The dashboard nobody reads, and the report that works

A dashboard that nobody opens is usually diagnosed as a discipline problem, and it almost never is. The first real reason is volume: a screen with twenty charts requires the reader to decide what deserves attention before they can learn anything, and that decision is work. Faced with it weekly, people stop. A report that has already made that decision — here are the five things, here is the one that is off — asks nothing of the reader except a response.

The second is latency. A dashboard showing figures that are days or weeks stale gets checked once, found out of date, and mentally filed as unreliable, which is difficult to reverse. The third and most fundamental is that it requires interpretation. A chart showing a line that moved does not say whether the movement is good, expected, within normal variation, or a problem. Supplying that judgement takes knowledge of the business and ten minutes of thought, and the person opening the dashboard usually has neither available at that moment. So the chart is looked at, nothing is concluded, and the habit dies — not from laziness but because looking produced nothing.

The version that survives is smaller than anyone expects. Five metrics, each with the current value, a target or expected range, and a status that says plainly whether it is fine or not. One sentence, written by a person, naming the most important thing that changed and what is being done about it. A fixed arrival time — the same morning each week — so it becomes part of a routine rather than something to remember. That is the whole thing, and it fits on one screen or one side of paper.

Each element is doing specific work. The target converts a number into a judgement, which is the interpretation step the reader cannot perform. The status makes the judgement readable in a second. The written sentence carries the causal reasoning no chart can express, and it is the part that makes the report worth opening. The fixed schedule makes it a habit. Notice how much of this is not visualisation: the reason the one-page version outperforms the elaborate dashboard is that a person did the interpreting before sending it, and interpretation is the scarce input, not display.

A figure without a reference point cannot be read. Revenue of ₹4,20,000 is meaningless until you know whether ₹4,00,000 or ₹6,00,000 was expected, and the comparison that matters is rarely last week. For seasonal businesses the honest reference is the same period last year; for anything volatile it is a range rather than a point, because a single target guarantees the metric is described as off most weeks and status colours that are always amber stop being read.

Setting the range is where the actual thinking happens, and it is uncomfortable because it requires committing to an expectation in advance. That discomfort is the value: a target written before the period forces you to state what you believe, which is what makes the outcome informative afterwards. Ranges set from your own recent history — the typical band over the last several periods — are usually more useful than aspirational figures, since a target nobody expects to hit trains everyone to ignore the status beside it. Where you genuinely do not know what to expect, write that down as the range's absence rather than inventing a number, and let the first few periods establish the band.

The reason most reporting produces no change is that it ends at information. A number that is off appears, everyone sees it, and no specific person is expected to do anything, so the same number is off next week and the week after until it stops being remarkable. This is the most common failure of small-business reporting and it has nothing to do with the report's design.

The fix is unglamorous: each metric has a named person, and when it is outside its range that person is expected to say what they are doing, in one line, in the next report. Even in a business of one this matters, because writing "margin down again, still not investigated" three weeks in a row is uncomfortable in a way that noticing it silently is not. It is also worth allowing the report to conclude that nothing needs doing. A review that must generate action generates busywork, and the credibility of the whole exercise depends on "all five within range, nothing to report" being an acceptable and common outcome.

It is worth being explicit about the assumption that underlies most dashboard projects: that decisions are limited by information. Occasionally true, usually not. The common constraints are that nobody has decided what the business is optimising for, that the person who could act is not the person reading, or that the required action is known and unattractive. None of those is relieved by adding charts, and all of them are disguised by a dashboard project, which feels like progress while changing nothing.

The practical consequence is that building a report should start from a decision rather than from available data. Ask which recurring decisions are made in this business — what to order, what to charge, who to chase, where to spend, whether to hire — and what figure would change each one. Metrics that survive that question belong in the report; the rest are available when someone has a specific question. This inverts the usual process, which starts from what the software can display and works outward, and which reliably produces something comprehensive and unread.

It reports what happened and not why, and the single written sentence naming a cause is a hypothesis rather than a finding. That is fine as long as it is labelled honestly — "margin down, I think because of the courier rate change, checking" is useful and appropriately hedged, while a confident false cause is worse than none because it closes the investigation. A weekly report is also a lagging document: it describes a period that has closed, so it can only ever be an early warning system, not a steering wheel.

It will also miss anything you did not think to measure, which is by definition where the surprises come from. A short report is a set of choices about what matters, and those choices were made with last year's understanding of the business. That is the argument for reviewing the list itself once or twice a year, and for keeping some capacity to investigate questions the standing report cannot answer. The report handles the known; it is not a substitute for someone occasionally going and looking at the underlying transactions, talking to customers, or asking why a number nobody tracks has started to matter.