Why are so many SEO audits useless?
In 25 years in industry, I have seen the same pattern in every diagnostic trade: when the client cannot verify the measurement, some providers stop measuring. SEO is no exception. Many "audit reports" are automated exports from generic tools, dressed up with a logo, whose real purpose is to justify a quote. The problem is not automation, it is opacity.
The 4 elements of an audit that actually helps
- Measured, verifiable facts. "Your homepage contains 240 words of real content" is a fact: you can count it. "Your content lacks impact" is an unverifiable opinion. Every finding in a good audit must be checkable by a third party.
- One fix per finding. A problem without a fix is not an audit finding, it is a source of anxiety. The fix must be precise: which file, which tag, which target value.
- Clear prioritization. Thirty problems all at the same level means zero priorities. The audit must say what is blocking (fix this week), what is slowing you down (this month) and what is optimization (once the rest is done).
- A deliverable you can act on. The simple test: can you hand the report to your developer or webmaster and have them know what to do without calling you back? The gold standard: the fix files already generated, ready to upload.
The red flags of a decorative report
- A score with no breakdown. If nobody can explain how the grade is calculated, it measures nothing.
- Jargon with no translation. "Optimize your crawl budget" without explaining why or how is intimidation, not advice.
- Problems with no sense of scale. 4 images missing an alt attribute and a fully deindexed website do not deserve the same color badge.
- Fear as a sales argument. A serious audit also tells you what is working. A document that only lists catastrophes is trying to sell, not to inform.
How do you check an audit yourself, without being technical?
A good audit survives spot checks. Pick three findings at random from the report and verify them yourself, in your browser, with nothing to install. Here is how, depending on what the finding is about:
- A title tag or meta description reported as missing. Open the page, view the page source (Ctrl+U on a PC), and look for
<title>and<meta name="description">. What you read there settles it. - A number of indexed pages. Type
site:yoursite.cominto Google: you see what Google actually knows about your website. A large gap with the number of pages you have published is a genuine issue. - A robots.txt or sitemap problem. Open
yoursite.com/robots.txtandyoursite.com/sitemap.xml. Both files are public on every website in the world, including your competitors'. - A speed problem. PageSpeed Insights, Google's free tool, measures the three official metrics and shows the thresholds you have to stay under: main content displayed within 2.5 seconds, responsiveness under 200 milliseconds, visual stability under 0.1. An audit that says "your site is slow" without quoting those values has measured nothing.
- Structured data. Google's rich results test tells you what it really reads on a given page.
If a finding does not reproduce, it is not necessarily wrong: websites change, and a measurement taken three weeks earlier may simply be out of date. Just ask for the method and the date of the measurement. A serious provider answers in one sentence; a provider who dodges the question has just told you what you needed to know.
A score out of 100: useful or decorative?
Both exist. A score is decorative when it comes out of a black box: impossible to verify, impossible to track. It becomes useful under two conditions: every point must be traceable back to a measured fact, and the score must be re-measurable in exactly the same way so you can track your progress after fixes. That is exactly the principle we apply: our score breaks down into 160 checks, each with its measured value, and we publish our scoring method, including what the audit does not do. The public leaderboard takes the reasoning all the way: the websites that opt in display their score and the date it was last measured, which makes the grade verifiable by anyone.
How often should you run a new audit?
Not on a calendar, on events. A yearly audit booked out of principle always arrives after the damage; an audit triggered by a change arrives at the right time. Four triggers are worth it:
- After a fix, to confirm it actually reached production and not just the developer's machine. Give Google a few days to crawl the pages again before drawing conclusions about indexing.
- After a redesign or a domain change. This is the most dangerous moment in a website's life: one forgotten redirect, or a
noindextag left over from the staging site, wipes out years of search visibility overnight with nothing showing on screen. - After changing host, theme or plugin. These operations rewrite files nobody ever reads again, starting with robots.txt.
- Before signing with an agency, so you hold a dated baseline. That is your point of comparison six months later, and it costs nothing.
One rule makes the comparison meaningful: keep the same tool from one measurement to the next. Two tools do not weight the same criteria, so going from 62 with one to 78 with the other would prove no progress at all. What matters is not the level of the score, it is the gap with your previous measurement, and the list of items that turned green in between.
Free audit or paid audit: which one to choose?
A serious free audit gives you the diagnostic: the score, the categories in trouble, the main problems. That is enough to know where you stand. A paid audit is worth it when it adds the executable part: the detail of every check, the exact fixes, files ready to upload, a comparison with a competitor. If a paid audit only makes the PDF longer, walk away.
And in both cases, stay in control: an audit tells you what to do, it does not force you to buy the fixes from whoever wrote it. An honest provider lets you leave with the plan. To go further: the 7 measurable causes of a website that is invisible on Google.