How to Audit On Page SEO for an Existing Website
The list of elements is well known and freely available everywhere. Knowing which ones matter for a given page, plus which are not worth touching at all, is the part almost nobody supplies.
The List Is Not The Hard Part
Titles, headings, content, images, internal links. Anybody can name the elements, as any number of published guides do. What none of them tells you is which of those matters for the page in front of you.
Why lists are unhelpful. Everything is equal weight.
A list presents twenty items as twenty tasks. In practice two of them account for most of what could change and several are worth ignoring entirely, so the list conceals that.
What ordering requires. Knowing the page.
Which search it targets, whether anybody reaches it, whether it earns anything. The same finding matters enormously on one page and not at all on another.
What this page provides. An order, with reasons.
What tends to move a page most, what rarely does, then how to decide between them on your own site. The reasons matter more than the order, because they let you apply it to cases we have not described.
What it does not provide. A form to complete.
No checklist appears here, for reasons block six sets out. Handing somebody a list of ticks produces a list of items nobody can act on.
Who this page is for. Somebody holding a report.
A business given a long list of findings needs to know which of them to act on. That is a harder question than producing the list was. It is also the question the list itself never answers.
Where the technical side sits. Elsewhere.
Whether a page can be reached and read at all is a separate review covered in our technical material. This page assumes the page works and asks whether it says the right things.
Biggest Impact
Six things, in rough order of effect. The order is the point, because most on-page effort goes to the bottom half of this list while the top half goes unexamined.
First, matching the page to the right search. The largest by far.
Whether the page serves what the search actually wants. A page aimed at the wrong intent cannot be improved into working, which search intent explains.
Second, one subject per page. Structural.
Whether the page is about one thing, plus whether another page competes with it. Cannibalisation covers the second half.
Third, a title that describes it accurately. Cheap and effective.
Whether the title states what the page contains. Minutes of work and the most common single fault we find on pages that are otherwise fine.
Fourth, structure that holds up. Readable from headings alone.
Whether somebody scanning can follow the argument. Frequently a symptom of a page with no argument rather than a heading problem.
Fifth, genuinely covering what the reader needs. Comparison work.
What the current results cover that your page does not. Optimising existing content covers the method.
Sixth, internal links pointing at it. Underrated and controllable.
Whether related pages on your site link to it. Entirely within your control and almost always neglected in favour of editing the page itself.
Why that order. Reach against refinement.
The first two determine whether the page can compete at all. The last four improve a page that already can. Working upward from the bottom is why most audits change nothing.
What Is Rarely Worth Touching
Four things routinely audited and routinely charged for that we would not recommend paying to change. Naming them costs us billable work, which is the reason this block is worth reading.
The proportion of a page a phrase occupies. No target exists.
Nothing has ever published one, so there is nothing to audit against. The keywords pillar traces where the figures in circulation come from.
Exact heading counts. Nothing to count towards.
Whether headings describe the page matters. How many of each level exist is a number with no meaning attached. Reports flag it because it is easy to measure.
Image filenames on an existing library. Real cost, marginal gain.
Renaming files on a live site changes where they live, so every reference needs updating. The image pillar covers when it is worth doing, which is during a rebuild.
Descriptions on pages nobody reaches. Effort with no possible return.
Writing snippet summaries for pages that never appear in results cannot produce anything, because nobody ever sees them. Do the pages that already appear.
What they have in common. They are countable.
Each produces a number that can be reported before and after. That makes them appealing deliverables regardless of whether changing them achieves anything.
Why saying so matters. It is most of what gets sold.
A great deal of on-page work invoiced across this industry consists of these four. A business recognising that can redirect the same budget at the top of block two.
Prioritising Fixes
Impact against effort, applied to the pages that matter. A change to a page nobody reaches achieves nothing however correct it is. This is where most audit output is wasted.
The first filter. Does this page matter.
Whether it earns enquiries, brings visitors or serves a commercially important search. Findings on pages failing all three are correct and irrelevant.
The second filter. Impact against effort.
What each change would plausibly achieve, weighed against what it costs to make. A serious problem taking an hour outranks a moderate one taking a fortnight.
Why the order of filters matters. It removes most of the list.
Applying the page filter first discards findings on pages nobody visits, which is usually the bulk of an audit. Prioritising within what remains is then a manageable exercise.
What gets discarded deliberately. Correct findings.
Plenty of accurate observations will never be worth acting on. Recording that decision matters, because otherwise the same items resurface in the next audit.
What a prioritised output looks like. Three lists.
Do first, do eventually, ignore. The third is the one that saves money and the one most reports omit.
Start With The Pages That Already Rank
Pages appearing just below where they would earn traffic are the cheapest wins available in on-page work. Everything has been established except the last stretch.
Why they are cheap. Most of the distance is covered.
The search exists, your site is considered a plausible source and the page is being shown. None of that has to be built, which is what a new page faces.
How to find them. Your own search reporting.
Which searches show your pages, plus where. Anything appearing on the edge of visibility is a candidate. Our measuring material covers reading that properly.
What to do with them. Block two, from the top.
Check intent first, then whether anything competes, then the title. On a page already appearing, one of those three is usually the constraint.
Why businesses skip this. It is invisible work.
Improving something published feels like standing still next to writing something new. The arithmetic favours the invisible work considerably.
The related shortcut. Pages that used to earn.
Anything that performed and stopped is worth examining, because something changed. That is a shorter investigation than a general audit.
The Checklist
A checklist is a memory aid for somebody who already knows what they are looking at. Handed to somebody who does not, it produces a list of unactioned items and a sense of failure.
What it does for an expert. Prevents omission.
Somebody who could do this from memory still benefits from a reminder, because the risk is forgetting rather than not knowing. That is what checklists are for in every field.
What it does for everybody else. Produces anxiety.
A list of items you can neither assess nor prioritise. Each reads as a possible problem and none can be resolved, which is worse than not having looked.
Why they circulate anyway. They cost nothing to publish.
A checklist looks generous and takes an afternoon to produce. It transfers the difficult part, which is judgement, to somebody who does not have it.
Why we do not publish one. It would contradict this block.
Supplying a runnable list here would hand you exactly the artefact this paragraph describes as unhelpful. Our technical material holds the same position for the same reason.
What to use instead. The questions.
Each guide in this section ends with questions a business owner can answer without technical knowledge. Those are usable in a way a checklist is not.
Auditing A New Page
Reviewing a page before it goes live is quicker than reviewing one afterwards. It also prevents most of what a retrospective audit finds. Almost nobody does it.
Why it is shorter. Nothing has accumulated.
A new page has no history, no stale content and no broken links. The review is a handful of questions rather than an investigation.
What to check. Four things.
Which search this page owns, whether anything else on the site already owns it, whether the title describes it, then what will link to it.
The link question specifically. Before publishing.
Naming which existing pages will point at the new one, then adding those links as part of publishing rather than later. This is the step that prevents orphaned pages.
Why the timing matters. Everything is cheap now.
Changing a title before publication costs nothing. Changing it afterwards means changing something already established, which is a different decision.
Who should do it. Whoever publishes.
This is a habit rather than a service, belonging with the person adding the page. Four questions at the point of publishing removes work that would otherwise be bought back later as an audit.
What it prevents. Most of the audit.
The bulk of what a retrospective review finds is decisions never made at publication. Making them at the right moment removes the finding entirely.
Reporting Is Not Fixing
An audit delivered and never acted on has achieved nothing. That position is argued in full in our technical material and held in our measuring material. It applies identically to on-page work.
The short version. A document is not an outcome.
Findings tell you what to change. The site is exactly as it was until somebody changes it. The two are separate purchases.
Why we are not arguing it here. It is held elsewhere.
Repeating a position across several pages weakens it rather than reinforcing it. Our technical material owns this argument and this page defers to it.
What it means for on-page work specifically. Most is implementable.
Unlike technical findings, most on-page changes can be made by whoever edits the site. That is an advantage worth noticing, because it removes the developer step.
What to establish before commissioning. Who edits.
Whether the audit includes making the changes or only identifying them. On on-page work the answer is frequently that you could do it yourself.
The question to ask. What do we do on Monday.
Any review that cannot answer that in one sentence has not been prioritised, whatever else it contains.
When The Problem Is Not On-Page
A page can be perfectly optimised and still not rank. Recognising that saves money. It is the reason this section does not present on-page work as sufficient by itself.
The site has no standing. The commonest reason.
A well built page on a site nothing links to and nobody references will lose to a worse page on an established one. No amount of on-page work closes that gap.
The search is unwinnable. Sometimes obvious.
Where every result is a national brand, a marketplace or a directory, a small business is not going to displace them. Better established before writing than after.
The content genuinely is not the best answer. Uncomfortable and common.
Sometimes what already ranks is simply better, produced by somebody with more to say. Improving your page may not be enough. Knowing that is worth the assessment.
How to tell which. Read the competition.
If the results are all substantially better resourced, the problem is standing. If they are comparable and yours is thinner, the problem is the content.
What to do instead. Choose a different search.
Specific searches nobody has written for are winnable where broad ones are not. That reframing is the most valuable outcome of an audit that concludes the page cannot win.
Why we say this. It prevents wasted spend.
A business repeatedly refining a page against an unwinnable search is spending money on the wrong problem. Our advanced material covers building the standing that changes it. The full series is on the on page SEO guide.
Which pages does
this finding even
apply to?
Filter by page before filtering by severity. A change to a page nobody reaches achieves nothing however correct it is, so applying that filter first usually discards the bulk of an audit before anybody schedules a single task.
In order of effect:
Density, heading counts and filename renaming are not on the list, deliberately.
Every guide.
One practice.
What a page says about itself, headings, keywords, search intent, images, page types, user behaviour, credibility and how to review an existing site.
