A report you will actually read

The usual outcome of setting up analytics is a dashboard that is opened weekly for a year and never changes a decision. The fix is not a better tool. It is deciding in advance what each number would make you do.

Three horizons, three different reports

Most reporting fails by mixing them: a weekly report full of numbers that only move over quarters, which teaches you to ignore it.

• Weekly, and short. Only things that can genuinely change in a week and that you would act on: outcomes, spend, anything broken. Five lines.
• Monthly. The rate, the sources, the reconciliation against your own records, and last year's same month for comparison.
• Quarterly. The slow ones: repeat rate, cost per customer over time, whether searches for your own name are growing.

Keeping them separate is what makes the weekly one survive, because everything in it is genuinely weekly.

Write the decision next to the number

The thing that turns a report into a tool. Beside each line, the action if it moves.

Illustratively: enquiries, and if they fall two weeks running, check the form and the top landing page. Spend, and if cost per enquiry exceeds a stated figure, pause the campaign. Site speed, and if the field data leaves the good band, look at what was installed recently.

Written this way, the report is a list of conditions rather than a collection of figures, and a week where nothing crossed a threshold is a week with nothing to do, which is a legitimate and under-rated outcome.

Keep the record where it survives

A spreadsheet with one row per month and a dozen columns. Unimpressive, portable, and it will still be readable in five years when you have changed analytics tool twice and lost the history in both.

Record the figures and the context: what you changed, what launched, what broke. A year later the context is what makes the numbers interpretable, and it is the part nobody writes down.

What to leave out

• Anything with no action attached.
• Anything you cannot influence.
• Vanity totals: pageviews, followers, impressions without outcomes.
• Charts that look impressive and answer no question. A number with a comparison beats a graph of it.
• Anything requiring a tool nobody else can open. A report only one person can produce stops existing when they are busy.

The alert that replaces most of the watching

Most of what a weekly report is actually for is noticing that something broke. That job is done better by an alert than by a human reading figures, and it frees the report to be about decisions.

• Zero outcomes in a period where you normally have some. The single most valuable alert there is, because a broken form or checkout is invisible in a chart until the week is over.
• Site down, from an external checker rather than from your own hosting dashboard.
• Advertising spend above a daily figure, which catches the misconfiguration before it costs a month's budget.
• A tracking tag missing from a page, if your tooling can tell you. This is how a redesign silently stops measurement.

Four alerts, all of which fire rarely, and between them they cover the failures that a weekly review would catch days late. The point is not vigilance; it is that a human reading a dashboard is a poor smoke detector and an excellent decision-maker, and the two jobs should be separated.

Then the report can be what it should be: a short monthly conversation about what to do, rather than a weekly check that nothing has fallen over.