An SEO audit is not a verdict. It is a dated snapshot. It answers three questions: what Google sees of your site today, what stops it seeing the rest, and in which order to fix things with the resources you actually have. Anything beyond that is a sales promise rather than a diagnosis.
I have been running these since 2012. I began with the long published audit guides, then with shorter checklists — I still point beginners towards the short ones, because they can be applied without experience. My own method has changed since: fewer boxes to tick, more ranking of what matters. An audit that lists a hundred anomalies without saying which ones count helps nobody.
What an audit examines, area by area
Six areas, always the same, in this order. The order matters: there is no point optimising text Google never crawls, and no point building links to a page held in noindex.
| Area | Question it answers | What gets examined |
|---|---|---|
| Indexing | Are your useful pages in the index? | Page indexing report, robots.txt, noindex tags, canonicals, sitemaps, orphan pages |
| Technical and speed | Is the site crawlable and fast? | HTTP status codes, redirect chains, JavaScript rendering, Core Web Vitals, mobile rendering, HTTPS |
| Content | Does it answer what your customers type? | Search intent, titles and meta descriptions, heading hierarchy, duplication, pages with no traffic |
| Links | Who cites you, and how do people move around inside? | Referring domains, anchors, internal links, click depth, broken links |
| Measurement | Are your numbers trustworthy? | Search Console and GA4 properties, tagging plan, conversions, duplicate measurement |
| Accessibility | Can every agent read the content? | Text alternatives, heading structure, form labels, contrast |
Indexing: the first question, and why the site: operator is not a measure
For years I counted indexed pages by typing site:yourdomain.com into Google. It is handy for a quick impression, but it is not a measurement. Google's own documentation is explicit: search operators are subject to indexing and retrieval limits, and the URL Inspection tool in Search Console is the reliable way to diagnose a specific page (Google Search Central, Search operators documentation, accessed 11 September 2026).
The real source is the Page indexing report in Search Console. It gives the status of every URL Google knows about and, more importantly, why a URL is not indexed: blocked by robots.txt, excluded by a noindex tag, duplicate without a user-selected canonical, soft 404, page with redirect (Search Console Help, Page indexing report, accessed 11 September 2026). Each label names a cause rather than a symptom, which saves a great deal of time. In most audits I run on small business sites, the bulk of the unindexed volume comes down to two or three structural causes, usually inherited from an old rebuild or a badly configured plugin. That is why the indexing question is settled before anything else is touched.
Technical checks and speed
The technical work starts with a full crawl. I have used Screaming Frog for years: the interface is austere, but it sees things no human will spot across several thousand URLs. The free version is limited to 500 URLs per crawl, and the licence is listed at £199 or $279 per year (vendor pricing page, accessed 11 September 2026). The best tool is the one you genuinely use.
On speed, I work from Core Web Vitals measured in the field rather than in the lab. The thresholds Google publishes are 2.5 seconds for LCP, 200 milliseconds for INP and 0.1 for CLS, to be met at the 75th percentile of page loads (web.dev, Web Vitals, accessed 11 September 2026). INP replaced FID as the stable responsiveness metric in 2024; if your last report still mentions FID, it is out of date. Where speed is the blocking issue, it leaves the audit and becomes its own piece of work: site speed optimisation.
Content: intent, structure and duplication
Take the vocabulary your customers actually type, and put it where it counts — the title, the section headings, the opening paragraphs. One correction I make often: length is not a criterion. A long text is not mechanically better placed; it ranks when it covers the intent more completely than a short one. On the sites I audit, the gain comes more often from merging pages that cannibalise each other than from adding a thousand words.
A content audit also picks up neglect at scale. The 2025 edition of the Web Almanac, which analyses the HTTP Archive crawl corpus, reports that at the median 15% of images on mobile pages carry an empty alt attribute (Web Almanac 2025, SEO chapter, accessed 11 September 2026). That matters on an online shop whose product photographs could be attracting image traffic. This strand continues naturally into ongoing SEO content work.
Links: count referring domains, not links
Popularity is measured in referring domains — the number of distinct sites citing you — not in raw link counts. Majestic, Ahrefs and the other suites all report it, and their numbers differ because each crawls the web with its own robot. I use them to place an order of magnitude against competitors, never as a score.
A position I used to defend — "more pages means more revenue" — is false as stated. A thousand weak pages dilute internal linking and consume crawl budget. What counts is the number of useful pages, connected to each other and reachable in few clicks from the homepage.
Measurement: without reliable data, an audit is only an opinion
An audit rests on your real data, so I check first that it exists and is clean: a correctly declared Search Console property (domain rather than URL prefix), GA4 without double tagging, conversions defined on events that match your actual objectives, and a dashboard somebody can read in Looker Studio or an equivalent. Without that, no recommendation will be assessable in six months, which is the whole point of measurement and traffic management.
Accessibility: a neighbouring subject, looked at in passing
Accessibility is not SEO, but the two overlap: text alternatives, a coherent heading hierarchy, form labels, contrast. Legal obligations vary by country, and an SEO audit is not a compliance review. I flag the visible problems anyway, because correcting them also improves how machines read the page.
What the audit produces: a dated, prioritised plan
The deliverable is not an 80-page report exported from a tool. It is a short document with a table of actions, each described with its expected effect, its level of effort, the person who carries it out — you, your agency, your host or me — and a deadline. Actions fall into three batches: what blocks indexing and is dealt with immediately, what improves the existing site over a quarter, and what belongs to a project that must be budgeted. A spoken debrief comes with it, along with the raw crawl files your technical supplier will want.
If you are commissioning this from outside, agree the boundaries first. Settling what the audit should cover beats comparing tool lists.
What an audit does not promise
It promises no position, no traffic volume and no guaranteed timescale: nobody controls Google's ranking or the calendar of its updates. It does not replace implementation — an audit that is never applied produces nothing. It does not cover your commercial strategy either: if your prices or your offer do not hold up against competitors, SEO will not compensate. And it has a shelf life. Beyond a year, or after a rebuild, it needs doing again.
Timescale and access to arrange
How long it takes depends on the size of the site and, above all, on how quickly I get the access. Allow a few days of analysis for a brochure site of a few dozen pages, and appreciably longer for a catalogue of several thousand URLs. To start, I need:
- read access to the Search Console property for your domain, or its creation if there is none;
- read access to GA4 or whichever analytics tool you use;
- an administrator or editor account on the CMS, or failing that a screen-sharing session;
- access to the hosting, to read server logs and configuration files;
- the names of the suppliers currently involved and the date of the last rebuild;
- your commercial objectives and your margins by product or service family, even roughly: they decide the priorities.
That last point is the one most often forgotten, and it changes everything. Two sites with identical technical faults will not get the same plan if one lives on online sales and the other on local quote requests.
Common questions
How long does an SEO audit take?
It depends on the number of URLs and how fast the access arrives. A brochure site of a few dozen pages can be analysed in a few days; a catalogue of several thousand URLs takes appreciably longer, because the crawl and the cross-checks expand with size.
What access do I need to provide?
At minimum, read access to Search Console and to your analytics tool, an administrator or editor account on the CMS, and access to the hosting for logs and configuration files. Sharing the date of the last rebuild, the suppliers involved and your commercial objectives also helps, since those set the priorities.
Does an SEO audit guarantee more traffic?
No. An audit identifies obstacles and an order in which to correct them, but no supplier controls Google's ranking or the timing of its updates. Results depend on the actions being implemented, on the competition for your queries, and on the quality of what you are selling.
Is the site: operator enough to know which pages are indexed?
No. Google states that search operators are subject to indexing and retrieval limits, and recommends the URL Inspection tool for diagnosis. The Page indexing report gives the real status of every known URL and the specific reason it is not indexed.
Do I have to buy SEO tools to get my site audited?
Not on your side: the supplier already owns them. A crawler such as Screaming Frog is limited to 500 URLs in its free version, enough for a small brochure site but not for a catalogue. Annual licences and rank-tracking suites are a cost that sits with the consultant.
How often should an audit be redone?
A full audit stays relevant for roughly a year on a stable site. Redo it sooner after a rebuild, a domain migration, a change of host or an unexplained drop in traffic, since each changes what search engines see.