Shoppers now ask ChatGPT, Perplexity, and Gemini what to buy, what it costs, and where to pick it up. Those systems read your storefront mechanically — and if the data they find is empty or ambiguous, they quote the wrong price, send the sale to a marketplace, or recommend a competitor. We audit the problem, fix it on your own domain, and prove the difference with transcripts.
No signup. The audit reads only your public pages and takes about thirty seconds.
From our testing, July 18, 2026 — timestamped transcripts archived
“[This brand] does not appear to have any physical retail locations open for walk-in purchases.”
Perplexity — a hedged guess from stale third-party pages, not the brand’s own data: the storefront gives machines no way to confirm its fulfillment model. The same answer quoted Amazon’s $130 price instead of the $115 on the brand’s own site.
“I wasn’t able to retrieve specific current pricing from the search results.”
Claude — after five consecutive web searches against a storefront whose product data is rendered by JavaScript that AI crawlers never execute.
“The ‘Add to Cart’ button appears alongside an ‘Out of Stock’ indicator in some variants.”
ChatGPT — the best answer of the five: right site, right price range, unable to say what is actually buyable.
None of the five platforms could verify live inventory. Re-tested July 23: asked for the best products in the brand’s flagship category — with no brand named — ChatGPT, Perplexity, and Gemini all answered with lists of competitors. None mentioned the brand. Its own catalog even serves the wrong product’s description on two pages, repeated verbatim by every AI that reads it. This is not a marketing problem; it is a readability problem, and it is measurable. The store above scored 62 of 100 on our audit. Our own reference deployment, running the full remediation, scores 100.
The market, measured
These are not our figures. They come from Adobe’s analysis of more than one trillion visits to U.S. retail sites and Salesforce’s data across 1.5 billion global shoppers.
$262B
in online sales influenced by AI and agents over the 2025 holiday season — one in five retail transactions.
Salesforce, Jan 2026
+393%
growth in AI-referred traffic to U.S. retail sites, first quarter of 2026 versus a year earlier.
Adobe Analytics, Apr 2026
+42%
higher conversion rate for AI-referred visitors than all other traffic — a full reversal from a year ago, when they converted 38% worse.
Adobe Analytics, Mar 2026
9×
AI-referred shoppers buy nine times more often than visitors from social media — they arrive having already done their research inside the chat.
Salesforce, Jan 2026
The catch: Adobe’s same study found that a third of retail homepages, product pages, and FAQs are unreadable by AI systems. The channel converts 42% better and pays 37% more revenue per visit — but only for the stores whose data AI can actually read. That readability gap is precisely what we audit and repair.
Sources: Adobe Digital Insights, Quarterly AI Traffic Report (Q1 2026) — over one trillion visits to U.S. retail sites and a survey of 5,000+ U.S. consumers. Salesforce Shopping Insights HQ, 2025 Holiday Shopping Report (Nov 1 – Dec 31, 2025) — activity of 1.5 billion global shoppers.
1
We read your store the way AI crawlers do — no JavaScript, public pages only — and grade five pillars: crawler access (including header-level blocks robots.txt never shows), agent protocols (agents.md, llms.txt, UCP, and live MCP endpoints), catalog readability, product schema (down to the identifiers product graphs match on), and local physical availability. Scored 0–100, findings in plain English.
2
A translation layer computes correct, current data from your commerce platform and places it on your own pages: server-rendered product schema, unit-level inventory, and store-by-store availability with addresses and distances. One template installed once; no redesign, no checkout changes. Then it stays current on its own: every inventory or price change triggers a resync within minutes, backed by a full-catalog pass every six hours. Your schema never goes stale — that is the point of the subscription.
3
Every day we re-run the audit and re-ask the AI platforms the questions your customers ask; every week we re-grade every page of your catalog individually. You watch it all in your private results portal — scores, trends, per-product detail, and what each platform is actually saying about you. If a theme update breaks the data, we catch it on the next daily pass.
Everything we sell runs on this site. The audit tool is public. Our own deployment serves the complete machine-readable surface — catalog, per-store inventory, agent instructions — and scores 100 of 100 on the same unmodified audit.
Audit any store: machineread.io/audit
Audit us: our own scorecard
For machines: llms.txt · agents.md · products.json
| Tier | Monthly | Includes |
|---|---|---|
| Core | $149 | Translation layer with continuous sync — schema updates within minutes of inventory changes — plus the client portal and scheduled audits. |
| Local | $399 + $99 per location | Store-level inventory certainty, synced in near-real time, and “in stock near me” answers for every retail location. |
| Growth | from $1,500 | Daily five-platform monitoring with transcripts, weekly full-catalog re-grades, priority protocol updates, dedicated support. |
| Enterprise | custom | Legacy or custom platforms, service-level agreements, dedicated integration. |
Pilot program. Our first engagements in each vertical run sixty days at no charge, in exchange for a published case study. Success is defined by measurable targets agreed in writing before we begin — audit score, platforms verifying stock, platforms quoting your own price — and re-tested daily.