We Built a Dashboard, Then Deleted Most of It

Our first dashboard had fourteen numbers on it. It looked like the dashboards I had seen at bigger companies, which was the point. Somebody would open it in the morning, everyone would feel informed, and the day would proceed exactly as it would have anyway.

That last part is what eventually bothered me. There are three of us. If a number cannot change what one of us does this week, it is costing attention and returning nothing.

So we applied a single test to every metric on that screen, and eleven of them failed it.

The test is what you would do differently

A number earns its place only if you can name the specific action attached to it before you start watching it. Not a vague intention to improve. An actual move: we stop feature work, we rewrite that screen, we call the partner.

Try it on your own reporting and it gets uncomfortable quickly. Page views, follower counts, total registered users, the breadth of features people touched. These are all real quantities and most of them are going gently up, which is pleasant. Ask what any of them would cause you to do on a Tuesday and the answer is usually nothing.

Lagging financial summaries fail this test too, which surprises people. Last month's revenue tells you what already happened. It is essential for planning and useless as an operating signal, because by the time it moves, the thing that moved it happened weeks ago and you were not watching that.

The three we kept

The first is how long it takes a new person to do the thing the product exists to let them do. Not sign-ups. The first real action. When that time gets worse, we stop building anything new until it recovers, because everything downstream of it is being taxed.

The second is how often the same confusion appears in support conversations. One person misreading a screen is a person. Two people in the same week misreading the same screen is a design defect that we have chosen to answer manually instead of fixing. That reframing matters, because it moves the problem out of the support queue and into the product backlog where it can actually be closed.

The third is unusual and it is the one I would recommend most to any owner of a very small business. We track how long work sits waiting on a decision. Not how much we produce. How long things wait. Anything unowned for more than about a day gets pulled into the open, because in a company this size waiting is almost always the largest cost and it is invisible in every conventional report. Nobody logs the hours a task spent parked.

Why smaller teams need fewer numbers, not simpler ones

The usual advice is that small businesses should track simplified versions of what large businesses track. I think that gets it backward. Large organizations need broad instrumentation because no single person can see the whole operation, so the dashboard substitutes for direct observation.

At three people we can see the operation directly. Anything the dashboard tells us that we already know from doing the work is pure duplication. What we need instrumentation for is the narrow set of things that are genuinely invisible from inside the work, and there are far fewer of those than the software industry would like to sell you.

That is why the answer was not a simpler dashboard. It was a much smaller one.

The part where this can hurt you

I would rather name the risk than pretend it does not exist. Watching three numbers means accepting that something slow and structural could deteriorate for months without appearing anywhere. A gradual decline in a segment you are not looking at will not announce itself.

Our mitigation is deliberately low technology. Once a month we look at the whole picture properly, including the boring lagging figures, specifically hunting for slow movements rather than daily signals. It is a different activity with a different rhythm, and conflating the two is what produces the fourteen-number screen in the first place.

There is also a real failure mode where a team picks three numbers, attaches actions to them, and then never revisits whether those are still the right three. Ours have changed twice. The test survives; the metrics do not have to.

What to do with this on Monday

Open whatever report your business currently runs on and go line by line asking what you would actually do if each number moved by a fifth in either direction. Write the action next to the number. The lines where you cannot write anything are not information, they are decoration, and they are competing for the same attention as the lines that matter.

Most owners I have said this to find they are left with two or three. That is not a sign that the business is too simple to measure. It is a sign that most reporting is built to look thorough rather than to be used, and a company small enough to act quickly is exactly the kind that cannot afford the difference.