A steady flow of articles that are not rewrites of the SERP

Thirty posts imported on a Monday and nothing for the rest of the month gives a crawler little reason to come back, and buries post twenty-nine under post one on the day it matters. IT Master releases articles on a steady drip instead: passing drafts go out across the days you choose, at the rate you set, so the archive grows at a regular pace.

Each release arrives finished rather than as a body of text to dress:

• JSON-LD schema for the article and its FAQ block, so any parser reading the page gets clean question and answer pairs instead of a wall of prose.
• Internal links into your existing archive, so a new post joins a cluster instead of sitting as an orphan.
• A hero image and an FAQ section, with the questions taken from People Also Ask and related-query data pulled during discovery.

The moment a post is live, IndexNow tells the engines that participate, Bing among them. Google does not take IndexNow, so it finds the page the ordinary way, through your sitemap and the links pointing at it. On a large archive the cadence is also a crawl budget question.

The fastest way to produce an article is to read the top ten results and summarise them. The output is an eleventh page saying what the first ten said: the thin-content pattern above, produced faster.

IT Master reads the competitors too, but as a benchmark rather than a source. Research fetches the ranking pages, the forum threads and the People Also Ask questions around the topic, then looks for what those pages leave out. Once a draft exists it is judged by models that did not write it, each check including at least one model from a different vendor: a novelty score against every competitor page fetched and a fact-check, both led by Gemini and GPT, plus an E-E-A-T critique and AI-tell detection from panels pairing a Claude judge with a GPT or Gemini one. A failing draft goes back to the writer with the specific objections, for up to three rounds, and is rejected if it still fails.

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 rejections are the part that protects an archive. Cross-model validation describes the mechanism in full.

An archive decays in a predictable way. A post ranks, somebody publishes something fresher, the query drifts, and the page slides to the third page without anyone noticing until the traffic review.

A nightly loop reads your Search Console data for every article the engine has published and watches impressions, clicks and position over a trailing window. Pages losing ground are queued for a refresh: re-researched against what ranks now, rewritten where they have gone stale, put back through the same checks, and republished at the same URL.

The same loop puts a fixed set of questions from your subject to GPT and Perplexity and records whether your page is named in the reply. That citation rate sits beside the Search Console figures, so you can see which posts assistants use as a source and which they pass over. Nobody can promise a citation; the point is that it is measured rather than assumed. The free AI visibility checker gives a first reading for any site before you connect anything.

A retailer grounds its articles in datasheets and stock data. A publisher has no catalogue, so it is fair to ask what first-party data means for a niche site.

More than you would expect. Your archive is first-party: the interviews, test notes, reader comments and corrections that no competitor holds. So are the questions readers send through a contact form or forum, the results of any survey you have run, and any dataset you have compiled for your subject. Where that material is available to the engine, it is retrieved before writing, and the article carries sentences that appear nowhere else on the web, which is what the novelty judge rewards.

Where it does not exist, the engine works without it rather than inventing it. The article then rests on the research stage: competitor gaps, forum threads and the questions people search. Grounded articles are harder for a competitor to copy; research-only articles are competent and checked, but they compete on the same terms as everyone else's.

You stay editor-in-chief. You set the scope and the cadence, and you can take any article down.

Most publishers run WordPress, and the WordPress plugin is the shortest route: install it, paste one key, and articles arrive as posts with the Yoast or Rank Math fields filled. Next.js sites use @itmaster/sdk; anything else can read the REST pull feed or take a signed push webhook. Shopify and Webflow we set up on request; Wix and Squarespace get a copy-paste queue.

You pay per published article from a prepaid balance. There is no subscription, no seat fee and no minimum, 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.

At twenty Standard articles a month that is £380; at sixty, £1,140. The four tiers differ in the writer model and in how demanding the judges are. When the balance runs out the next article does not start, which is the only scheduling constraint the billing model imposes.

This does not replace original reporting, and no tool can guarantee a ranking. What you get is a flow of articles that each clear a fact-check, a novelty threshold and an E-E-A-T critique before they touch the site, and a loop that notices when they slip. Start with the free site check, or read how it works.

Being specific about the misfits is more useful than another paragraph on the fits.

• You publish original reporting. Interviews, testing, court records, attending a thing and writing about it: no model does any of that, and a checked article about a subject is not a substitute for a piece about an event.
• You want ten thousand templated location or long-tail pages. We would argue against it, and the novelty check would refuse most of them anyway. If that is the plan, programmatic SEO sets out why.
• Your archive is already large and mostly decaying. The nightly refresh loop only touches articles the engine published itself. It will not rehabilitate ten years of posts written by your own team; it can write new articles that link into the ones worth keeping.
• An editor must read every piece before it publishes. Clearing the four checks is the gate. Takedown, per-site pause and a global off switch all operate after publication.
• Your revenue depends on a display network with a page-view floor you need hit this quarter. The engine drips articles out deliberately, and a new page takes months to earn traffic. Anyone promising you a step change on that timescale is selling you the thing this page exists to argue against.

A last honest limit: we cannot tell you how a search engine will treat any particular article, and nobody who can see your Search Console can tell you either.