The shopper questions your category pages were never built to answer

Retail content decays for a reason most content does not: the catalogue changes underneath it. A model sells out, a supplier discontinues a line, a successor arrives with a different part number, and a guide written six months ago recommends three products you cannot ship. The shopper clicks through to a dead product page and leaves.

Two mechanisms keep the articles and the catalogue in step. Grounding comes first. The engine retrieves your product data before it writes, so a guide recommends what your catalogue says you sell rather than what a model half-remembers from training. The draft then goes to a fact-check judge from a different vendor than the writer, which tests the claims against that retrieved material instead of trusting fluent prose.

Then the nightly loop reads how each published article performs in Search Console and queues a refresh when it starts to decay. The refreshed copy passes the same checks as a new article before it replaces what is live, which is how a superseded model gets swapped for the one that replaced it without anyone opening the post editor.

Retail demand runs on a calendar. Garden equipment, dehumidifiers and gift guides each have a season, and the searching starts before the spending does. A guide published on the day demand peaks has not been crawled, indexed or linked to by anything; it arrives as the wave is breaking.

The engine treats seasonality as a scheduling problem. Articles drip out on a steady cadence rather than in batches, which puts a seasonal guide on the site ahead of the season it serves and gives it time to be indexed and to gather internal links. IndexNow tells participating search engines the URL exists on the day it publishes rather than waiting on a crawl.

Seasonal pages are also the clearest case for refreshing rather than republishing. Last year's guide keeps its URL, its inbound links and whatever history it has earned, and the nightly loop can queue it for a rewrite against this year's range; the rewrite goes through the same checks as a new article before it goes live. Retiring the URL and publishing a fresh one every autumn throws that history away.

Structured data does two different jobs in online retail, and it pays to keep them apart.

The article layer is what the engine ships. Every published article carries JSON-LD describing the article, an FAQ block with matching schema, a hero image and internal links into the products and category it discusses. Google narrowed FAQ rich results to government and health sites in 2023, so what that block earns now is a clean question-and-answer pair for any parser to lift, which is the unit an assistant quotes when it is summarising rather than listing.

Product schema sits on your product pages, and it is the layer most often broken: a missing price or availability field, markup that disagrees with what the page displays, or an offer block that never changes when the item sells out. No article layer compensates for that, because the rich result a shopper sees against a product is generated from that product page's own markup.

The free site check reports what your pages emit today, and the AEO Checker looks at whether your answers are structured so an assistant can lift them. Repair the product pages first, then build the article layer above them.

Each retailer selling a given SKU received the same manufacturer description. Search engines have met that text on every site stocking the line, and so have the models trained on the open web. Content written from it is a summary of a summary, which is why so much retail content reads interchangeably.

Your own data is on nobody else's site. Where it exists, IT Master ingests it read-only and grounds the articles in it:

• product data, variants and what is in the box, from your catalogue
• supplier datasheets and manuals you hold for the lines you sell
• support questions, and the reasons customers actually sent things back

A comparison written from that can say which of two models needs a separate power supply, or which one buyers returned over a firmware fault. See first-party data for how the retrieval works.

Checking is deliberately adversarial. Writers are Claude models; the judges are Gemini and GPT, from a different vendor, running a fact-check, a novelty score against the competitor pages fetched during research, an E-E-A-T critique and an AI-tell pass. Across 186 Standard-tier runs, 51% of drafts passed every check first time; the rest were revised or rejected.

Retail stacks vary more than most. WordPress and WooCommerce stores install the plugin and paste one key, and posts arrive with the Yoast or Rank Math fields filled. Next.js storefronts use @itmaster/sdk. Anything else reads the REST pull feed or takes a signed webhook at an endpoint you name. Shopify and Webflow are set up by us on request; Wix and Squarespace get a copy-paste queue instead.

You pay per published article from a prepaid balance, with no subscription, and a draft that fails the checks is not charged at the full rate. The four tiers — Lite, Standard, Pro and Premium — are on pricing.

After publish, the nightly loop reports what each article did in Search Console, and asks GPT and Perplexity the questions those articles answer to record whether your store is named in the reply. That is your citation rate, measured rather than assumed, next to the impressions and clicks. If a persona view suits you better than a sector one, ecommerce stores describes the same engine from the owner's side.