Articles

An article has one page in Trakkr, from its first streamed word to a year after it ships. That page is where you watch it being written, edit it, publish it, and later find out whether an AI engine ever quoted it.

The layout stays the same and the contents change with the state.

The five passes

Generation runs in order and the stage label on the row tracks it: Research, Draft, Enhance, Verify, Score. Research is the part that matters most and is easiest to miss. Before a word is written, Trakkr assembles your knowledge base entries, your voice profile, the template sections, internal link candidates from your site index, and the pages AI engines currently cite for that query, including the snippet each engine used.

Blocks stream into the page as they are produced, so the opening is readable while the closing sections are still being written. A typical article finishes in a couple of minutes.

If a run fails, the page shows the error with a Retry generation button and the credit is returned automatically. Retrying re-runs the whole pipeline against current research, not a cached copy.

Editing the draft

Click any paragraph, heading or list to edit it in place. Escape cancels, Cmd+Enter (or Ctrl+Enter) commits, and clicking away saves. The header shows when the last save landed.

Each section carries a quiet menu with Regenerate section. That rewrites the section's body prose while leaving its heading, quotes, statistics, lists and source blocks untouched, so a weak paragraph can be replaced without disturbing the structure around it.

Every save writes a checkpoint. The History menu in the header lists them as the generated draft, your edit checkpoints, and a snapshot taken before any restore, so rolling back never loses the version you rolled back from.

Editing needs edit access on the brand. Viewers see the same page as read-only.

Checks and citation likelihood

The rail carries two instruments.

Citation likelihood is a 0 to 100 score with a bar for each engine underneath it. It is a heuristic read of the finished document, weighted from articles that were and were not cited: answer-first structure, question-shaped headings, sourced statistics, quotes, freshness and length. It updates when you save.

Checks are deterministic. Eight of them run on every draft, and each failure is written as a sentence you can act on rather than a lint code.

Publishing

Publish offers three routes, and all three end in the same place: a live URL Trakkr can watch.

Connecting a CMS is covered in Sites. A push from this sheet always lands as a draft post, never a published one, so the final say stays with you.

Binding a URL runs a health check first. The result is honest about what it found.

Exporting and forgetting to come back is the common failure, so Trakkr scans your sitemap for a page whose slug matches an unbound article and offers it as a one-click suggestion on the article page. Recheck re-runs the health check whenever you want.

After it goes live

Published articles gain a Performance view alongside the content.

The gap between AI crawls and citations is the useful reading. Crawls with no citations means the page is being read and passed over, which is a content problem. No crawls at all is a reachability problem, and Crawlers is where to take it.

The page also joins your Pages records, and the article header links straight to its page record once it appears there.

Campaigns

A campaign is a topic you are writing to own, not a folder. It holds a set of target queries, a brief that steers every article in the set, and the articles themselves.

Coverage is the number that matters. For each target query, Trakkr asks whether AI engines cite your brand when they answer it, using the same citations snapshot the ideas engine reads. A baseline snapshot is frozen when the campaign starts, so the header can show today's coverage against where you began rather than a number with no reference point.

The campaign page lists each article with its stage, checks passed, live URL and citation count, alongside the target queries with their coverage state and the editable brief. You can add or remove target queries at any time, though a campaign needs at least one.

When articles have been live for more than 21 days without earning a citation, the header offers Refresh stale. That re-queues those articles through the full pipeline so they pick up current research. Each refreshed article spends a credit.

Common questions

How much should I edit a generated draft?

Enough to add what only you know. The structure, sourcing and schema are handled, so your time is best spent on judgment: the claim being made, the angle, and anything the knowledge base did not carry. Rewriting from scratch defeats the point.

What happens if I edit after publishing?

Edits save to the article record in Trakkr. They do not push back to your CMS. Update the live page yourself, because the live page is what the engines read and what citation tracking measures.

Can I write in Trakkr without generating anything?

Yes. Create an article from the composer, then edit the blocks by hand. You can also skip Trakkr's editor entirely, publish elsewhere, and bind the URL so the tracking still works.

Why did my score drop after I edited?

Citation likelihood recomputes on save and the checks re-run with it. Cutting a sourced statistic, removing a question-shaped heading, or pushing the article well past its word target will each move it. The rail names which check flipped.

Does regenerating a section cost a credit?

No. Credits are spent per article generated, including retries that come from a refresh. Editing, regenerating a single section, restoring a checkpoint and rechecking a URL are all free.

What counts as a citation for the article?

An AI answer that references your published URL. Trakkr matches on the bound URL, which is why binding matters more than anything else in the publish step. An article with no URL on record can never show a citation, however well it was written.