A number that predicts, and a number that confirms

What a KPI is

A key performance indicator is a measure chosen because it tracks something the strategy depends on — key because a company can watch only a handful of numbers closely, performance because it moves when people act, indicator because it stands for a goal that is larger than itself. Two classifications organise every KPI on the exam.

Financial or non-financial. A financial measure is stated in dollars or derived from the statements: gross margin, revenue growth, ROA. A non-financial measure is a count, a rate or a time: pick accuracy, training hours, complaints per 1,000 orders.

Leading or lagging. A lagging measure confirms a result after it has happened; a leading measure moves first and predicts the result. Financial measures are almost always lagging — ROA for 2025 tells you about 2025 once the year is closed. Non-financial measures tend to lead: a rise in complaints this month predicts the repeat-purchase rate next quarter and the revenue after that. The trap is to assume non-financial means lagging because it is “softer”; it is the reverse.

Eight Tamarack measures, classified

Measure                              Financial?   Leading / lagging
Gross margin %                       financial    lagging
Revenue growth %                     financial    lagging
ROA                                  financial    lagging
Warehouse pick accuracy              non-fin.     leading  (→ complaints, returns)
Staff training hours                 non-fin.     leading  (→ service quality)
Website load time                    non-fin.     leading  (→ online conversion)
Complaints per 1,000 orders          non-fin.     leading  (→ repeat purchases)
Net promoter score                   non-fin.     leading  (→ revenue growth)

The six characteristics of a good measure

Not every measurable thing is worth measuring. A good KPI is:

1. Aligned with the strategy — it moves when the strategy is working.
2. Controllable by the person judged on it — they can move it by what they do.
3. Timely — available soon enough to act on.
4. Reliable and verifiable — two people measuring it get the same number.
5. Cost-effective to collect — the data is worth more than the cost of gathering it.
6. Hard to game — the easiest way to improve the number is to do the job well.

Testing “sales per employee”

Tamarack proposes to pay store managers a bonus on sales per employee. Run it through the six. Aligned? Partly — it rewards revenue, but says nothing about the fitting and service the strategy rests on. Controllable? A manager controls staffing and, less directly, sales. Timely? Monthly, yes. Reliable? Yes, both numbers come from the systems. Cost-effective? Free. Hard to game? No. The fastest way to raise sales per employee is to cut part-time staff: the numerator barely moves in the short run and the denominator falls at once. The store gets emptier, the fittings get shorter, the repeat rate falls — and the KPI reports success while the strategy fails. The measure is also lagging and financial only. The fix is a pairing: sales per employee and a customer-satisfaction floor, or contribution per store, so that gaming one number costs the other.

The habit to build is the sixth characteristic as a question: how would I game this? Ask it before adopting any measure, because a measure changes behaviour whether or not you intended it to.