Blaze AI Discovery for WooCommerce

Only if its crawlers can fetch your pages. ChatGPT uses OAI-SearchBot to index sites for its search answers, ChatGPT-User to fetch a page on demand when someone asks about it, and GPTBot for training — and any of them can be turned away by robots.txt, a firewall or a CDN without you noticing. The plugin's AI crawler access check (AI Discovery → Store Audit) tests each crawler from your server and, with cloud checks on, from outside your network, then shows a per-crawler verdict with a link to the fix.

Usually one of three things: the AI crawlers cannot reach the store, the product pages do not give the assistant enough facts (brand, identifiers, price, availability, shipping and returns) to recommend a product with confidence, or the store is simply not mentioned anywhere the assistant reads. The plugin addresses the first two directly, with the crawler access check and the per-product AI readiness score, and the AI Visibility page shows how often ChatGPT and Google AI mention you over the trailing 30 days, which pages they cite and which competitors get named instead. No plugin can make an assistant recommend you; it can make sure the reason is not something on your side.

They fall into two groups. Training crawlers (GPTBot, ClaudeBot, Google-Extended) collect pages to train models; blocking them is a legitimate choice and does not by itself hide you from shoppers. Retrieval crawlers (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, PerplexityBot, bingbot) are what the assistants use to find and cite your pages when a user asks — ChatGPT-User is the on-demand fetcher that visits a page the moment someone asks about it — so if those are blocked, the store cannot be recommended. Google-Extended is a robots.txt-only token rather than a user agent that visits your site, so it is judged from robots.txt alone. The plugin lists every crawler with its class and verdict, and the audit row passes when every retrieval crawler can get in.

Often, yes. Since mid-2025 Cloudflare has blocked known AI training crawlers by default on newly created zones, and Super Bot Fight Mode or a managed challenge can stop the retrieval crawlers as well; some of these crawlers are on Cloudflare's verified-bot list and some are not, so what gets through depends on your settings. Because the block happens at the edge, a check from your own server passes while the crawler is turned away. With cloud checks on, the plugin fetches your pages from outside with each crawler's user agent, reports which layer is blocking (Cloudflare, Wordfence, Sucuri, Kinsta or robots.txt) and links to the fix for that layer.

Probably not, and it is not where the gains are. llms.txt is a proposed convention — a plain-text summary of your site served at /llms.txt listing your store name, description, categories and key pages — that some tools read but the major assistants have not committed to; asked directly, ChatGPT itself calls it low priority for ecommerce and puts crawlable pages, structured data and clear product facts ahead of it. The plugin generates and serves one anyway because it costs nothing, but spend your time on the crawler access check and the product readiness scores first.

Yes, in the sense that it does the practical, measurable parts of answer engine or generative engine optimization for a WooCommerce store: making sure AI crawlers can reach you, making product data complete and machine-readable (schema, feed, identifiers, policies), and measuring whether ChatGPT and Google AI mention you. It does not write content, build links, submit your store to assistants or promise placement, and it does not replace conventional SEO — Google's AI Overviews are built on Google's index, so the ordinary SEO work still matters.

ChatGPT Shopping shows products inside ChatGPT answers and, for approved merchants, lets shoppers buy without leaving the chat. The plugin generates a product feed in OpenAI's current CSV spec, validates every product against it and scores each product's readiness so you know what to fix first. Joining the merchant programme, uploading the feed and enabling instant checkout happen with OpenAI (chatgpt.com/merchants), not inside the plugin, and agent-driven purchasing is still early — the plugin prepares the data, it does not enrol you.

Google's thresholds, measured at the 75th percentile of real visits: Largest Contentful Paint (LCP) 2.5 s or under, Interaction to Next Paint (INP) 200 ms or under, and Cumulative Layout Shift (CLS) 0.1 or under are "good"; above 4 s, 500 ms and 0.25 respectively is "poor", with "needs improvement" in between. Time to First Byte (TTFB, good under 800 ms) is shown for context and does not decide the pass. With cloud checks on, the audit shows mobile field data — what real Chrome users experienced on your store over the last 28 days, from Google's Chrome UX Report — rather than a lab test. "Not enough data" means Google has not yet collected enough real-user samples for your origin; the numbers appear once the store gets steady Chrome traffic.

No. The tracking script is delay-loaded using requestIdleCallback (runs only when the browser is idle) and only fires a network request when an AI referral is detected — which is less than 1% of visits. Schema injection adds under 5ms to page render. Heavy operations like batch scoring run in the background via Action Scheduler.

Yes. The plugin detects which SEO plugin you're using and augments its schema output using the plugin's own filter hooks. It never creates duplicate schema blocks. Tested with Yoast SEO, Yoast WooCommerce SEO, and Rank Math.

ChatGPT (via UTM parameters and referrer), Perplexity (referrer), Google AI Mode/Overviews (URL text fragments), Microsoft Copilot (referrer), Claude (referrer), Gemini (referrer), and Grok (referrer). It also detects AI bot crawlers like GPTBot, PerplexityBot, and ClaudeBot via user-agent matching.

Yes. When a visitor arrives from an AI platform, the plugin sets a first-party cookie. When that visitor completes checkout, the order is tagged with the AI source. The dashboard shows AI-attributed orders and revenue.

Industry benchmarks show approximately 70% of AI-referred traffic arrives without identifiable headers (especially from mobile apps). We detect what we can via referrer headers, UTM parameters, and text fragments, but some AI traffic is fundamentally undetectable with current technology. We believe in being transparent about this.

Not yet. The plugin generates the CSV; you upload it to the SFTP endpoint OpenAI provides once your merchant application is approved. If you are in the merchant programme and want automatic delivery, post on the support forum and we will work with you. The generated feed file is saved under wp-content/uploads/blazeaid/ and, like the rest of your uploads directory, is publicly readable at that URL — it contains only the same product data already visible on your storefront.

Yes. The tracking uses client-side JavaScript (like Google Analytics) that works regardless of server-side page caching. Tested on Kinsta, WP Engine, SiteGround, Cloudways, and shared hosting. Tracking signatures embedded in cached pages expire after two full days; if your page cache keeps HTML longer than that, visits from those stale pages are silently skipped. Adjust with the blazeaid_signature_window_days filter.

All data is stored locally in your WordPress database. Optional cloud checks send only your store's public URL, one public product URL, your store's domain, country, locale, up to five product category names and version numbers — and only after you enable them. See External Services below for the full list.