Skip to content
SEOCompanyby Webso Digital

Free resource

The SEO audit checklist, 30 checks you can run on your own site today

Every check below can be run with Google Search Console, PageSpeed Insights and your own eyes, all free, in about two hours on a small site. Each one says what to look at, how to look at it, and what a pass actually looks like, because “check your titles” is advice and “no two pages share a title” is a test. Work the list in order: if the first section fails, nothing below it can help you yet.

Written by Daniel Stoychev. This is the order we work in on a real account, minus the part where we argue about which finding matters most, which is the part you are paying for when you buy an audit.

Can Google see it at all

Everything else on this list is wasted if the answer here is no. A page that is not indexed cannot rank, cannot be cited by an AI assistant and cannot be fixed by better content. Start here and stop here if it fails.

  1. The pages you care about are actually indexed

    How: Search Console, Pages report. Or search site:yourdomain.co.uk and count.

    Pass: The home page, every service page and your main location pages are all in the indexed list.

  2. Nothing important is blocked in robots.txt

    How: Open yourdomain.co.uk/robots.txt and read it.

    Pass: No Disallow line covers a page you want found. Watch for a staging rule left in place at launch.

  3. No accidental noindex

    How: View source, search for “noindex”.

    Pass: It appears on nothing you want ranked. This is the single most common cause of a new site that never appears.

  4. A sitemap exists, is current and is submitted

    How: yourdomain.co.uk/sitemap.xml, then Search Console, Sitemaps.

    Pass: It lists live pages only, no redirects, no 404s, and Search Console reports it as read.

  5. Each sitemap URL has its own lastmod date

    How: Open the sitemap and compare a few dates.

    Pass: Dates differ per page and reflect real edits. Every URL sharing one date is the same as having none: Google discounts it, and it is the only automatic recrawl signal you get.

  6. One version of the site, not four

    How: Try http, https, with www and without.

    Pass: All three of them 301 to the fourth, in one hop.

Does each page say what it is

Google has to work out what a page is for, and so does a person reading a result. Most sites fail here quietly: the pages are fine, the labels are missing or duplicated.

  1. Every page has its own title

    How: Crawl the site, or check your top ten pages by hand.

    Pass: No two titles are identical. Two pages with the same title are two pages Google has to choose between, and it often chooses neither.

  2. The title leads with the thing being searched for

    How: Read it as a search result, not as a heading.

    Pass: The keyword is near the front and the brand at the end, not the other way round.

  3. Every page has a meta description

    How: Same crawl.

    Pass: Present, under about 155 characters, and written as a reason to click rather than a summary.

  4. Exactly one H1, and it matches the page

    How: View source and count.

    Pass: One H1. Page builders often render the hero as a styled div, so a page can look like it has a heading and have none in the code.

  5. The canonical tag points at the page itself

    How: View source.

    Pass: Self-referencing, and in the raw HTML. If JavaScript rewrites it after load you have a problem that is invisible from the browser.

  6. Headings are in order and describe sections

    How: Read only the headings, top to bottom.

    Pass: The outline alone tells you what the page covers.

Is it fast enough on a phone

This is the one part of the list you can verify the same day, and the one people most often measure on the wrong device. Almost all of your visitors are on a phone on a middling connection.

  1. Largest contentful paint under 2.5 seconds on mobile

    How: PageSpeed Insights, mobile tab.

    Pass: Under 2.5s. Look at the field data at the top if it is there: that is real visitors, the lab score underneath is a simulation.

  2. Images are in a modern format and sized for the screen

    How: PageSpeed opportunities list.

    Pass: No “serve images in next-gen formats” or “properly size images” entries. A camera-sized file on a phone is usually the single biggest cost.

  3. Images have width and height set

    How: View source, check a few img tags.

    Pass: Both present, so nothing jumps as the page loads.

  4. Cumulative layout shift under 0.1

    How: PageSpeed, same run.

    Pass: Under 0.1. Above it, something is loading late and pushing the page around under the reader's thumb.

  5. No scripts that draw nothing

    How: PageSpeed, reduce unused JavaScript.

    Pass: Nothing large you cannot name the purpose of. Page-builder runtimes and abandoned tracking tags are the usual passengers.

  6. It works with JavaScript turned off

    How: Disable JavaScript and reload.

    Pass: The text is still there. Google renders JavaScript; no other major AI crawler does, so a client-rendered page can rank in Google and be blank to ChatGPT and Perplexity.

Does the content deserve the position

The part most audits skip because it cannot be automated. A technically perfect page that answers nothing will sit at position 50 forever, and no amount of the items above will move it.

  1. The page answers the search in the first paragraph

    How: Read the opening on its own.

    Pass: It answers the question without the rest of the page. This is also the part an AI assistant quotes.

  2. Location pages are genuinely different from each other

    How: Read two of them side by side with the place names covered.

    Pass: You can still tell them apart. If you cannot, Google treats them as one page and copies, and the copies get filtered out.

  3. The page type matches what the search rewards

    How: Search the term and look at what the top ten actually are.

    Pass: If the results are directories, listicles and map pins, a service page will not win it however well written. That is a page-type problem, not an optimisation problem.

  4. Prices, if you publish them, are visible and current

    How: Read the page.

    Pass: A figure on the page. Cost-intent searches want a number, and “packages available” is not one.

  5. No two pages target the same search

    How: Search Console, compare queries by page.

    Pass: One page per query. Several pages competing for one term is the most common self-inflicted ranking problem there is.

  6. The content is something only you could have written

    How: Ask whether a competitor could publish it unchanged.

    Pass: They could not, because it contains your jobs, your numbers and your view.

Local and trust

For anything with a service area, the map decides more of the outcome than the website does, and the profile is where that is won.

  1. The Google Business Profile is claimed and complete

    How: Search your business name.

    Pass: Claimed, with the right primary category, every service listed, and opening hours that are actually correct including Saturday.

  2. Name, address and phone match the site exactly

    How: Compare the profile against the site footer, character by character.

    Pass: Identical. “St” against “Street” counts as different.

  3. Reviews are arriving steadily, not in one burst

    How: Look at the dates.

    Pass: A steady trickle. Recency counts for more than most people expect.

  4. LocalBusiness schema is present and honest

    How: View source, or Google's Rich Results Test.

    Pass: Present, with the specific subtype where one fits, and every field true. Never mark up a rating you do not display.

  5. There is a real person on the site

    How: Open the about page.

    Pass: A name, a face and something checkable. This is the cheapest trust signal available and the most commonly skipped.

  6. The contact route works

    How: Send yourself a test enquiry.

    Pass: It arrives, and not in spam. Test this after every site change; forms break silently and nobody finds out for months.

When you have finished

The order to fix them in

Resist doing them in the order you found them. The first section is a gate: a page that is not indexed gains nothing from a faster image or a better title, so anything failing there goes first and alone. After that, the honest order is whatever affects the most pages for the least work, which is almost always the duplicated titles and the sitemap dates, not the content rewrite everybody wants to start with.

The content section is last on this page and first in value, and that is not a contradiction. It is last because it is the only part you cannot check mechanically, so it needs the other four out of the way before you can see it clearly. A page failing only the content test is a page with nothing wrong that nobody wants to read.

If you get through the list and the answer is that the site is fine and the searches you want do not exist, that is a real result and worth the two hours. Our guide on why a website is not ranking covers what that looks like, and SEO against Google Ads covers what to do instead.

A laptop showing search performance charts on a glass-top table
Search Console is the one free tool on this list worth more than every paid one, because it is the only record of what people searched before they saw you.

Questions about running your own audit

Can I really do an SEO audit myself?

Most of this list, yes, and you should: you know the business better than any auditor will after an hour. What you cannot easily do yourself is judge which of the 30 findings actually matters for your site, because that depends on your market and your competitors. Working the list and then arguing about the priorities is a better use of an audit than paying someone to tell you your titles are duplicated.

What do I need to run these checks?

Google Search Console, PageSpeed Insights and your own eyes. All free. Search Console is the one that takes a day to set up and is worth more than every paid tool combined, because it is the only source of what people actually searched before they saw you.

How long does it take?

A couple of hours for a small site if you are not fixing as you go. Write down findings, finish the list, then decide the order. Stopping to fix the first thing you find is how audits take a fortnight and miss the section that mattered.

What if something fails and I do not know how to fix it?

Note it and move on. The first section is the only one where a failure stops everything else being worth doing. If you want a second opinion on the order, the free check gives you one in writing within a working day.

Is this the same as your paid audit?

No. This is the checklist; the full seo audit is the written document that runs it against your site, with the Search Console data, the competitors above you and a prioritised fix list. The difference is the judgement and the data, not the list of things to look at, which is why the list is free.

How often should I run through it?

Twice a year on a stable site, and after anything structural: a redesign, a platform change, a migration. The items that break silently are indexing, the forms and the sitemap dates, and all three fail without a single visible symptom.

Or send it to us and skip the two hours

The free check runs this list against your site and replies in writing within one working day, with the three competitors above you and a quote from the public price list, from £399 a month. No call unless you ask for one.