How to run an SEO audit on your own site in an afternoon

A commissioned SEO audit typically takes two weeks and arrives as a document of a hundred-odd pages, most of which is tool output with severity labels attached. Some are excellent. Many are a crawler's export with commentary.

The findings that actually change what you do are reachable in an afternoon, by you, with free tools. This article is that afternoon, ordered deliberately: the checks that can change your whole strategy come first, and the ones that produce tidy-up tasks come last. If you run out of time, you will have done the part that mattered.

Set aside four hours. Open a blank document. Write findings into it as you go — an audit you do not write down is a browsing session.

Before you start

Two things need to exist.

Google Search Console, verified for your site. This is non-negotiable. It is the only source of what Google actually thinks about your pages, and without it half the checks below become guesswork.

A list of what you are trying to rank for. Not a keyword tool export. The three to ten things a customer would type when they want what you sell. If you cannot write that list, that is your first finding and the audit stops until you can.

Hour one: is the strategy right?

This hour is skipped in most audits, which is why most audits produce fixes rather than decisions.

Check 1: does the top of your target results page look like your site?

For each target query, search it and look at what is ranking — not the individual pages, the *type* of thing.

Write down, for each query, what the top five are: national comparison sites, marketplaces, local businesses, forums, big publishers, video, official sources.

Then answer honestly: is your site the same kind of thing?

If the query returns ten aggregator sites and you are a single business, you are not in a difficult race, you are in a different race. That finding is worth more than every technical fix below, because it redirects effort that would otherwise be wasted for a year. The method in detail is in judging whether you can realistically win a keyword.

Check 2: does each target query have a page?

Make a two-column list: query, and the URL meant to answer it.

Two failures show up immediately. Queries with no page — you are hoping to rank for something you have not written about. Queries with three pages — you have written about it repeatedly and split your own signals, which is cannibalisation.

Check 3: what does Search Console say you rank for now?

Open Performance, set the range to the last three months, sort by impressions.

Read the top thirty queries and ask one question: are these the queries you wanted? Frequently they are not, and the gap between what you intended to be about and what Google thinks you are about is one of the most useful findings available.

Then apply the near-miss filter: queries at average position 5 to 20. You are already visible there and not being chosen. This is the shortest path to more traffic on the whole site, and the highest-priority list you will produce today.

Hour two: can Google actually use your site?

The technical pass comes second deliberately. A perfectly crawlable site targeting the wrong queries is still a site with no traffic.

Check 4: the Pages report

Search Console → Pages. Compare indexed pages against how many pages you believe you have.

If a substantial share is missing, the report names the reason. The reasons and what they mean are set out in technical SEO for people who are not developers, along with the four failures that silently block a whole site.

Write the count down. "Indexed: 34 of 41" is a finding. "Indexing seems okay" is not.

Check 5: the four blockers

Mechanical, two minutes, occasionally catastrophic when one is present:

• View source on a key page and search for noindex
• Open yourdomain.com/robots.txt and look for Disallow: /
• Confirm www and non-www resolve to one canonical version
• Confirm http:// redirects to https://

Check 6: the phone test

Open your three most important pages on an actual phone. Note anything missing, unreadable, awkward to tap, or covered by a pop-up.

Google indexes the mobile version. What is missing there is missing.

Check 7: speed, once

Run the homepage and one key inner page through PageSpeed Insights. Read the field data at the top if it appears.

Record one thing: is the largest image on the page enormous? That is the finding that pays. Resist spending the rest of the afternoon on lab scores.

Hour three: are the pages any good?

The uncomfortable hour, and the one that decides most outcomes.

Check 8: read your five most important pages as a stranger

Not skim. Read.

For each, answer:

• What question does this page answer, in one sentence?
• Would someone searching that question be satisfied, or would they go back?
• Is anything here that they could not get from the four pages above you?
• Is anything stated that is untrue, out of date, or vague enough to be meaningless?

Most sites discover in this hour that their key pages are competent summaries of what everyone else says. That is the expensive finding, because the fix is writing rather than configuring.

Check 9: the title tags

List your titles and look for:

• Duplicates — two pages with one title usually means two pages that should be one
• Titles that are category labels rather than promises
• Titles where the distinguishing part sits at the end, after the brand
• Pages with impressions but almost no clicks

That last group is the highest-return list on the site: fifteen minutes per page, results within weeks. The method is in how to write a title tag that earns the click.

Check 10: intent match on the near-misses

Take the position 5–20 queries from Check 3. For each, search it and compare the *format* of the top results against the format of your page.

A page stuck at eleven often has the right subject and the wrong shape — a how-to guide where the results are all comparisons, or an essay where the results are all short direct answers. Reshaping is cheaper than rewriting and works faster. The diagnosis is in search intent.

Hour four: structure and links

Check 11: orphans

Which pages are linked from nothing but the navigation and the blog index?

If you cannot crawl your own site, approximate: take your ten most important pages and search your own site for links to each. Zero inbound internal links means the page is invisible in your structure even though it exists.

Check 12: where your internal links point

Count roughly where internal links go. If a large share point at the homepage, they are wasted — the homepage is already linked from every page.

Cross-reference the near-miss list from Check 3 against pages with few inbound internal links. That intersection is your immediate action list. The system for fixing it is in internal linking: a system, not a habit.

Check 13: who links to you

Search Console → Links → Top linking sites.

You are not counting. You are answering: is there anything here beyond directories and your own profiles? A site with no genuine external links has a ceiling, and it is worth knowing where that ceiling is before planning around it. What actually counts is in what makes a backlink high quality.

Check 14: local, if you are local

If customers come from a place, open Google Maps, search your main service plus your town, and see whether you appear in the map pack. If you do not, that finding outweighs everything else in this document, and it is a different discipline — see local SEO and how the local pack differs from organic results.

Turning findings into a plan

You now have twenty or thirty findings. The trap is working through them top to bottom.

Sort them into four groups instead.

Blockers. Anything preventing indexing or access. Fix today. Usually zero or one of these — occasionally the one that explains everything.

Near-misses. Position 5–20 pages needing a better title, a reshape, or two internal links. Best return per hour on the list, results in weeks.

Strategic findings. Wrong queries, wrong kind of competitor, key pages that are not actually better than what sits above them. These are quarters of work, not afternoons, and pretending otherwise is how sites stay flat for a year.

Hygiene. Alt text, minor speed items, tidying. Real, small, and last. Do not start here because it is comfortable.

What an audit cannot tell you

An audit describes your site. It does not describe the market you are trying to enter.

If Check 1 said your target queries are held by sites structurally stronger than yours, no configuration of your own site changes that this year. The honest options are narrower queries where you can genuinely be the best answer, patience measured in quarters, or paying for position outright rather than earning it — the trade set out in when renting attention beats owning it, with current prices on the rankings board.

Knowing which of those you are choosing is the actual point of the afternoon.

Common questions

How often should I do this? Twice a year, or after any change of platform, theme, host, or domain.

Do I need paid tools? Not for this. Paid crawlers make hours two and four faster on large sites; under a hundred pages they mostly add volume.

What if I find nothing wrong? Then the problem is competitive rather than fixable by tidying — itself a finding, and a far more useful one than a list of alt-text warnings.

Should I still hire someone? Possibly, and you will brief them far better having done this. The decision framework is in should you hire an SEO agency.