AI findability means whether an assistant, asked a shopper’s question, recommends your products, states their facts correctly, links to your store, or recommends a competitor instead.
New Our method is published and our on-site search evidence is strong; our own assistant results are not in yet. What we can and cannot say.
What we look for
Each intent becomes one or two prompts in the shopper’s words: an open request (“best sofa for a small apartment with a dog, under $1,500”) that shows whether the assistant finds you at all, and for some intents a store-aware request (“does [store] sell a sofa that works for a dog?”) that shows whether what it says about you is true.
We test five surfaces in a standard Evaluate:
best sofa for a small apartment with a dog, under $1,500
Here are three options that suit small spaces and pets:
Sources: competitorone.example, a review site. No link to your store
Platform facts, checked November 6, 2026
These facts come from each platform’s own documentation. They shape both the test and the fix.
| Platform | How it gets product data | What the store controls | Source |
|---|---|---|---|
| ChatGPT | Structured metadata from first- and third-party providers plus web content; Shopify stores included through Shopify Catalog; direct feeds via a merchant waitlist | Catalog data quality, OAI-SearchBot access, store policies | OpenAI Help: Shopping with ChatGPT search · OpenAI: Powering product discovery in ChatGPT (March 24, 2026) |
| Google AI Overviews, AI Mode, Gemini | Search index plus Merchant Center feeds (Shopping Graph); no special files or schema required | Merchant Center feed, structured data matching visible text, crawl access | Google Search Central: AI features and your website |
| Microsoft Copilot | Microsoft Merchant Center feeds inform organic results; Shopify stores auto-enrolled in Copilot Checkout after an opt-out window | Merchant Center feed, return and support settings | Microsoft Advertising blog (January 8, 2026) |
| Perplexity | Free US Merchant Program (feed via API, SFTP or S3); PerplexityBot for search | Program enrollment, feed accuracy, PerplexityBot access | Perplexity Merchant Program terms · Perplexity crawlers |
Eligible products reach ChatGPT and Copilot through Shopify Catalog with title, description, options, images, price and availability. Data kept in metafields needs the Channel mapping page to be sourced correctly. The question is what the listing says. Shopify Help
“You don’t need to create new machine readable files, AI text files, or markup to appear in these features.” We do not sell llms.txt as a fix for Google. Google Search Central
OAI-SearchBot is for ChatGPT search, GPTBot for training. PerplexityBot surfaces sites; Perplexity-User fetches on a user’s request. A CDN bot rule can block a bot that robots.txt allows, so we test actual responses. OpenAI · Perplexity
We tie every fix to something in the first list and never claim a fix will produce a recommendation.
What you get
Each prompt runs 3 times per surface on different days within one week, in fresh logged-out sessions with memory off where the product allows, from one fixed US state. A standard Evaluate is about 450 answers. Rates are printed with run counts (“present in 4 of 15 runs”); single runs are never findings. A product counts as present only if it is your listing or links to your store.
Technical checks explain each miss: crawler access by actual response, feed and catalog eligibility, structured data, and whether facts are stated as text. Want your search box too? Findability Evaluate covers both for $5,000 / $10,000.
New
AI assistants only
Catalog Fix improves both pillars at once; re-test the assistants for $1,250 when something changes, such as turning on ChatGPT or Copilot checkout or joining Perplexity’s Merchant Program.
Is AI findability the same as GEO or AI visibility tracking? No. Monitoring tools track brand mentions daily across many prompts. We run a one-off test with real shopper intents, check every answer against your own catalog, and tie each miss to a factor you control. We do not track daily, and we name monitoring tools in our fix lists if you want ongoing tracking.
Do I need llms.txt or special AI files? Not for Google: its documentation says you do not need new machine-readable files, AI text files or markup to appear in its AI features. We mention llms.txt only for stores with large content sections, and we label it unproven.
Our store is on Shopify. Are we already in ChatGPT? Probably listed: eligible Shopify products reach ChatGPT and Copilot through Shopify Catalog with title, description, options, images, price and availability. The question is what the listing says. Data kept in metafields needs the Channel mapping page to be sourced correctly.
Why is this labelled New? We have strong evidence for on-site search and none yet of our own for assistants. Until our repeatability and root-cause checks pass, AI findability is labelled New, sold with its method visible, and on-site results carry the proof.
Next Step · Engineering Architecture & Diagnostics
A 30-minute engineering call: your store, your search & recsys stack, and whether a Tier 1 Audit, Tier 2 Build, or Resident ML Engineer fits best. No slides, no obligation.