The SEO system: from search demand to article refreshes

The SEO module is a search content management system, not a “write an article” button. The goal is not the number of texts but the number of relevant search queries your site can earn visitors from. We do not guarantee rankings, traffic or indexing time: they depend on your site, competition and the search engine.

The full cycle

1. Site audit. Availability, robots.txt, sitemaps, existing pages, duplicates and technical problems.
2. Search demand. Queries from Search Console, Yandex Wordstat and Webmaster, SEO data providers and your CSV/XLSX exports. Unknown volume is shown as “no data”, never as an invented number.
3. Intent and clusters. Informational, commercial, transactional, navigational, local and support queries. Different intents are never merged into one page.
4. Topic map and priorities. What is covered, what is missing and where pages compete with each other. Every priority comes with an explanation.
5. Brief. Primary and secondary queries, audience, reader questions, structure, required sources and evidence, unique value and target action.
6. Article. The format follows the intent — each format has its own structure.
7. Quality control. A separate stage before publication.
8. Publishing. Through your connected CMS — drafts only by default.
9. Monitoring and analytics. Post-publish checks, indexing, impressions, clicks, conversions.
10. Refresh. The service finds articles that are worth improving instead of writing new ones.

Article formats

Informational article, step-by-step guide, comprehensive guide, FAQ, review, comparison, tools roundup, commercial landing page, e-commerce category, service page, reference article, case study, research and update of an existing article.

Rules for every format:

• the first paragraph answers the main question directly;
• facts, numbers, prices and specs come only from verified sources or from data you entered in the brief;
• no invented reviews, experts, certificates, case studies, statistics or personal experience;
• when data is missing the text keeps a “[DATA NEEDED: …]” marker and publication is blocked until you add it.

Case studies and research are written only from your own data. Structured data (Article, FAQPage, HowTo, Product) is added only when it matches the page content.

Sources

For factual content the service collects sources, fetches the pages and stores which claims each source supports and where it is used. A claim counts as supported only if the quote really exists on the source page. Unavailable or contradictory sources are flagged.

Quality control

Scores from 0 to 100 in eight dimensions, each explained: factual reliability, completeness, usefulness, originality, intent match, editorial quality, technical SEO and publish readiness.

• 85–100 — may be auto-published if there are no blocking issues and the mode is enabled;
• 70–84 — needs revision or review;
• below 70 — publication is blocked;
• any blocking issue (invented reviews or experts, personal data leaks, verbatim copying, a duplicate of an existing page, a missing-data marker) — publication is forbidden regardless of the score.

Thresholds are configurable per project. Weak articles are sent back for an automatic revision based on the findings, with limits on attempts and cost. The service also checks similarity with your published pages and query cannibalization, suggesting to update the existing page, merge (with a 301 redirect plan) or retarget the new article.

Sensitive topics

Finance, health, legal and safety topics are checked more strictly: every factual statement needs a source, auto-publishing is off, and before publishing you confirm that a professional has reviewed the content.

Internal links

Links are chosen from your site's pages by meaning, with a natural anchor that already exists in the text. A link is inserted automatically only with enough confidence and within a sensible limit; the rest stay suggestions. Orphan pages without inbound links are reported.

CMS publishing

Supported: WordPress (including Yoast SEO and Rank Math fields when exposed to the REST API), Ghost, 1C-Bitrix (REST webhook), a custom webhook or REST API, and our widget feed. Tilda does not allow creating pages via its API, so the service prepares an HTML block to paste and you confirm the page URL afterwards.

Modes: drafts only (default); auto-publish after a passed quality check; auto-publish with confirmation for selected categories; automatic updates of published articles within allowed changes. Retries never create duplicates, CMS-side edits are never silently overwritten (you resolve the conflict), and every version can be rolled back without changing the URL.

Indexing and monitoring

After publishing the service checks the status code, canonical, robots, sitemap and internal links, and notifies IndexNow if you enabled it and host the key file. Submitting a URL is not indexing: “indexed” is shown only from Search Console or Yandex Webmaster data. Daily monitoring reports 4xx/5xx errors, accidental noindex, robots.txt blocks, wrong canonicals, unexpected redirects and broken links.

Analytics and refreshes

Connect Search Console, Yandex Webmaster, GA4 or Metrica — every number shows its source, period and completeness. Search impressions, visits and conversions are never mixed into one metric. Opportunities include high impressions with low CTR, near page-one positions, declines (seasonality-aware), unanswered queries, outdated facts, few internal links, cannibalization and low conversion. Accepting one gives you a new brief version, an updated article, a quality report and a version diff. Changing only the date is not an update.

Costs

The report shows the cost of research, generation, checks, publishing and refreshes, plus the cost per published article, per article with clicks and per organic visitor — when the data allows it.