Audits have a credibility problem, and it is deserved. A great many of them are produced by running a tool, exporting the output, adding a cover page and invoicing. The result is technically accurate, entirely unprioritised, and impossible to act on.
That is a shame, because a real audit is one of the highest-value things a small business can buy: it is the difference between guessing and knowing, for a fixed and modest cost.
1. Numbers, not adjectives
“Your page is slow” is not a finding. “The first readable content appears 4.2 seconds after the tap on a mid-range phone on 4G” is one, because it can be checked, argued with, and measured again after the fix.
Every item should carry a number or a specific observation you could verify yourself. If a finding cannot be checked, you cannot tell whether it was real and you certainly cannot tell whether it got fixed.
2. What each problem costs — in the reader's terms
Not severity out of five. Severity scores come from the tool's opinion, and the tool does not know your business. What you want is a sentence about what happens to the person on your page: they cannot read the text outdoors; they cannot tell which of four buttons is the real one; they are asked for a phone number before they know if they want to talk to you.
3. An order, and the reasoning behind it
The most valuable line in any audit is which three things to do first. It is also the hardest to write, because it requires a judgement no tool can make: what is expensive here, given who this business is and who is arriving.
Watch for ordering by how easy something is to fix. That is the developer's order, not the customer's, and it produces a fortnight of tidy work that changes nothing anybody would notice.
4. Fixes specific enough to hand over
“Improve your calls to action” is not actionable. “These four links all say ‘read more’; here is what each should say instead” is. The test is whether a developer who has never spoken to the auditor could implement the item from the sentence alone.
5. An explicit statement of what it did not look at
A trustworthy audit says where it stops. It did not evaluate your pricing. It did not judge whether your product is any good. It could not see the logged-in area. It says so, in writing, so you do not treat silence as approval.
An audit that appears to cover everything is either enormous or dishonest about its edges.
6. Language you can repeat to somebody else
You will have to explain the findings to a developer, a partner or a board. If every item is written in jargon, that conversation goes badly and the work does not happen. A good audit gives you the sentence to use.
Four signs you are buying a report generator
| Sign | Why it matters |
|---|---|
| Over sixty findings, unordered | Nobody triaged it. You are being sold the tool's raw output |
| Severity scores with no reasoning | The number came from software that knows nothing about your business |
| No mention of what was not covered | Either it was not thought about, or it is being hidden |
| Generic recommendations with no example from your site | It was not written about your site |
What an audit cannot do, whoever writes it
- It cannot tell you whether people want what you sell.
- It cannot tell you whether your price is right.
- It cannot fix badly targeted traffic. It can tell you that is what you have.
- It cannot promise a conversion figure, and anyone who does is selling confidence.
Those limits are not a weakness. An audit that is clear about them is more useful than one that implies it answers everything, because you know what you still have to find out elsewhere.
A reasonable price
Conversion consultants generally charge between $100 and $300 an hour, and senior ones more, so a serious audit of a small site tends to land in the several-hundred-dollar range. Much cheaper usually means the tool output with a cover page. Much more expensive should come with a scope that explains it.
Whatever you pay, the price and the scope should be known before you start. An audit billed hourly with an open end is a strange thing to buy, because neither side knows what is being agreed.
Questions people also ask
How long should an audit take?
A single page, two to three days. A small site, one to two weeks. Much faster usually means automated; much slower usually means it was not scoped.
Should I get an audit before a redesign?
Yes, and it is the best possible time. A redesign without a diagnosis reproduces the same problems in a new visual style, and you find out a year later.
Can I just use free tools?
Partly, and you should start there — free tools cover the measurable checks well. What they cannot do is prioritise, look at the journey between pages, or write the findings so somebody else can act on them.
What if the audit says my site is fine?
That is a real and useful result: it tells you the money is going somewhere else, which is usually the offer, the price or the traffic. An auditor who cannot deliver that conclusion will always find problems, whether or not they matter.