Content SEO
Most of the available gains are in pages you have already published
Content SEO is the craft layer: matching a page to the job the searcher is doing, structuring it so an answer can be lifted out of it, linking it into the rest of the site, and deciding which existing pages to refresh, rewrite, merge or remove.
Where it starts
The same subject, asked in four shapes
One subject produces several distinct jobs, and each job wants a different kind of page. Getting the shape wrong is the most common reason good writing is never shown.
These are illustrative rather than a plan for any particular business. The point is the grouping: the words overlap heavily, the required answer does not.
Wants a definition
Wants the answer in the first two sentences, then the detail. An article that takes four hundred words to define the term is answering a different question.
- what is a heat pump
- heat pump meaning
- how does a heat pump work
- heat pump explained simply
Wants a comparison
Wants criteria side by side, with the trade-offs stated. This is table-shaped information, and prose is a worse container for it.
- heat pump vs gas boiler
- air source vs ground source heat pump
- best heat pump for a small house
- heat pump alternatives
Wants a procedure
Wants ordered steps with prerequisites and warnings. Readers here are mid-task and will leave immediately if the sequence is buried.
- how to size a heat pump
- heat pump installation steps
- preparing a house for a heat pump
- heat pump maintenance schedule
Wants a decision made
Wants a position, honestly argued, with the conditions under which the answer changes. Hedging everything is the failure mode here.
- is a heat pump worth it
- heat pump running costs
- do i need a new radiator for a heat pump
- heat pump grants eligibility
These are examples of how customers in this market search, drawn from keyword research and from the questions that come up on sales calls. They are illustrative, not a volume claim — the actual demand in your area is something we size before recommending anything.
The craft
Fit is decided before a word is written
Look at what is already being shown for a query and the shape of the required answer is usually obvious. If the whole page is comparison tables, the job is a comparison. If it is all short procedural pages, the job is a procedure. If it is dominated by forums, the job may be something no company page can do well, which is useful to know before commissioning one.
This is not about imitating whatever ranks. It is about recognising what the searcher has already told the search engine they want, repeatedly, through their behaviour. You can beat the existing results by being clearer, more current, more specific or more honest. You rarely beat them by answering a different question well.
The second half of the craft is structure. A page that answers a question near the top, uses headings that match how people phrase things, and puts list-shaped information in lists is easier for a person to scan and easier for a machine to lift from. Those are the same property, which is why writing well for a hurried reader tends to solve both problems at once.
Decisions
Every page in the library gets one of five verdicts
| Dimension | When it applies | What changes | What to expect |
|---|---|---|---|
| Refresh | The page fits the intent and earns something, but facts, examples or screenshots have aged. | Update the specifics, tighten the opening, add what has changed, keep the address and the structure. | The cheapest option, and the one that most often justifies itself. |
| Rewrite | The subject is right but the page answers a different question than the one being asked. | Rebuild the page around the actual job, usually changing the format entirely. Same address, new content. | Substantial work. Worth it where the page already has links or history behind it. |
| Consolidate | Several pages cover overlapping ground and none of them is the obvious answer. | Merge the useful material into one page, redirect the others to it, rewrite the internal links pointing at them. | Frequently the highest-value action available, and the one clients most resist. |
| Prune | No traffic, no links, no conversions, no strategic role, and nothing worth merging elsewhere. | Remove it, and return the correct status code rather than redirecting it somewhere irrelevant. | A small, unglamorous gain in how coherent the rest of the library looks. |
| Leave alone | The page works, fits its intent and has no evident problem. | Nothing, beyond adding it to the maintenance calendar for a scheduled review. | Underused. Not every page needs touching, and touching one that works carries its own risk. |
Connective tissue
Internal linking, done as writing rather than plumbing
Links are editorial decisions
A link belongs in a sentence where a reader would genuinely want the other page. Placed that way it says something about how the two subjects relate. Generated in a box at the foot of the page, it says nothing at all.
Anchor text is a description
The same anchor used four hundred times across a site tells nobody anything, and reads like software wrote it. Anchors should describe what is on the other end in the words that suit the sentence they sit in.
Direction carries meaning
Supporting pages should point at the page you want to be the definitive one, not merely at each other. When a cluster all links sideways and nothing points upward, no page is nominated as the answer.
Orphans are invisible
A page reachable only from a sitemap is barely published. Part of any library audit is finding pages nothing links to and either connecting them properly or accepting that nobody needed them.
How it runs
Working through a library that already exists
Inventory everything
Every published page with its traffic, queries, conversions, inbound links, publication date, last substantive edit and current format. Most clients have never seen their library laid out this way.
You get: A complete content inventory
Classify by job
What is each page trying to do, and what are the results for its target queries actually made of? This is where format mismatches become visible, and where near-duplicates cluster together.
You get: Intent and format classification per page
Issue the verdicts
Refresh, rewrite, consolidate, prune or leave alone, applied to every page, with the reasoning recorded so it can be argued with rather than accepted on faith.
You get: A decision log covering the whole library
Execute in priority order
Consolidations first, because they change the shape of the site. Then rewrites of commercially important pages, then refreshes, then pruning. Redirects and internal links are updated as part of each change, not afterwards.
You get: Published changes with redirects and links handled
Rebuild the internal linking
Once the surviving pages are known, the links between them are written deliberately: which page is definitive on each subject, and which pages support it.
You get: A linking map applied across the library
Put maintenance on a calendar
A library decays because nobody owns it after launch. Each surviving page gets a review interval and a named owner, so the audit does not have to be repeated from scratch in two years.
You get: A maintenance schedule your team can run
Questions
What people ask about their existing content
Should we update old posts or write new ones?
Usually update, and the reason is arithmetic. An existing page has history, links and a known set of queries it is already being shown for. A new page starts with none of that.
New pages earn their place when there is a genuine gap in the coverage — an intent nothing you own currently serves. If something you already published serves that intent badly, improving it is nearly always the better use of the budget.
Is deleting content really a good idea?
Deleting is one of four sensible options, alongside refreshing, rewriting and merging, and it is the right one less often than people assume.
It applies to pages with no traffic, no links, no conversions and no strategic reason to exist — event announcements from 2019, duplicate service pages, thin tag archives. Where a page has any value at all, merging it into a stronger page is usually better than removing it outright.
How does this differ from content strategy?
Strategy decides what subjects you should own, who you are writing for and why anyone should read you rather than somebody else. This page is about the execution and the maintenance of that decision.
They are separate jobs and they are frequently bought separately. If you have no strategy yet, start there — the craft work is more valuable once the direction is settled.
What does structuring for extraction mean?
It means the page is organised so a specific answer can be lifted out of it cleanly: a direct answer near the top, headings that read as the questions people actually ask, lists and tables where the information is genuinely list-shaped, and definitions stated plainly rather than implied.
It is not a trick and it does not involve markup that contradicts what is on the page. It mostly overlaps with writing clearly for a reader who is scanning.
How many internal links should a page have?
There is no correct number, and any answer that gives you one is guessing. What matters is that the links are editorial — placed because a reader at that point in the sentence would genuinely benefit from the other page.
The pattern worth avoiding is automated related-content blocks doing all the linking. They tend to point everywhere, use the same anchor text repeatedly, and tell nobody anything about how your subjects relate to each other.
Find out what your library is actually worth
Give us access to your analytics and Search Console and we will show you how much of the site earns nothing, where the near-duplicates are, and which twenty pages would repay attention first.
Related
Where to go next
- deciding what to publish in the first placeStrategy chooses the subjects; this page is about executing and maintaining them.
- the intent map underneath itOne dominant intent, one canonical page — decided before writing starts.
- being quoted by answer enginesExtractable structure matters in both places, for related reasons.
- writing the pages themselves
- making sure the pages can be crawled
- realistic timelines for content work
Last updated · Reviewed by Zubair Afzal