What will the fix cost? Quantifying effort with JMX

The report has gone out, 340 findings, the client reads as far as page three and calls: “So what does this cost?” The Issues view counts findings, not hours — and a number pulled out of thin air is either too high, in which case the job goes to somebody else, or too low, in which case you work off the difference. What gets expensive is not the estimate but that it cannot be evidenced: if you cannot say which line causes which hours, you cannot negotiate what gets cut either.

How to go about it in JMX

1. Open the “Tasks” tab. Above the table sit four tiles — Do now, Schedule, Alongside, Defer — each with the number of tasks and the effort inside. Those four totals are the answer to the question, in working time.
2. Read the “Kind” column. It gives effort per finding and reach, for instance “Trivial, one-off” against “Small, per page”. 412 pages without a meta description are 412 texts, 412 pages without a viewport meta are one line in a template. In the issue list both show the same number.
3. Bring in the “Leverage” column — points per person-day. Sorting is by the field of the matrix first and only then by leverage: 0.2 points in ten minutes would otherwise stand above twelve points in three days.
4. Set the “At most” filter. It shows everything up to a given effort per finding — the list for the question of what can still be done today.
5. Store segments (Settings → Segments). Shop, blog, careers: the breakdown with findings per page then sits above the rule list in the Issues view — who takes on the work depends on it.
6. Check the detail column. For the selected row it gives the rule’s recommendation verbatim and up to five affected addresses; Show every occurrence jumps into the Issues view filtered to that rule.
7. Export. The file extension decides: PDF with metrics and every task including its instruction, for the quote; Excel with a cover sheet and one row per task, to tick off. Without a window, jmx-cli --tasks tasks.pdf writes the same thing.

What to watch out for

The hours are calculated, not measured: 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 is costed at three times what a single finding of the same rule costs, because a template wants checking. Your hourly rate is in none of these numbers. JMX knows neither your CMS nor your approval loop — converting working time into a price stays your job, and that is why the column is in hours and not in euros.

The reach is deliberately cautious. Only what structurally sits in one place counts as one-off: robots.txt, server configuration, language markup in the shell. Everything else counts per page. If your CMS generates the missing descriptions from a single template field, JMX estimates too high — that is the direction in which an estimate may be wrong.

A critical finding always counts as high impact, even where it barely registers in points: a redirect loop on ten thousand pages comes to almost nothing arithmetically and is still the first thing somebody has to touch. Findings from the log analysis stand in the list with 0 points of impact, because the score knows nothing about crawl budget. And without a license every analysis stops at 25 pages — extrapolating an effort total from 25 pages to four thousand is not an estimate.

A number that comes out of a list survives the follow-up question: if the budget runs out at line 14, you are no longer negotiating the price but lines 15 to 40.