After a failed contractor: find what's broken and fix it

By measurable things. We look at how the domain responds to the request, whether the test order arrives, whether the analytics data is really collected, whether there is a site map and whether there are no ads in the markup about something that does not exist on the page. All this is either yes or no - there is no room for taste here. Each item in the report comes with a way to check it yourself, so you don't depend on either our rating or the previous performer's rating. But "I don't like the font" is not included in such a report: this is a matter of redesign.

Those that are not visible at first glance. In the tool store we measured, the certificate was valid, the pages opened, everything looked fine — but the bare domain returned a 301 to an insecure protocol, and anyone who dialed an address without an internal page ended up on an insecure connection. Internal addresses worked correctly. The language redirect from the root was broken, one line in the configuration. The certificate itself does not save anything: it encrypts the channel, and that's where its role ends.

Constantly. In the clinic site we measured, the inline tag manager code blocked itself: the identifier variable was written with the analytics identifier instead of the container identifier, and the exit condition compared the variable to that value - so the exit always fired and the container did not load at all. The code was on the page. Such things are found by the reader, not by the scanner. And separately: the restored meter itself does not improve anything, it only gives an opportunity to see; blockers, consent mode, and browser cookie restrictions make data a trend, not an accounting truth.

Might, and usually unintentionally. In the app we measured, the structured data announced a rating with 50 ratings, even though the page had no mention of reviews. This self-imposed feedback marking is a direct reason for manual sanctions. We clean such places first, along with critical ones. Just don't get confused: honest markup does not increase positions and does not force the search engine to show an extended snippet, it only gives a chance. A false one creates a risk, and this asymmetry is not in your favor.

We fix the "as received" status. We start the repository, if it was not there, take a full copy of the files and the database, set up the test platform - and only then touch the first line. This is our own recommendation even to those who will fix the site not with us: without a fixed rollback point, any correction in someone else's code is a blind game, and the dispute "it worked for you" becomes insoluble. A day of work on this step saves a week on the next.

No. We write what we found and how to check it, without evaluations of personality and qualifications. Some of what looks like a mistake turns out to be a conscious decision in its circumstances — a limited budget, a customer requirement, outdated hosting. The part is really a mistake. It is not our job to understand the motives; ours is to show the facts you'll move forward with, including talking to the same contractor if you plan to.

Sometimes cheaper — and we will say so if the list of defects exceeds the cost of a new assembly. But more often it turns out that there are three or four critical places, and their repair costs many times less than a new development, and the most important thing is that it does not entail a change of page addresses. Therefore, we give the estimate by points: you can see the cost of each correction separately, and you decide what to take and what to leave as it is.

Diagnostics — 3–5 working days. Correction — from 11 to 38 working days depending on the list. Critical, which is losing money now, we suggest closing with the first outlay in a few days, without waiting for the rest. The longest part of this work is not the code, but the approval of the disputed points of the report: if you have one person with the right to say "yes", the deadline is significantly shortened.

We do not promise this and do not advise anyone to promise it. Removing duplicates does not add traffic - it stops scattering it, and in the first weeks the number of pages in the index may even drop, this is the expected course of events. The sitemap does not guarantee indexing, the real numbers are visible only in your Search Console. Speed ​​doesn't make an offer more attractive either: a site that loads in 200ms will simply give a person a reason to leave sooner. We remove technical waste - the thing that causes people and orders to get lost along the way. Demand is not created by repairs.