Content and the editor

Briefs, drafts, the editor, the live scorecard, claim checks, internal links and images.

Content is where a move becomes a page. LogNorm researches a brief from what ranks and from your Company Brain, writes a draft you edit in the browser, scores it while you work, and takes it through review to your CMS. This page walks through each step. For the overview, see Content studio.

The content pipeline

Open Content. The top of the page shows your pipeline (briefs, drafting, in review, approved, published), words drafted, average quality and pieces published in the last 30 days. Below it, every piece for the current website sits on a board or in a list. You can filter by stage, content type and owner (Everyone, Mine, Unassigned or a teammate), or search by title or keyword.

StageStatusesWhat it means
IdeasIdeaPlanned in the Blog planner, no brief yet.
BriefsBriefA researched brief, no draft yet.
DraftingDrafting, Draft, Changes requestedThe AI is writing, or the draft is being edited.
In reviewIn reviewWaiting on a reviewer.
ApprovedApprovedReady to publish.
PublishedPublishedLive, and being measured.

On the board you can drag a card to the next column: to Briefs to research, to Drafting to write, to In review to pick a reviewer, to Approved or back to Drafting with a note, and to Published to open the publish flow. LogNorm explains when a move isn't allowed, for example dragging a piece without a draft into review. Published pieces stay published; open the article to publish an update.

From move to brief

Most pieces start from a move. Pick New content on the Content page and choose the move under From a move; the title, target keyword, type and URL fill in from it. You can also write about any keyword without a move.

  1. Set the basics

    Enter a Working title (at least three characters) and a Target keyword.

  2. Pick a type

    Blog post, Landing page, Comparison, Alternatives, Integration page, Use case or Refresh. The type sets the brief's guidance and its default length. Refresh also asks for the Live page URL.

  3. Research now, or later

    Leave Research a brief now on and choose Create and research. The switch shows the most credits the job can use. Turn it off to create an empty piece and research later from the Plan tab or the ⋯ menu.

What the research does

Research usually takes about a minute. LogNorm:

  1. Pulls the Google results for the keyword, with People Also Ask questions and any AI Overview (a recent snapshot is reused if one exists).
  2. Reads the top ranking pages, skipping your own sites, and notes their headings, length and which sub-topics most of them cover.
  3. Collects the keywords for the page: the target, the keywords your cluster plan assigns to this article, and Search Console queries you already get impressions for.
  4. Reads what AI engines answer today for your tracked prompts on the subject: whether they mention you, who they name and which pages they cite.
  5. Looks for facts worth citing. A fact is kept only when LogNorm opened the source, found the figure on that page, and rated the site as original research, an official source or an established publication. Competitors, vendor marketing and content farms are never cited.
  6. Searches your Company Brain for the passages most relevant to the keyword and title, and picks pages on your site worth linking to.
  7. Plans the article in one AI call, using the linked move's context when there is one.

The plan lands in the Plan tab: audience and search intent, the angle, What only we can say, the outline, questions to answer, Must cover items, a target length and the ranking pages analysed. Every outline section says why it is there and what evidence backs it, and Why this plan explains the larger choices: the angle, the length, which keywords were prioritised and what was left out. The Research tab lists the verified sources, the AI answers, your Search Console queries and every page and Brain passage used. Use Re-research to run it again.

Your notes for the writer (on the start screen and in the Plan tab) steer both the plan and the draft: who to write for, what to lead with, what to leave out.

A thin Brain makes a generic article
Differentiation comes only from your Company Brain and company details. If the Research tab shows no Brain passages, add positioning and product docs before you draft.

The draft and the editor

Choose Research and write to do both in one go, or Research only to check the plan first and then Write the draft. Write it myself starts from a blank page. The AI writes the whole article in one pass with the plan in view (two passes for long pieces), so sections don't repeat each other. Then it reviews its own draft: stock phrases and other signs of machine writing, sections that are generic or don't answer their heading, claims that contradict your Company Brain and must-cover topics it missed. Sections that fall short are rewritten once. It usually takes two to three minutes, and you can leave the page.

When it finishes, LogNorm places internal links, opens every link in the article (a dead link, a competitor or a weak source is removed), builds JSON-LD schema, saves a version and runs the quality checks. How this draft was written, at the top of the Plan tab, shows what the self-review changed, which links were removed and why, and what the AI models cost.

Drafts follow strict grounding rules. Facts about your company (features, pricing, customers, numbers) may only come from your Company Brain and company details. Statistics, customer names and quotes are never invented: a figure appears only when it came from a verified source, linked where it is used. When no figure passed the checks, the article states none.

The editor saves as you type; the header shows Saving…, Saved or Unsaved. It is read-only while the AI is working on the piece. The toolbar has paragraph and heading levels, bold, italic, strikethrough, inline code, lists, quotes, code blocks, dividers, links and image upload at the cursor.

Feedback, and how LogNorm learns your style

Tell LogNorm what you would change and it fixes the draft and remembers the preference, so the next draft needs less.

  • On a passage. Select any text in the article and choose Give feedback. Say what should change, or pick Shorter, Simpler, More specific or Less like AI. Fix now rewrites that passage and shows you the result before anything changes; Save comment keeps it for later. Passages with open comments are highlighted in the article.
  • On the whole article. Write it in the Feedback tab ("too formal throughout", "cut the intro in half").
  • Apply feedback rewrites only the sections your open comments touch. Each comment then shows what was changed, and the text from before is kept in version history. A reviewer's note can be handed to the AI with Have the AI apply this.

When a comment is a lasting preference ("never open with a question", "say teams, not users") rather than a one-off correction, LogNorm saves it as a rule and tells you. Saying the same thing again makes the rule stronger. When you approve a piece, LogNorm also compares it with what the AI wrote and suggests rules from the pattern of your edits; those wait for you to keep or dismiss. Rules, and the voice LogNorm can learn from two or three of your own posts, live in Company Brain under Writing style, where you can add, switch off or delete them. Every draft, revision and refresh follows them.

The right rail

TabWhat it's for
PlanThe plan and its reasons: angle, outline with why each section is there, questions, must-cover items, ranking pages, your notes for the writer, and how the draft was written. For refreshes, the refresh plan.
SEOTitle (H1), meta title, meta description, slug, target keyword, a search preview and the Schema (JSON-LD), with Regenerate. Keywords shows each keyword the plan chose and whether the article uses it, updating as you write.
LinksSuggested internal links and the ones already in the article. Link check opens every link and rates outside sources; run it again after edits and before publishing.
MediaThe hero image, body images and the image library.
ScoreThe live scorecard: sections, must-cover items and claims.
QualityThe quality gate: pass, warn and fail checks.
ResearchVerified sources the article may cite, what AI engines answer today, Search Console queries, ranking pages, Company Brain passages and the live page used.
FeedbackComments for the AI writer and notes between people. Change requests land here.

The ⋯ menu has Regenerate draft (the current text is kept in version history), Re-research brief, Save a version, Version history (restore any snapshot; your current text is kept too), Export as Markdown or HTML, and Delete. Deleting removes the brief, draft, versions and comments; the linked move stays in your backlog.

The Score tab

The Score tab is a live scorecard from LogNorm's decision AI. It re-scores a few seconds after you stop typing and marks itself Edited since the last score until it catches up. Use the refresh button to score again immediately. Unchanged sections are reused, so re-scoring costs little.

  • Whole article: search intent, audience fit and originality (Strong, Fair or Weak), plus two flags. Answers first asks whether you get to the point, and Quotable by AI answers whether it has passages AI assistants can quote. When the model is unsure, a flag reads Your call.
  • Sections: each H2 is judged on whether it delivers on its heading, stays on intent, is specific and avoids filler. Issues are listed under the heading. Click a section to jump to it in the editor.
  • Must cover: each item from the brief, marked Covered (with the section it's in) or Missing (with the closest section).
  • Claims: every figure, superlative and product claim is checked against your Company Brain, company facts and the research.
Claim verdictMeaningWhat to do
SupportedA passage backs it. The evidence and its source are shown.Nothing.
ContradictedYour sources say otherwise.Fix the sentence, or update the Brain if the Brain is wrong.
Not in your BrainNo passage mentions it.For product claims, add a source or remove the claim.

Contradicted claims and unbacked product claims are listed first. Click a claim to select the sentence in the editor. Judgements inform the gate; you decide what ships.

The Quality tab

Quality checks run after every draft. Choose Re-run after edits. Each check passes, warns or fails, and the score (0 to 100) counts a pass as a full point and a warning as half. Checks are grouped into Search, Structure, Readability and Trust, failures first. Checks marked AI-judged come from the scorecard.

CheckPasses when
Meta title length30–60 characters
Meta description length70–155 characters
Target keyword placementIn the title, H1, intro and slug
Heading structureOne H1, at least two H2s, no skipped levels
Length vs target85–150% of the brief's word target
Readability (Flesch)Score of 45 or more
Internal linksAt least three
External sources citedAt least one
CannibalisationNo page or other draft targets the same topic
Claims backed by Company BrainNo unbacked figures or claims
Hero imageA hero with useful alt text
Body images and alt textEvery image has alt text
Structured dataJSON-LD present, including FAQPage if there's an FAQ
FAQ sectionThe article has one

The AI-judged checks add Matches search intent, Answers first, Section quality, Covers the brief, Originality, Fits the audience and Quotable by AI answers. Failing checks don't block publishing, but reviewers see them when you request review.

When LogNorm drafts, it matches the article against your crawled pages and inserts up to five internal links using phrases already in the text. Other good candidates, and pages the brief suggested, wait in the Links tab under Suggested, each with a reason. Choose Apply to link the suggested anchor, or select your own words in the editor first and then apply. Remove a link from In the article. No suggestions usually means your site hasn't been crawled; run a site analysis.

Images and hero covers

If your plan includes AI images and you have an image model set up (a Google, OpenAI or xAI key; see AI providers), LogNorm can generate a hero image and body images for the most visual sections, each with alt text and a caption. By default this runs automatically after each draft. Configure the model, house style, hero aspect ratio, body images per article (up to four) and the automatic setting under AI providersImage generation.

In the Media tab you can generate missing images, regenerate the hero or all body images, insert library images at the cursor, set any image as the hero, edit alt text and captions, copy an image URL or delete it. You can always upload your own.

Cover layouts

A hero is assembled from a layout, your logo, the article title and the generated art. In the hero image toolbar, choose Layout, title and logo to pick a preset: Split, Overlay, Card, Editorial, Spotlight, Type only or Art only. You can also toggle the title and logo, add a label above the title and set an accent colour. Changing the layout re-renders the cover instantly without an AI call; the art is reused. With no art at all, Design a cover builds a type-only cover from your logo and title, with no AI and no credits. Set the default layout and logo under AI providersArticle covers.

Review and approval

  1. Request review

    Choose Request review, pick a reviewer (Owner, Admin or Editor) and add an optional note. The piece and its linked move go to In review, a version is saved and the reviewer is notified.

  2. Approve or request changes

    The reviewer chooses Approve or Request changes. A change request needs a note, which is posted to Comments; the piece goes to Changes requested.

  3. Publish

    Approved pieces show Publish. See Publishing to your CMS.

Contributors can create and edit pieces (they see only pieces assigned to or created by them) but can't approve or publish. Use Reopen as draft to pull an approved or in-review piece back. If the review workflow isn't in your plan, Owners, Admins and Editors approve drafts directly. Assign an owner from the avatar in the article header.

Refreshing a published page

A Refresh rewrites a live page instead of starting from scratch. LogNorm reads the live page, checks what ranks for its keyword, pulls the page's Search Console queries for the last 90 days and searches your Company Brain. Before rewriting, it records its call on whether a full rewrite is warranted. That's a recommendation; the refresh runs either way. The Plan tab then shows the refresh plan: a summary, every change and the queries behind it. Connect Search Console for the best refreshes.

The published-content audit

At the bottom of the Content page, Published content audit checks your published blog posts: crawled blog pages with at least 150 words, plus pieces you published from LogNorm. Choose Run audit. Each post is judged as Outdated, Thin, on or off search intent, and Needs refresh. Posts worth a look are listed with the reasons.

LogNorm automatically creates refresh moves for up to ten posts it is confident about. For any other flagged post, choose Create refresh move, or View move if one exists. The audit costs a fraction of a cent, and unchanged posts are free on the next run.

The Score tab and the content audit run on LogNorm's decision AI. If it's ever unavailable, the Quality tab's measured checks still run.

Costs

Briefs, drafts, refreshes and article images use credits. Each button shows the most a job can use, and you're charged only for what it actually used. With your own AI key, AI tokens are free but a small platform fee and data calls still apply. See Credits and billing.