SEO and content strategy
The work that happens before anyone writes a word, and the technical work that decides whether the page gets indexed once they have. This is the half most content suppliers leave to you.
Before anyone writes
Four decisions, in order. Get them wrong and the writing budget goes into pages that were never going to rank, which is the single most expensive mistake in this vertical.
-
Keyword research and clustering
We parse your Ahrefs or Semrush export, classify the terms, and group them into clusters. One cluster becomes one page. A keyword list that has not been clustered is a list of pages you are about to build too many of.
-
A SERP check on anything that matters
What actually ranks for a term beats what the term looks like it should want. Localised head words are the trap here. The bare word for roulette is a wheel-picker tool in most markets, and in several languages "table games" returns board games. We check before planning a page around a term, not after it fails.
-
A topical map
Hubs, sub-clusters, supporting pages and URL paths, with an owning page per cluster. Delivered as the spreadsheet your team or ours builds from, in the format you already use rather than a new one.
-
Content briefs
Heading structure, a word count per section, which keyword each section owns, the internal links to place, and the questions the page has to answer. A brief this specific is what makes two different writers produce the same page.
On-page and technical
Where a task has one correct answer, we compute it rather than estimating it.
| Work | How it is actually done |
|---|---|
| Titles and metas | Written, then measured in rendered SERP pixel width against the real truncation limit. A page fails the build if its title is over. Every title and meta is checked for duplicates across the whole site. |
| H1s and slugs | The H1 carries the exact target keyword. Slugs are short, lowercase and stable, and we flag any change that would need a redirect rather than making it quietly. |
| Internal linking | Anchor text uses the target page's own keyword, not a natural-sounding phrase that happened to sit nearby. Links are woven into sentences, never dumped as a list at the bottom. Anchor diversity is checked across the site, not per page. |
| Schema markup | JSON-LD matched to what is on the page. Article, review, FAQ, breadcrumb, organisation, product, collection. Validated before delivery, never written from a template and left unchecked. |
| Technical audit | A 40-point checklist as a four-sheet workbook, built from your Screaming Frog crawl and Search Console export. Items that cannot be judged from the data are marked as such with the reason, because a missing row reads as a forgotten row. |
| Fixes, not descriptions | Where the fix is mechanical you get the file. The corrected robots.txt. The hreflang tag set for your URLs. The canonical values. Not a paragraph explaining what a correct one would contain. |
| Content refresh | Which pages are decaying and what specifically changed in the SERP, so the rewrite budget goes to the pages that will move rather than across everything published before a given date. |
| AI search visibility | Whether AI answer engines can reach your pages, whether your brand facts contradict each other across the site, and which sources currently get cited for your terms. Contradictory facts are the thing that stops an engine citing a site confidently. |
No CMS access is the normal case
Most clients cannot give a contractor admin rights, and most of the time they should not have to. It changes the shape of the deliverable rather than the quality of it.
Anything you cannot hand us edit rights to comes back as a find-and-replace instruction: the exact original sentence so you can locate it, then the replacement. That applies to on-page fixes, internal link insertions and content edits alike. Someone on your side pastes it in. Nobody has to interpret a recommendation.
The same goes for spreadsheets. If you send a tracking sheet, you get your own sheet back with columns added, in your own format, with any columns we did not recognise left exactly where they were. Reformatting it into something tidier only moves the reconciliation work onto you.
Send your Screaming Frog crawl and your Search Console export if you have either, even if you think they are out of date. Working from the better data costs nothing, and working around its absence costs a great deal.
Common questions
We already have an SEO team. What would you actually add?
Usually the content layer they do not have time for. In-house teams tend to be strong on technical and weak on the volume of briefs and pages a market launch needs. We take the research and brief work off them, or we take the writing and work to their plan.
If your team is already doing all of it well, you do not need us for this.
Can you work without Search Console or analytics access?
Yes, and most of the time we do. We work from Ahrefs or Semrush exports, Screaming Frog crawls, live SERP checks and whatever you can share. If a recommendation genuinely needs data we do not have, we say what is missing and what it would make possible rather than estimating and presenting the estimate as a finding.
Why measure titles in pixels instead of counting characters?
Because Google truncates on width, not character count. A title of 60 narrow characters fits where 55 wide ones do not, so a character limit either wastes space or gets you cut off. We measure the actual rendered width against the limit and fail the page if it is over.
What does the technical audit cover?
A 40-point checklist delivered as a four-sheet workbook: issues, crawl data, index coverage and sitemap. It parses your Screaming Frog and Search Console exports rather than asking you to interpret them.
Where a fix is mechanical, you get the corrected file. The audit generates the actual robots.txt or the hreflang tag set for your URLs, not a paragraph describing what a correct one would look like.
How do you stop our pages competing with each other?
A page ledger, one row per page, one topic per page, and no topic owned twice. It is checked before any new page is planned. For multi-brand clients it runs across brands too, because two sites in the same group targeting the same term is the same problem wearing a different logo.