From the findings export to a report the client understands

The crawl is done, the Excel export has 3,000 rows, the mail has gone out. Two weeks later nothing has happened. At the follow-up meeting comes the question that costs you the next job: “And what does that mean for us?” A list of findings answers what is wrong — but what was asked is what to start with, how long it takes and who does it, and as long as that second question stays open, an audit gets paid for once and never commissioned again.

How to go about it in JMX

1. Describe the areas of the site under Settings → Segments — shop, blog, careers — with a name and a pattern each: URL contains, path starts with, regular expression, host. The Issues view then shows the breakdown above them with findings per page. “412 pages without a meta description” turns into the question of which department handles it.
2. Open the Tasks tab. Above the table sit four tiles: Do now, Schedule, Alongside, Defer. Every row carries the impact in points, the estimated effort in working time and the leverage — points per person-day.
3. In the Report tab, create a template per client: logo (embedded, up to 2 MB) and accent colour, then move, switch off and rename the 15 sections. If you talk internally about a “visibility audit”, the client should read that word in the document too.
4. Fill in the opening note. It is the one field no tool can produce: why the audit was done, what prompted it. That is what every client asks first.
5. Under Level of detail, decide who the document is for. Turning “list affected URLs” off turns forty pages into two; “from this severity” set to Error takes the notices out; “print the recommendation” stays on, otherwise you produce a list of defects.
6. Go through the “Why this matters” boxes. Every section brings a text that is preset in the editing field; your own wording replaces it permanently in that template.
7. Generate report — the extension in the save dialog decides between HTML and PDF. For whoever ticks things off, add the task list as an Excel workbook with a cover sheet and one row per task.

What to watch out for

The most common misreading concerns the points. The impact in the task list says what a finding currently costs the score — not by how much the score rises after the fix. A site sitting at 0 only moves once enough else is gone. The same applies in the other direction to the accessibility score: 100 means “these classes of defect are absent”, not “accessible”. Colour contrast, the actual focus order and the quality of a description are seen by no static check, and that sentence is therefore in the report as well.

The effort is an estimate and meant as one: around six minutes for something trivial, a good twenty for a short text, an hour and a half for real work on a page, three quarters of a day for a project. A change that only needs making once on a template is costed at three times the respective individual effort. Anyone with figures of their own from experience will expect different hours — say so in the meeting, before the client takes the number for a quote.

Two special cases otherwise cost rework. A section without data is silently skipped when generating; tick Core Web Vitals and never measure, and without the list above the fields you notice only after sending. And the logo stays out of the PDF: the built-in PDF generator sets text, lines and areas but embeds no images. If you need the logo, send the HTML version.

Whoever has the order explained to them can commission the second step — and that is decided not by the crawl but by the half hour afterwards.