In September 2026, ResaHost's Ahrefs audit showed 116 errors and a Health Score of 43. That is an uncomfortable result for your own website. Particularly when building websites and doing SEO is part of your business.
It would have been easy to start working through the report one line at a time. It would also have been slow. Some rows described the same problem appearing across different pages. We first needed to establish where the fault originated and how widely it affected the site.
One important finding involved email addresses. Cloudflare's email protection was modifying them in the site's public HTML response. The crawler received protection links that returned errors. In our own browser, an email link could still appear to work.
This was a useful reminder for me too. A page that looks right does not establish what a search engine receives. The source file, the server's response and the final view in a browser can differ. We needed to examine the response that the audit's finding actually referred to.
We corrected how the email protection was handled and added a check of the public response. We investigated other findings at the same time. One guide linked to a status page that did not exist. The fix was to correct the guide and direct the customer to the proper place to ask about service availability.
Another page had appeared slow in a single measurement. We repeated the measurement before making changes. The original delay did not recur. That did not justify the same corrective work as a link that repeatedly failed.
In the completed follow-up audit on 10 September, the Health Score was 100 and the error count was zero. Four warnings remained. Those figures describe that particular audit. The site continues to change, so later updates need to be assessed on their own terms.
I cannot turn that result into a claim that sales increased by the same proportion or that Google moved every page to the top. This measurement tells us neither of those things. It tells us that the errors identified in that audit had been removed. Traffic, enquiries and sales need their own evidence.
There are two practical lessons here for a CEO. The first concerns commissioning work. When a specialist presents a hundred findings, ask them to group the findings by cause. Otherwise, a hundred rows can quickly look like a hundred separate jobs. The second concerns completion. After a correction, return to the original observation and establish whether it has gone away.
We need to do this just as well for ourselves as we do for clients. The ResaHost website gave us a concrete piece of work to resolve. In Resaco's client work, the customer deserves an equally practical answer: what was broken, what changed and what evidence shows that the change worked?
The most useful lesson was that the length of a report and the size of a problem are different things. A repeated fault can look enormous. A single misleading conclusion can be easy to miss. Both require us to investigate beyond the first observation.
That is also why I consider the interpretation in a website analysis report part of the product's quality. The customer's next decision needs to rest on what we actually know.
