Content review
Our content review policy and how pages get signed off
Our content review policy: every page is checked and scored before it can appear in search, and each approval is recorded by name and date.
Review stages
The stages every page goes through before it can appear in search
Every page has one of these statuses stored with it, and the status controls what happens to the page.
Brief
The brief names the reader, what they are searching for, the doubts the page must answer and the claims it must not make. If it is incomplete, the page fails validation before any writing starts.
Outcome: An approved brief, stored with the page
Draft
One writer drafts the page against the brief, starting with one question: which sentence could we not defend if a client asked where it came from?
Outcome: A complete draft with no gaps to fill in later
Subject check
Someone who knows the subject reads the page for accuracy: on a technical page, that the platform still works as described; on an industry page, that customers in that trade buy the way we say.
Outcome: Corrections, or a note that a claim needs a source
Search and conversion check
Two separate checks. The search check covers intent, overlap with our other pages, headings, metadata, internal links and whether the structured data matches the page. The conversion check asks whether the page is clear about who it is for, what it solves and what to do next.
Outcome: A page that answers the search and offers a clear next step
Legal or compliance check, where needed
For pages with commercial promises, regulated industries and anything that could read as professional advice. Our guarantee, privacy, terms and cookie pages have not been reviewed by a lawyer yet. The owner chose to publish them anyway, and each one says so.
Outcome: Sign-off, or wording that has to change first
Final QA
Every automated check runs at once: overlap with every other page, figures without sources, accessibility, mobile display, speed, broken links, status codes, canonical tags and schema.
Outcome: A pass, or a list of specific problems
Approval
The approver records the decision to publish, with his name and the date. The register keeps two kinds of approval apart. An editorial review means someone read that page from start to finish. Owner authorisation means the owner published a group of pages after the automated checks passed, and reviews them while they are live. Until one is recorded, the page is live and linked but kept out of search.
Outcome: An entry in the approval register, with its basis
Update or retire
We re-read published pages on a schedule, and whenever something they describe changes. If a page can no longer be defended, we update it or withdraw it.
Outcome: An up-to-date page, or a page taken out of search
Approval
Why no page can approve itself
Writers cannot fill in approval. The approval fields in the content files are always empty, and the record of who approved what lives in a separate register edited by hand. The two are combined only when the site is built.
So publishing is a deliberate step. A page can exist, display correctly and appear in the menu while search engines are told to ignore it. Only a name and a date against its address in the register lets it into search results and the sitemap.
Deleting that entry takes the page out of search on the next deploy, while the content stays and links keep working.
This is slower than publishing the moment a change is merged, and the site always has fewer pages in search than it has pages. We accept that trade-off.
Where the site stands today. Every route in the register carries a name and a date on the basis of owner authorisation. None yet carries a page-by-page editorial review. Each page shows its basis at the foot of the page, generated from the register so it always matches.
Extra care for money, health and legal topics
If a mistake could cost someone money, their health or their legal position, a qualified expert reviews the page or it does not go live.
Pages about marketing for healthcare, legal, finance and insurance businesses sit close to regulated advice, where a marketing claim can easily be read as advice about the service itself.
So we follow two rules. We keep marketing advice clearly apart from professional advice, in the main text rather than small print. And when a page describes what a regulator, professional body or platform requires, we cite that body directly, not someone else’s summary.
If we cannot get a qualified review for a claim in these areas, we remove the claim. A shorter page that is correct is better.
We apply this test to each page, not to the client’s industry. A page selling marketing to a dental practice or a mortgage broker is written for a business owner, not as advice about health or money. Those pages carry a compliance section on what the regulator requires.
Roles
Who checks what on each page
Brief owner
Decides whether the page should exist, then sets out the reader, the proof it needs and the claims it must not make before anyone writes. Skipping this causes more bad pages than weak writing does.
Subject reviewer
Checks the page is accurate for its subject, from how a platform works to how customers in a trade buy.
Search reviewer
Checks the page answers the search it targets without competing with our other pages, plus the technical basics: canonical tag, metadata, headings, internal links and matching schema.
Conversion reviewer
Checks the page is clear about who it is for and what happens next. A page that asks for nothing has failed, and so has one that asks on every screen.
Approver
Puts his name against the page and records the basis: an editorial review of the page, or authorisation to publish a group of pages. Either way, the approver answers for the page afterwards.
Scoring
The twelve things we score every page on
Each is scored from zero to five before a page goes live. A priority page needs an average of four or more, and no page goes live with a critical score below three, so a high average cannot hide a page that is wrong.
- Intent fit: does it answer what the person searched for?
- Reader focus: could this page have been written for any other audience?
- Originality: does it say anything the pages already ranking do not?
- Expertise: does it show hands-on knowledge, or just summarise others?
- Clarity: can a busy reader find the answer quickly?
- Persuasion: does it make a case, or just list features?
- Accuracy: has every claim that can be checked been checked?
- Coverage: does it answer the whole question, including the awkward parts?
- Internal links: does it point to the pages this reader will want next?
- Visual quality: has the page’s design been thought about?
- Next step: is the call to action right for this reader?
- Technical SEO: canonical, metadata, headings, status codes and structured data.
Before a page is indexed
Four things that must all be true before a page can be indexed
These do not trade off against each other. If a page fails any one, it stays live, readable and linked, but out of search.
- An average of four or more
- Across the twelve checks, on a priority page. A high score in one area cannot make up for a low one in another.
- A minimum score of three
- This minimum stops a high average from carrying a page that is wrong.
- A name and a date
- Kept in a register the page cannot edit. No entry means no search listing, with no override and no expiry.
- Still accurate today
- If a page stops being true, we withdraw it. We do not leave it collecting traffic under our name.
What stops a page
What keeps a page out of search, or takes it back out
The build checks the first four on every deploy. The last two are judgement calls about pages that are already live.
- No approval recorded
- The most common blocker, by design. Until the page is in the approval register, search engines are told not to index it and it is left out of every sitemap. There is no override and no expiry.
- Fewer than three sections, or no internal links
- Thin stub pages are what this system is built to stop. A page with too little to be useful, or no links onward, is held back rather than published thin.
- A figure with no source
- Any number without a source fails validation. The fix is a link to the primary source or removing the number, and usually we remove it.
- Too similar to another page on this site
- Every page is compared against every other page. High overlap usually means a nearby page was copied with a few words changed, and the check catches it before it goes live.
- A claim we can no longer defend
- Live pages come down for this: a platform changes, a source is withdrawn, or we find out we were wrong. Withdrawing a page takes a one-line change and applies on the next deploy.
- The page went out of date
- Easy to miss. A page that was accurate when approved but is now badly out of date is misinformation with our name on it.
Questions
Questions about our content review policy
- Who approves a page for publication?
Zubair Afzal, who publishes this site. Each approval is recorded with his name and the date against that page. A team, a job title or an automated check can never approve a page.
Approvals are kept in a register separate from the content, so a page cannot approve itself. Until its entry exists, a page stays out of search results and the sitemap, though you can still open and read it.
Each entry also records the basis: an editorial review of that page, or the owner authorising publication after the automated checks passed. We never record the second as the first, because that would claim reviews that never happened. Every page shows its basis at the foot of the page.
- What happens to a page that fails review?
It keeps its current status and stays out of search. Nothing is half-published, and no page goes live to be fixed later.
Most failures are small: a claim needs a source, two pages target the same search, or a section is thinner than the outline promised. The page goes back with each problem listed.
- How do you decide a published page needs updating?
Three things trigger an update: something in the world changed, the data behind the page changed, or the page is old enough that we would re-read it before defending it.
The first matters most. When a platform changes how it works, an accurate page can become misleading overnight, and only someone who follows the topic will catch it.
- What counts as YMYL content for a marketing company?
YMYL means "your money or your life": any page where acting on our writing could affect someone’s money, health or legal position. We apply the test to the page, not the client’s industry. A page selling marketing to a dental practice or a mortgage broker is read by a business owner, and is not health or financial advice.
Those pages carry a compliance section instead, citing the relevant regulator directly and keeping marketing advice apart from professional advice. A page giving advice a reader could act on to their own harm could not go live without review by a qualified expert, and our bulk publishing tool refuses to touch it. If we cannot get a claim properly reviewed, we remove the claim.
- Does an approved page ever get withdrawn?
Yes. We remove the page from the approval register and it drops out of search on the next deploy. Nothing is deleted and no links break.
We would rather withdraw a page quickly and bring it back after a rewrite than leave up a page we can no longer defend.
Last updated · Published by Zubair Afzal, responsible editor