What to Actually Measure

The measurement industry's quiet assumption is that more visibility is better — more dashboards, more metrics, more real-time everything. For a practice, the assumption runs exactly backwards: every number you track costs attention, attention is the scarcest resource you have, and a dashboard nobody reads is a vanity artifact in furniture form. The winning move is radical smallness — few enough numbers that they actually get looked at, real enough that looking changes what you do.

The index card — five numbers, monthly:

1. Enquiries — how many new people asked about working with you.
2. Booked share — what fraction of them became actual work.
3. Source — where each new client actually came from, in their words.
4. Days-to-paid — the average gap from invoice to money.
5. Repeat & referral share — what fraction of this month's work came from existing clients or their introductions.

Written down monthly, these five read the health of the entire chain — and a practice that keeps them honestly knows more about itself than most companies with analytics teams.

Why these five

Each one is a gauge on a different section of the system, which is the whole design: enquiries reads the demand side (found, trusted) in one number; booked share reads the conversion chain — demand arriving but not booking points squarely downstream; source is the cheapest attribution system that works and the input for every where-should-effort-go decision; days-to-paid reads the money chain; and repeat-and-referral share is the flywheel gauge — the one number that says whether the business is compounding or refilling a leaking bucket at full price.

When one gauge moves, the cluster that owns it tells you what to inspect. That's the index card's real function: it doesn't diagnose — it tells you where to look, which is the constraint question in numeric form.

Keeping it honestly

The card fails one way: by becoming aspirational. Defenses:

Define each number once, plainly. What counts as an enquiry (a real approach from a potential new client — not newsletters, not vendors)? When does the clock start on days-to-paid? Write the definitions on the card itself; drift in definitions is how numbers stop meaning anything.

Pull from where the truth already lives. Enquiries and sources from the record; days-to-paid from the invoice tool; repeat share from the client list. If assembling the card takes over fifteen minutes, the problem is disconnection, not measurement.

Same time, every month, written down. The card's power is the row above — this month against the last six. A number without its history is a mood; the trend is the instrument. One spreadsheet tab, twelve rows a year, kept for the life of the practice.

Read it with one question: which number would I most want to move next month? — and then notice you've just chosen where the month's improvement effort goes. That's the diagnose-improve-measure loop running on fifteen minutes of arithmetic.

Questions practices actually ask

Only five? What about profit, utilisation, my pipeline value...? Your accounting already owns profit; utilisation matters for capacity planning; add a sixth gauge if your practice's shape truly demands it. The constraint is the card staying readable at a glance — every addition taxes all the others.

Weekly or monthly? Monthly, for reading; the underlying habits (states, the source question) run daily. Weekly reads of small numbers produce noise-chasing, which is worse than not measuring.

Do I need software for this? You need the habits feeding it and one spreadsheet. Dashboards that auto-assemble the card are a pleasant luxury after the numbers are defined and trusted — never a substitute for them.

What are good benchmark values? Your own last six months. Cross-practice benchmarks mislead more than they inform — the card's job isn't to grade you against strangers, it's to catch your system changing.

---

Part of Measure What Matters — the instrument panel.