KPI dashboard for a small business: what to track and who owns it

Pick few, and pick ones the team can move

Two tests will cut most candidate KPIs. First: can the person who owns it change it by what they do next month? Market share usually fails that test in a company of forty people; jobs quoted per week, or average days to collect, usually passes. Second: would a bad reading start an argument about the number itself? If it would, fix the definition before you put it on a wall. The examples are deliberately not software ones, because most small companies are not software companies. A services firm tracks utilisation and realisation against quoted hours. A distributor tracks stock turns and the margin it actually achieves after volume discounts. A manufacturer tracks scrap rate and on-time delivery. A clinic tracks chair or room utilisation and the proportion of bookings that no-show. In each case the KPI is close enough to the work that the owner can point at what they would do differently.

Cascade: company target, department target, individual target

A KPI that lives only at company level is a scoreboard. The cascade is what turns it into a plan: the company target breaks into department targets, and those break into targets held by named individuals. CEODash models exactly those three levels, and a target at any level carries its own period and unit rather than inheriting a vague one. One rule in that structure is worth understanding, because it is counter-intuitive the first time you see it. A manager target is held deliberately above the sum of the targets beneath it — the tree is generated so that the parent is at least ten per cent clear of its children. That is not double counting. It is the acknowledgement that some of the individual targets will miss, and that the manager owns the shortfall rather than discovering it at the end of the quarter. A parent target that exactly equals the sum of its children is a plan with no tolerance in it at all.

Every KPI needs a named owner

A target in CEODash is in one of three ownership states, and the distinction does real work. It is owned, meaning a named person holds it. It is awaiting an owner, meaning the role exists but the seat is empty. Or it is unowned, meaning nobody has been assigned at all. An unowned target is a reporting artefact. It will be measured, it will be reviewed, and when it is missed the conversation will be about the number rather than about a decision anyone could have made. Making the state explicit is what stops a quarter-end from turning into an argument about whose job it was — the answer was visible all along, including when the answer was "nobody".

When a target lands on a role you have not hired

This is the case a spreadsheet handles worst, and it is common in a growing company: the plan needs someone who is not there yet. When a target lands on a role that is missing or unfilled, CEODash opens a hiring order against it and holds the target in the awaiting-owner state rather than quietly parking it on the founder. The same thing happens in reverse when someone leaves. Their targets move back to awaiting-owner and an order is opened, instead of the numbers silently going stale on a departed person’s name — which is the usual way a KPI tree rots. For a chief executive the useful reading is the count: how much of the plan currently depends on people who do not work here yet. That number is a hiring forecast and a risk register at the same time, and it is rarely what founders assume before they look.

Reviewing without a weekly reporting ritual

The reason KPI systems die is that maintaining them becomes a job. The way out is to make the review arrive rather than be assembled: a briefing generated on a schedule from the numbers already in the system, and flags raised when something moves outside its usual range rather than a human scanning for it. What sits on which plan is set out on the pricing page, and that page is the source of truth rather than this one — plans change and a guide is a bad place to pin them. What is worth saying here is that the discipline matters more than the tier: a short review at a fixed hour each week, against targets that each have a name beside them, will outperform a richer system nobody keeps up.

Frequently asked