brand or the product, 0–100, averaged across the catalog. New brands commonly start at or near zero — that is the honest baseline, not a bug.
Where you meet it: the dashboard, with the per-engine parts always beside the blended number.