What a real SEO audit looks like, and what the scary ones are.

How the scary one is made

A crawler is pointed at your site. It visits every page it can find and checks each one against a list of a few hundred rules — is there a meta description, is it under 160 characters, are the images tagged, is there an H1, is the H1 unique, does the page load fast, is the link text descriptive. Every rule it fails becomes an "error." The errors are counted, the count is turned into a grade, and the grade is coloured red.

Then a single PageSpeed run is screenshotted — the worst one available, if several were tried — and placed next to a sentence about how many customers a slow site loses. The number in that sentence comes from a study about e-commerce checkout pages, or from nowhere.

None of this is fabricated. The crawler really did find 340 things. The trouble is what the report does not contain: which of those 340 things Google actually weighs, which pages on your site anyone searches for, and what any of it costs you in a search you are losing. A report that counts findings without weighting them is not an audit. It is an inventory, dressed up.

Why it looks alarming

• Rules are counted, not weighted. A missing alt tag on a decorative image and a homepage that takes twelve seconds to load are both "one error." One of them matters.
• There is no baseline. A score of 41 sounds bad. Against what? Every site fails dozens of crawler rules, including the ones ranking first.
• The speed number is one run. PageSpeed scores move from run to run with nothing changed; that variance is documented here. A single screenshot of a single run is the least honest measurement the tool can produce, and it is the one scare audits use.
• The customer-loss figure is borrowed. "A one-second delay costs 7% of conversions" is a real sentence from a real study, about retail. It has nothing to do with a pet owner searching for an emergency vet, and the person who put it in the PDF knows that.
• It is free because it is automated. The report cost nothing to make, which is why it can be sent to two hundred practices a week. The sale is the retainer that follows.

What a real audit starts with

Not the crawler. The search data. Google gives every site owner a free tool, Search Console, that shows exactly which searches your pages appeared for, how often, where they ranked, and whether anyone clicked. That is the ground truth, and an audit that does not begin there is guessing.

From that data, the questions a real audit answers are specific:

• Which of your pages are earning impressions, and for what? Most practice sites have two or three pages doing all the work and twenty doing nothing. The audit says which.
• Where are you appearing but not being clicked? A page with many impressions and no clicks is a page ranking too low, or a page that ranks but promises the wrong thing. Those are different problems with different fixes.
• Does each page answer the search it appears for? A page that ranks for "emergency vet near me" and opens with a paragraph about the practice's founding year is losing the click to whoever put the phone number first.
• How slow is it, in seconds, across several runs? Largest Contentful Paint against Google's 2.5-second threshold, tested more than once, mobile first.
• Can Google read the things that matter? Hours, location, services, and service area as structured data, not just as text on a page.
• What does the Business Profile look like? For a local practice, the map result often matters more than the organic one, and the audit says whether it is claimed, complete, and correct.

And then the part that makes it useful: a short list, ranked by what it would change. Three things, not three hundred. The fix that moves the most searches first.

What one looks like in practice

This site's own SEO page is the example, because the audit was run on it this month and the result was not flattering. Search data showed the page was appearing for over two thousand searches in three months — the biggest source of impressions on the whole site — at an average position around 84, page eight or nine, with zero clicks. A crawler report would have listed its missing alt tags. The actual audit found the page was technically clean: indexed, fast, well linked, longer than the pages outranking it. What it found instead was that the page argued a position while the searches wanted an explanation. The title stated a thesis; the search results were all guides. The phrasing people actually typed did not appear on the page once.

The fix was a rewrite, not a retainer. The new version is at veterinary SEO, and the audit's receipts — what was measured, what was ruled out, and why — are the reason the rewrite took the shape it did. Whether it works is a question for the search data in a few months, which is where the next audit will start.

How to read one you have been sent

Three questions, asked of whoever sent it.

Which of these findings does Google actually rank on? If they cannot separate the twelve that matter from the three hundred that do not, they are reading you the crawler's output.

Show me the search this is costing me. A real finding names a query, an impression count, and a position. "Your site has 47 errors" is not a finding. "You appear for 'cat vet [your town]' 300 times a month at position 22" is.

Run the speed test again, now, while I watch. If the number moves ten points, they will have to explain why, and the explanation is the one this site already wrote.

What the free audit here is

One page. Your search data if you grant read access, or your public page data if you do not; your speed in seconds across several runs; whether Google can read your hours and location; and the three highest-leverage fixes, ranked, with what each one would change. Delivered in 24 hours, no call required. It is short because a useful audit is short, and it is free because it is the honest way to start a conversation about the one thing I do sell, which is the ads.