Every growing business we meet has data. Remarkably few have measurement. The distinction matters: data is what accumulates in your systems; measurement is the deliberate practice of using a small number of trusted figures to steer decisions. The first is a by-product. The second is built.
The typical pattern in a company under £10 million looks like this: the accounts arrive weeks after month end, accurate but historical; the sales figures live in a CRM that half the team updates; operations run on spreadsheets maintained by one indispensable person; and the numbers quoted in Monday's meeting depend on who compiled them. When figures disagree, the meeting debates the figures. The decisions wait.
Lagging tells you the score. Leading changes it.
Revenue, profit and cash are lagging indicators: they report the consequences of decisions made months earlier. They are essential, and insufficient, because by the time they move, the cause is history. Leading indicators sit upstream: enquiries received, quotes issued, conversion rate, order book cover, capacity utilisation, error rate, time from order to invoice. These are the dials a management team can actually turn during the month in question.
A useful test of any management pack: what fraction of it could still change this quarter's outcome? If the honest answer is none of it, you have a scoreboard, not an instrument panel.
Fewer numbers, held to a higher standard
The instinct, once measurement begins, is to measure everything. Resist it. Attention is the scarcest resource in a growing business, and every measure on a dashboard taxes it. We would rather a client tracked eight numbers they trust and act on than forty they skim.
Each measure that earns a place should have four properties: a precise definition, so two people cannot compute it differently; a named owner; a target or expected range; and a stated action for when it strays. A number with no owner and no consequence is decoration.
A KPI without an owner and a consequence is not a measure. It is wallpaper with digits on it.
Respect variation, or it will fool you
Numbers wobble. Sales dip in a short week, complaints cluster by coincidence, and a metric drifts within its natural range. Management teams unfamiliar with variation react to every wobble, praising the noise upwards and punishing it downwards, and in doing so they teach the organisation that the numbers are a lottery.
The statistical remedy is old and underused: understand the normal range of each measure before reacting to any single value. A simple run chart, showing the last eighteen months rather than the last data point, prevents most false alarms. Where the stakes justify it, control charts distinguish signal from noise formally. The mathematics is not exotic; the discipline of using it is what is rare.
One source of truth, one rhythm of review
Two builds make everything else work. The first is a single source of truth: each measure calculated one way, from one agreed system, automatically. The monthly ritual of reconciling competing spreadsheets is not quality control, it is waste wearing a tie.
The second is a decision rhythm: a short weekly review of leading indicators, where exceptions are examined and actions assigned, and a monthly review that connects the numbers to the plan. The meetings must consume the measurement, or the measurement will quietly die. Dashboards nobody opens are the digital equivalent of the unread board pack.
Where to begin
Start with decisions, not data. List the ten decisions the business makes repeatedly: pricing, hiring, scheduling, purchasing, chasing. For each, ask what number, known earlier, would improve it. That list is your first dashboard. Automate its production, define each figure precisely, and put it in front of the same meeting every week.
Within a quarter, the tenor of management conversation changes: less adjudication of whose spreadsheet is right, more argument about what to do. That argument, conducted over trusted evidence, is what measurement is for.