Content SEO fails in predictable places: a page written for a query that wanted a different kind of page; a brief that was really a keyword list; a draft nobody checked against the brief; a published page nobody looked at again. The Content Loop closes those holes with six stages and four gates. The stages are Intent, Page type, Brief, Production, Optimisation and Refresh-or-prune. The gates sit between stages and have a simple rule: a page that fails a gate goes back, not forward. The loop is the operating version of my content SEO guide.
- Intent
- Page type
- Brief
- Production
- Optimisation
- Refresh / prune
When to use this
Use the loop when content is produced regularly by more than one person: an in-house team, freelancers, an agency, or a mix. It is designed for the hand-offs, because that is where quality leaks. It is also the right tool when a site has a few hundred pages and nobody is sure which ones are earning their keep; the Refresh-or-prune stage is built for that.
It is overkill for a founder writing one post a month, though the four gates still make a useful personal checklist. It also assumes you have decided which topics matter. That decision belongs to the topical authority framework; this loop takes a node on that map and turns it into a page that performs.
The six stages and four gates
Stage 1: Intent. What does the searcher actually want to happen?
Inputs: the target topic node, the live SERP for its main queries, People Also Ask, the query variants Search Console already shows. Actions: classify the dominant intent (learn, compare, do, buy, find) and, more importantly, the outcome the searcher wants: a decision made, a task completed, a number found. Note where Indian searchers phrase things in Hinglish or regional terms. Outputs: an intent statement in one sentence: "Someone searching this wants to X so that they can Y." Checks: the live top results agree with your classification. If they do not, either your reading is wrong or there is an opportunity, and you should know which. My search intent guide covers the classification.
Gate 1, the Intent gate: no page proceeds without an intent statement that a second person agrees with.
Stage 2: Page type. What shape of page serves that intent?
Inputs: the intent statement, the SERP's current page types, your site's templates. Actions: choose the page type: guide, comparison, tool, template, category page, product page, glossary entry, FAQ, calculator. Decide whether it is a new URL, an addition to an existing page, or a merge. Outputs: a page type, a URL decision, and a target template. Checks: the chosen page type exists in the current results, or you have a specific reason it should. A 3,000-word article is not the answer to a query the SERP answers with a table.
Stage 3: Brief. Specify the page so that a stranger could write it.
Inputs: intent statement, page type, topic node and its siblings, subject-matter expert input, competing pages. Actions: write the brief: the question the page answers, the sections and what each must contain, the expert's facts and examples, the pages it must link to and from, the entities it should naturally cover, and what it must not cover because a sibling page does. Outputs: a brief document. Checks: the brief contains at least one thing no competitor page has, sourced from a real expert. The content briefs guide has the full structure.
Gate 2, the Brief gate: no drafting starts until the brief has an expert contribution and a linking plan.
Stage 4: Production. Write it, and check it against the brief, not against taste.
Inputs: the brief. Actions: draft, with the writer free to restructure if the brief is wrong, provided they say so. Review in two passes: brief compliance (sections covered, expert material included, sibling topics avoided) and editorial quality (clear, specific, correct). Outputs: a reviewed draft and a compliance note. Checks: every section in the brief is present or consciously removed; every claim has a source or an owner.
Gate 3, the Publish gate: the page ships only when it passes brief compliance, has its planned internal links in place, and has a title and description written for the SERP rather than for the CMS.
Stage 5: Optimisation. Tune after Google has seen it, not before.
Inputs: Search Console query data after a few weeks, click-through by query, the live SERP again. Actions: look at the queries the page actually gets impressions for. Move the title and headings towards the phrasing searchers use. Add a section if the page ranks for a question it answers badly. Check the snippet with the SERP snippet preview tool; add structured data where the page type supports a rich result. Outputs: a short change log for the page. Checks: impressions are rising for queries that match the intent statement. If the page is picking up impressions for a different intent, that is a signal for a new page, not for stretching this one.
Stage 6: Refresh or prune. Decide, on a schedule, what each page deserves.
Inputs: twelve months of page-level performance, the coverage map, the page's age and last update. Actions: every page is reviewed on a cadence (quarterly for commercial pages, half-yearly for the rest) and given one of four decisions: keep, refresh, merge, remove. Refreshing means re-running from Stage 1, because intent drifts. Merging means consolidating with a redirect. Removing means a 410 or a redirect to the closest parent. Outputs: a dated decision per page. Checks: no page goes a year without a recorded decision.
Gate 4, the Review gate: a page without a review date is a page that will quietly decay. Every URL gets one.
Worked example (hypothetical)
Consider a made-up personal finance site in India covering mutual funds. The topic node is "SIP versus lump sum investment". This is an example, not a client.
Intent. The live results are mostly explainers with a comparison table and a calculator or two. The intent statement: "Someone searching this wants to decide which method to use for money they have right now, so that they can start investing without regretting the timing." Note the outcome: a decision, not an education.
Page type. A comparison page with a decision section, not a generic guide. The site already has a thin page on the topic ranking on page three, so this is a refresh and expand on the existing URL, not a new one.
Brief. Sections: the difference in mechanism, when each wins, a decision section driven by three questions (is the money already available, how volatile is the fund, how disciplined is the investor), an illustrative calculation, and an FAQ. The expert contribution is a financial planner's rule of thumb about splitting a lump sum, which no competing page has. Must link to the SIP guide and the fund selection page; must not explain what a mutual fund is, because a sibling page does.
Production, Optimisation, Refresh. The draft passes brief compliance on the second attempt, after the writer had dropped the decision section as "too opinionated". Six weeks after publishing, Search Console shows impressions for "SIP vs lump sum during market fall", so a section is added. The page is tagged for quarterly review because it is commercial.
The quality gate scorecard and the refresh-or-prune table
Each gate is scored on three criteria from 0 to 2, so a gate's maximum is 6 and a pass is 5 or more. A writer and an editor can then disagree about one criterion rather than about the whole page.
| Gate | Criterion 1 | Criterion 2 | Criterion 3 | Pass at |
|---|---|---|---|---|
| G1 Intent | Intent statement names an outcome | Live SERP agrees with classification | Second reviewer agrees | 5 / 6 |
| G2 Brief | Expert contribution present | Linking plan (to and from) present | Sibling exclusions listed | 5 / 6 |
| G3 Publish | Brief compliance confirmed | Internal links live | Title and description written for the SERP | 5 / 6 |
| G4 Review | Review date recorded | Performance data attached | Decision recorded (keep, refresh, merge, remove) | 5 / 6 |
For the Refresh-or-prune stage, I use a second table that turns two questions into a decision: does the page have demand (impressions for queries matching its intent), and does the page have a place (a node on the coverage map that nothing else serves)?
| Demand | Place on the map | Performance trend | Decision | Action |
|---|---|---|---|---|
| Yes | Yes | Stable or rising | Keep | Light optimisation; next review per cadence |
| Yes | Yes | Falling | Refresh | Re-run from Stage 1; intent may have drifted |
| Yes | Shared with another page | Any | Merge | Consolidate into the stronger URL, redirect the other |
| No | Yes | Flat | Refresh or keep | Check if the node matters; if yes, rewrite; if no, keep as support |
| No | No | Flat or falling | Remove | Redirect to nearest parent, or 410 if nothing fits |
Common failure modes
The brief is a keyword list
"Include these 40 terms" is not a brief. It produces pages that mention everything and explain nothing. A brief specifies questions, sections, expert input and links.
Treating AI drafts as finished
Generated drafts are fluent and generic. Without the expert contribution and the brief-compliance pass, they are the same page every competitor can produce. See the AI SEO guide for where these tools genuinely help.
No sibling exclusions
Every page tries to be complete, so five pages explain the same basics, and they cannibalise. The brief must say what the page leaves to its siblings.
Measuring publishing volume
Pages per month is an input. Queries per cluster, clicks on intent-matched queries and conversions from organic are outputs. Report the outputs.
Run it yourself
- For the next page in the queue, write the intent statement with an outcome, and get a second person to agree.
- Check the SERP's page types; choose yours; decide new URL, expansion or merge.
- Write the brief with sections, expert input, links to and from, and sibling exclusions.
- Score G1 and G2 before drafting. Do not draft on a failed gate.
- Review the draft for brief compliance first, editorial quality second.
- Score G3; ship only with links live and SERP-facing title and description.
- Four to six weeks after publishing, tune title, headings and sections against real queries.
- Assign a review cadence and record it. Quarterly for commercial, half-yearly for the rest.
- Run the refresh-or-prune table across the whole library at least once a year.
When I run content programmes through my content SEO service, this loop is the operating system, with the technical side alongside it, because a well-briefed page on a template that fails the audit gates still goes nowhere. The SaaS organic growth case study shows the loop in a product context.
FAQ
Does every page really need a written brief?
Every page that costs money to produce, yes. A brief takes an hour and prevents a rewrite that takes a week. For short FAQ additions it can be a few lines, but the intent statement and sibling exclusion still need to exist.
Can this framework work with AI-assisted writing?
Yes, and it is designed to. The gates are where human judgement is required: the intent decision, the expert contribution, brief compliance and the review decision. Drafting between those points can be assisted. What cannot be automated is the thing that makes the page different from every other page.
How does the loop handle pages written for AI search engines?
The same way. Pages that get cited by AI search systems tend to be the ones with clear structure, specific claims, and a real answer near the top, which is what a good brief produces anyway. I am testing this directly in the AI search citation experiment.