Six steps, and none of them skip ahead.
The order is the method. An observation is not a diagnosis, a diagnosis is not a proven cause, and an intervention has no effect until it is re-measured.
- 1
Observation
Something measurable is recorded. "Search agents received a 403 on every request for nine days." No interpretation yet, just what was seen and when.
- 2
Diagnosis
The mechanism behind the observation, identified and stated as a claim that could be wrong. "A firewall rule is challenging non-browser clients." This is where most agencies skip straight to selling.
- 3
Opportunity
What would be worth having if the diagnosis is correct, sized honestly. Not a revenue projection, because that would require causal evidence we do not have yet.
- 4
Recommendation
The specific change, at the level of a named rule or a named page. Vague advice cannot be checked afterwards, which is precisely why vague advice is popular.
- 5
Intervention
The change is applied, by us or by your team, and the date is recorded. Without a date, before and after are not distinguishable.
- 6
Re-measure
The same measurement runs again on schedule. Until it returns, the intervention has an unknown effect, and that is exactly how it is reported.
What people ask before signing.
How long does the first audit take?
The automated retrieval checks run in about a minute. The reviewed version, where a person reads the result and writes the diagnosis, is returned within two business days.
What happens if a measurement fails?
It is reported as failed and excluded from the score. A check that produced no verdict is never counted as a failing check, because that would make a broken measurement look like a broken website.
Do you need access to our website?
Not for the audit, which only reads public pages. Implementation work needs either access or a developer on your side to apply the changes. We are equally happy handing a specification to your team.
Step one is free.
The audit is the observation stage of the loop. Everything else follows from what it finds.