Why do so many SEO audits disappoint?
Because most of them are generated, not performed. Crawling software made findings cheap: point a tool at a domain and it will happily produce four hundred “issues,” missing meta descriptions on paginated archives, alt text on decorative icons, a duplicate title on a page that shouldn’t be indexed anyway. What the tool cannot produce is judgement: which of those findings is costing money, which is cosmetic, and which is actively wrong for your situation.
The economics make it worse. An audit is usually a sales document, and page count photographs better than insight. So the deliverable swells, the prioritisation shrinks, and the buyer is handed a list only an expert could rank. That buyer commissioned the audit precisely because they lacked the expertise to rank the findings in the first place.
A useful audit inverts that. It treats findings as raw material and the ranking as the product: what to fix first, roughly what each fix costs, what it’s expected to return, and what to consciously ignore.
What are the four layers a real audit covers?
A complete audit examines technical health, content and intent, authority, and AI-search visibility, in that order, because each layer depends on the one before it. A site that can’t be crawled can’t rank; a site with nothing worth ranking gains little from links; and none of it matters to answer engines that can’t parse you.
Layer 1: technical health (can engines reach and render you?)
This layer asks whether anything else you do can work. It covers crawlability and indexation (what engines can reach, what they’re told to ignore, what’s accidentally blocked), redirect and canonical signals, site speed and Core Web Vitals, and structured data. Technical problems are the quietest ones (nothing looks broken to a visitor), which is why they survive for years. It’s also where fixes are most objective: a broken redirect chain is simply broken, and technical SEO is the discipline that repairs it.
Layer 2: content and intent (do your pages answer real questions?)
This layer compares what you’ve published against what your buyers actually search. It should surface coverage gaps (queries you have no page for), cannibalisation (two pages competing for the same query and splitting its equity), thin pages that answer nothing, and pages that hold the right answer but bury it. The output feeds a content strategy: a map of what to publish, merge, rewrite or retire, commercial queries first.
Layer 3: authority (does anything vouch for you?)
Authority is the layer engines use to break ties between competent pages. An honest audit reads your backlink profile three ways: what’s genuinely carrying weight, what’s dead weight, and what’s a risk. Risk means bought links, spun directories, the residue of an old agency’s shortcuts. It should also size the gap between you and the sites outranking you, because that gap sets the realistic timeline for link earning. Any audit promising to close it in a month is telling you something about itself.
Layer 4: AI-search visibility (can answer engines cite you?)
The newest layer, and the one most audits skip because most audit templates predate it. A growing share of research now ends inside ChatGPT, Perplexity and Google’s AI Overviews. Being citable there is a distinct, checkable property: are AI crawlers allowed in, is your entity story consistent everywhere machines read about you, and is each page structured so its answer can be lifted verbatim? This is answer engine optimisation. If your audit doesn’t mention it, you’re buying a report about the web as it used to be.
What are the red flags of a junk audit?
Most junk audits announce themselves in the first five pages, once you know the signs.
- The PDF dump. Ninety pages, of which eighty are screenshots of a crawler’s dashboard. Length is being used as a proxy for effort, and effort at the expense of clarity is a cost you’re paying, not a service you’re receiving.
- No prioritisation. Four hundred findings, all presented with equal weight. If the audit doesn’t say what to do first (and what to skip entirely), the ranking work has been handed back to you, and the ranking was the entire job.
- Findings that don’t survive contact with your site. Warnings about pages that are deliberately not indexed, “duplicate content” flags on legally required boilerplate, “missing” descriptions on utility pages. It means nobody senior read the export before sending it.
- The hostage audit. A teaser listing headline problems, with the detail held back for a sales call. An audit you can’t act on independently isn’t an audit; it’s a brochure with your logo on it.
- A single score. “Your SEO score: 61/100.” Composite scores compress incomparable things (a crawl block and a missing alt attribute) into one number engineered for alarm. Real audits price findings in impact terms, not points.
- One template for every business. If the audit for a five-page local plumber and a fifty-thousand-SKU retailer would read the same, neither site was actually audited.
How should you read a fix list?
Read it as an investment schedule, not a to-do list. Each item should carry three properties: expected impact, cost to fix, and who fixes it. Impact-over-effort sets the order; the owner column tells you whether the work lands with your developers, your writers or your agency. Anything with high effort and vague impact goes to the bottom, however technically “correct” the finding is. That ranking is also what turns an audit into a priced plan: the variables that move an SEO quote are the same ones the fix list exposes.
Expect sequencing logic, not just ranking. Some fixes unblock others: if crawlers can’t reach a section of the site, improving the pages inside it is wasted effort until the block lifts. A good list says so explicitly: “do nothing in part two until items one to three in part one have shipped”.
Expect a “consciously ignore” section too. A confident auditor tells you what not to fix: findings that are real but immaterial for your model. Its absence usually means the auditor was optimising for list length, or never understood your business well enough to know what’s immaterial. Finally, fixes that need engineering should be written in your engineering team’s language: a finding that pastes straight into a ticket for your development team is worth ten that need translating first.
What should an audit cost, and when is a free one legitimate?
Free audits are lead generation, and that isn’t a scandal. It’s a business model, and sometimes a good deal. The test is whether the free version is complete: full findings, full prioritisation, actionable without hiring the auditor. A free review that withholds detail for the sales call fails the hostage test above. A free review you could hand to a different agency entirely is the real thing: the auditor is betting that showing you the work wins the engagement honestly.
That’s the bet behind our own free SEO review: all four layers, a fix list ranked by revenue impact, delivered in 48 hrs, yours to keep either way. Paid audits earn their fee at scale (tens of thousands of URLs, migrations, penalty forensics) where the reviewing genuinely takes weeks. Between those poles, judge the deliverable rather than the price tag. Either way, a full SEO engagement should start from an audit, because a programme without a baseline can never prove what it changed.
