The dashboard a company director actually reads

“Can you send me the report?” The report went out on its own, on the first of every month, straight to his inbox. In six months he had opened it three times.

So I asked him what he actually looked at when he wanted to know whether the marketing was working. The answer fit in one sentence: “whether revenue went up, and whether it cost me more than before to get it”. The report held twenty-two charts. Not one of them answered that sentence.

A dashboard is not judged on how many numbers it displays, but on how many decisions it has caused.

1. The test I put every number through

Before adding a line, one question: if this number moves 20 % next month, what do I do differently?

If the answer is “nothing”, the number comes off. It is not wrong and it is not useless, it simply has no business on the page a company director reads. There are two families of indicators. Those that describe what happened, and those that trigger a call: put more in, cut, shift budget from one channel to another, phone somebody. An executive dashboard holds only the second family.

On the accounts I take over, that test removes two thirds of the lines. Twenty-two charts become five numbers, and the document starts getting read. The rest does not vanish: it moves down to where it is genuinely useful, which is the last section of this article.

2. Three numbers that almost never pass the test

They do not lie. They measure something other than what people credit them with.

Bounce rate. It no longer means “the visitor left straight away”. Google Analytics documentation defines an engaged session as one that lasts longer than ten seconds, or contains a key event, meaning an action you yourself flagged as important, or counts at least two page views; bounce rate is simply the share of sessions that were not engaged. Somebody who reads your article for forty seconds and leaves satisfied is therefore “engaged”. Somebody who works out in eight seconds that they are on the wrong site and turns back is a “bounce”. Both behaviours are healthy. A telling detail: neither metric appears by default in most reports, you have to go and switch them on. A number that takes effort to surface and triggers no decision afterwards does not go up to a director.

Average position. Google states plainly what it contains: the topmost position held by a link to your page, averaged across every query where it appeared. A page can therefore watch its average position slide while the work is going well, purely because it has started surfacing on dozens of new, distant queries. The documentation says as much itself: it is a complex metric that can be misleading if you do not grasp the subtleties, and it recommends watching change over time rather than the absolute value. I keep it for my own work. I do not put it on a director’s page.

Sessions. The most commented number in meetings and the least actionable. A rise in traffic tells you neither where it came from, nor what it is worth, nor whether it turns into anything. I have seen a site gain 30 % of visits in a quarter without a euro landing: the rise came from a blog post that was pulling in students. Good news for the domain’s authority, no decision to take.

3. The five numbers that stay

These pass the test. Each one can change a decision within the month.

1. Revenue, and the margin behind it. Google itself invites you to feed back the real value: its best-practice page on value-based bidding asks you to specify what you want to maximise, sales revenue, profit margin or lead score. A return on ad spend of 4 on a product carrying 12 % margin and a return of 2.5 on a product carrying 40 % are not comparable. Until margin reaches the tool, you are optimising a leaderboard of false winners.
2. Acquisition cost set against what a customer is worth. A cost per lead of 40 € is neither good nor bad on its own. Against an average basket of 90 € that never repeats, it is fatal. Against a subscription kept for two years, it is a bargain.
3. The conversion rate of the enquiries themselves. This one does not come out of the measurement tool, it comes out of the sales software, or the notebook. It is what separates “we are short of enquiries” from “we are short of good enquiries”, and the two call for opposite moves: in one case you add budget, in the other you tighten the targeting. I have seen cost per lead halve without a single euro of revenue moving.
4. The split between brand and non-brand. If most clicks come from people typing your name, you are measuring your reputation, not your acquisition. It is an uncomfortable number, and it is the one that reframes a budget conversation best.
5. The lag between the first click and the money landing. On a two-month cycle, a director judging the month on the 5th is judging work that has not had time to produce anything. Putting that lag at the top of the page saves ten conversations.

4. A number with nothing to compare it to triggers nothing

“3,412 clicks” says nothing. “3,412 clicks, against 2,890 last month and 3,380 this time last year” says something, and the third column is the one most often left out. Without it, a seasonal retailer reads an alarming collapse where he is simply living his usual August.

There is a technical trap on that third column. Data retention can be set to 2 or 14 months for user-level data, and to the same values for other event data, with 26, 38 and 50 months reserved for the paid 360 edition. Standard aggregated reports are unaffected, but every exploration is. Past fourteen months, in other words, the tool will not let you rebuild a two-year comparison. The setting sits at 2 months by default on plenty of properties: it is the first thing I check when I arrive on an account. And I have the month’s five numbers copied into a spreadsheet, line by line. A lost history cannot be bought back.

5. Do not call it too early

The instinct of a director finally handed a readable dashboard is to open it every morning. That is the surest way to make a bad call.

Google gives its own orders of magnitude on the page quoted above: aim for at least fifteen conversions a month at account level, and allow a ramp-up of at least two weeks or three conversion cycles before evaluating a strategy. On an account running twenty conversions a month, comparing one week with the previous one amounts to reading noise, and cutting a campaign that had done nothing wrong. It needs to mature. The executive dashboard is read once a month, on the trend; fine-grained steering is another job and another document.

6. Two dashboards, not one

This is the practical conclusion of everything above, and the mistake I see most often: wanting one single document that serves both the director and the person doing the work. What you get is a document that serves neither, too dense for one and too thin for the other.

On one side, one page, five numbers, three comparison columns, read monthly. On the other, the detail by campaign, by query, by landing page, opened every week by whoever has their hands in the account. Both feed on the same data, they do not tell the same story. I described how to build those two documents in the article on Looker Studio reporting; this one is about what goes inside them, which is a different problem.

The five numbers above assume the measurement holds up. When conversion tracking loses half the path, no dashboard rescues that, and it is a subject in its own right that I covered in the article on what Google Analytics 4 no longer sees. You fix the thermometer before arguing about the temperature.

What is left at the bottom of the crucible

A twenty-two chart report is raw ore presented as a result. The data is there, nobody disputes its presence, and nobody does anything with it. The work is to melt it down, let the spoil run off, and keep only the metal you can weigh. Five numbers a director can actually rule on are worth more than the whole mine. The rest waits below, available to whoever is digging.