Analytics & Attribution

Reporting Cadence That Actually Gets Read

A report nobody reads is worthless, no matter how correct the numbers behind it are. Most reports fail for the same reason: they leave nothing out. Everything gets shown — every keyword, every impression, every dashboard screenshot — and it reads as noise. The fix isn't prettier charts. It's discipline about how often you report, how long the report is, and, hardest of all, what to cut before it goes out.

The One Test

"So what would you do differently?"

Before anything else, run every number through one question: so what would you do differently because of this? If the answer is a specific action — change the page, rewrite the title, kill the campaign, double the budget, do nothing on purpose — the number stays. If the answer is a shrug, it goes. Not into an appendix. Out. Apply this honestly and most reports lose half their length in an afternoon: impressions on a query you'll never target, average position across 400 keywords, bounce rate with no follow-up — all real, none of them changes what anyone does on Monday.

How Often

Monthly is usually right

Monthly is long enough for the work to show up in the data and short enough that nobody feels ignored. Search especially is slow: you change a page today, it gets recrawled over days or weeks, and only then does anything reach a chart. Watch that weekly and you're mostly watching variance — the line goes up, the line goes down, you explain both, you've explained nothing. Weekly reporting is usually theatre; it rewards fast moves over the slow ones that compound, and trains everyone to react to noise.

There are real exceptions. Report weekly when the spend is large enough that a bad week is expensive — an active campaign, a launch, a migration, a live incident. In those cases you're not reporting, you're monitoring: say so, keep it to a handful of numbers, and stop when the reason stops. And once a quarter, step back for the questions a monthly report can't answer — is the strategy working, what did we learn, what changes next. That's a different document with a different job; don't make one report do both.

How Long

Length is unsorted effort

A 40-page report isn't a deliverable — it's a confession that nobody decided what mattered, so everything got included and the reader was left to sort it out. The work of a report is the sorting. A good monthly report is two to four pages, or one screen you can scroll, and it holds five things: the headline (one or two plain sentences), the three-to-seven numbers that matter (each with a comparison and a reason), what you did (briefly, only where it connects), what's next and why (with owners and dates), and what you need from anyone (if anything). If something doesn't fit one of those slots, it probably doesn't belong. Keep the raw data available — just don't mail it out every month as if volume were value.

Write For The Reader

Lead with the answer

Be honest about the audience: whoever reads this has ten minutes, half a coffee, and eleven other problems. They read the first two sentences, the one number they already care about — usually leads, calls, bookings, or revenue — and whatever's in bold. Then they stop. So put the answer first. Don't build to a conclusion; open with it: "Enquiries were up 18% on last month, driven by the service pages we rewrote. One problem to flag: the contact form broke for four days and it's fixed." The rest of the document exists to support that opening for whoever wants to check the work. Most months, nobody will — and it should still hold up.

The Hard Case

A bad month goes in the first paragraph

Every account has bad months, and how you report one decides whether trust survives it. The instinct is to bury it — lead with something positive, mention the drop in the middle, use the passive voice. Everyone can smell that, and it costs more trust than the bad number ever would. Instead: say it first. Separate cause from blame honestly — was it outside anyone's control, something that got changed, or something you missed? If it's the third, own it early: "We missed that the staging site was indexed for nine days. That's on us. It's fixed, and here's the check we've added." Then say what changes and what number should move if you're right — and next month, report against exactly that.

If You Report to Clients

The same discipline, one more reader

If you're an agency reporting to clients rather than to yourself, none of this changes — it just gets a name. The person who approves your retainer is often not the person you sent the report to, so "lead with the answer" and "two to four pages" become survival traits: one page of clear judgement survives the hand-off; a 40-page deck dies in a Slack thread. Report outcomes first and use activity to explain them, never as a substitute — a client shown only activity eventually notices they're paying for effort rather than results, usually on the day they're looking for something to cut.

What ARIA Shows You

The report, already sorted

The mechanics of reporting — pulling the data, checking it, deciding what moved and why — are hours a month before you've written a word, and when it's tedious the temptation to export everything and call it a report gets stronger. ARIA reads Google Search Console and produces a plain-English audit with graded scores, each one backed by the metric it came from, so you can hand it over without translating first. A read might come back as:

"This month's one-line headline: organic enquiries rose from 34 to 41, driven by the service pages rewritten last quarter. One flag: three pages slipped in ranking after a competitor refreshed theirs. Everything else held — not worth a line."

What ARIA won't do is decide what matters to your reader. That's still your job. It's the job.

Keep Reading

Report what matters, not everything

Connect your site and let ARIA do the collection and the arithmetic, so the report you send is short, honest, and already sorted — leaving you the one part that was always yours: the judgement.

See My Site