Buying guides written from your catalogue, linked to your products

No, and it is worth being precise about why, because it is the question retailers ask first.

Google distinguishes two product experiences: product snippets, for “product pages where people can't directly purchase the product”, and merchant listings, for “pages where customers can purchase products from you” (developers.google.com/search/docs/appearance/structured-data/product, checked 4 September 2026). Both are generated from the markup on the product page itself, or from a Merchant Center feed. An article about a product is neither.

So the division of labour is fixed. Product and offer markup, with price and availability that agree with what the page displays, is your storefront's job, and it is the layer most often broken — a stale offer block on a sold-out item costs you more than any article will earn. The engine ships the article layer: Article JSON-LD, an FAQ block with matching schema, a hero image with alt text, and internal links down into the products and category the guide discusses.

The free site check reports what your product pages emit today. Repair that first; the article layer sits above it, not instead of it.

Research that reads only the pages already ranking produces a competent summary of pages that already exist, and Google has no reason to rank an eleventh.

IT Master reads the competitors too, but the facts in the article come from your side of the fence. Before a word is written, the engine retrieves whatever first-party material the store holds:

• product data: titles, variants, specifications, what is in the box;
• supplier datasheets and manuals for the lines you sell;
• support Q&A and returns notes, where you keep them, which record the questions buyers actually asked and why things came back.

A comparison built on that can say which of two models has the wider field of view, and which one buyers returned over a firmware fault. Those sentences are on no competitor page, because no competitor holds the data. That is what a novelty check rewards, what an AI assistant has reason to quote, and what a rival cannot reproduce by rewriting your page. Where the data is thin the article stays inside what can be verified instead of filling the gap. First-party data explains the retrieval side.

A guide that ranks but does not link to the products it recommends is a leak. IT Master places internal links from the same retrieval layer that grounds the article, so the link to a product page goes in where that product is named, not into a footer block of related posts.

They run in three directions:

• down to the product pages the guide names;
• across to the category the guide supports, so that category collects a link from every article in its cluster;
• sideways to sibling guides and comparisons, which builds a topic cluster instead of a pile of orphaned posts.

New articles link back into earlier ones in the same cluster, so the cluster thickens as it grows and nobody edits old posts by hand. Article and FAQ schema ship in the same payload, and IndexNow pings the search engines as soon as the page is live rather than waiting for a crawler to find it.

For a store, a false spec is worse than a dull paragraph. It is a return, a chargeback, and a customer who does not come back. So the check is structural rather than a promise: the model that writes is never the model that checks.

Writers are Claude models, and no model ever marks its own work. Four checks then run in parallel. None is run by the model that wrote the draft, and every check includes at least one model from a different vendor: Gemini and GPT lead the fact-check and the novelty score, while the E-E-A-T and AI-tell panels pair a Claude judge with a GPT or Gemini one. What they check is the fact-check against the retrieved catalogue and datasheet material, a novelty score against every competitor page pulled during research, an E-E-A-T critique, and an AI-tell detection pass. A draft that fails any of them goes back to the writer with the specific objections, for up to three revision rounds, and is rejected if it still fails.

The numbers are worth stating plainly. Across 186 Standard-tier runs, 21% of drafts cleared every check on the first attempt with no revision, and 51% got there within the three revisions allowed; the rest never published. Pro-tier judges are stricter: 25% cleared every check across a small sample of 16 runs. The rejects are the point. How it works walks the full pipeline.

cctvmaster.co.uk sells CCTV to UK trade and retail buyers, and it runs this engine against its own catalogue. The store holds what the approach needs: specifications for cameras, recorders, cabling and kits, supplier datasheets for its main lines, and a history of buyer questions.

So a guide to choosing a recorder can quote channel counts and storage figures from the datasheet the store already holds, and a comparison between two camera lines can use the specifications of the exact models on the shelf and link straight to them. Nothing in either article depends on a model remembering a number.

Every store on the engine runs the same pipeline against its own catalogue, each with its own brand voice and its own retrieval collection, kept apart from one another. What is shared between stores is the method, never the content or the data. The security and CCTV page goes through that in more depth for the trade.

Articles arrive on whatever the store runs. On WordPress, install the plugin, paste one key, and posts appear with the Yoast or Rank Math fields filled. Next.js storefronts use @itmaster/sdk; anything else can read the REST pull feed or take a signed webhook. Shopify and Webflow we connect for you on request; Wix and Squarespace get a copy-paste queue your team posts from.

You pay per published article from a prepaid balance, with no subscription, and a draft that fails the checks is not charged. Fixed prices per published article: Lite £9, Standard £19, Pro £39 and Premium £69 — about $11, $24, $50 and $88, charged in GBP. Length does not change the price, only the tier does; the tiers are compared on pricing. Twelve Standard guides a month is therefore £228. When the balance runs out the next article does not start, so top-ups are the one piece of housekeeping this asks of you.

Publishing is not the end of the job. A nightly loop reads Search Console per article, spots pages that have started to decay and queues a refresh, which runs the research and grounding stages again against the catalogue as it stands then. The same loop tracks whether GPT and Perplexity cite you. Start with the free site check.

The cases where this is the wrong purchase are specific enough to list.

• A catalogue of under about fifty lines with no supplier documentation. Grounding is what makes a guide hard to copy. With a thin catalogue and no datasheets, articles rest on public research alone and compete on the same terms as every other store's.
• You want product descriptions or category copy rewritten. The published unit is an article. Per-product description work is a different job with different constraints and is not what arrives through the plugin, the SDK, the feed or the webhook.
• You are on Shopify and need it live this week. There is no self-serve Shopify connector yet; we set the store up on request, which means a conversation and a short wait rather than a plugin install.
• You need to approve each guide before it publishes. Passing the checks is the gate. Takedown, per-site pause and a global off switch are the controls, and they all operate after publication.
• Your margin problem is conversion, not traffic. If the pages you already rank for are not converting, more articles feeding them will not fix it, and we would rather you spent the money on the checkout.

There is also a limit worth naming. The engine reads your catalogue to write and to refresh, but it does not watch stock levels minute by minute; a guide is re-grounded when the nightly loop queues it for a refresh, not the moment a line sells out.