The monthly review starts well. Sales hit the revenue target. Operations hit on-time dispatch. Procurement brought purchasing costs down. Customer service is answering faster, and finance followed up more overdue accounts than last month.
Every manager's scorecard is green. Meanwhile margins are thinner, cash is tighter than you planned, customers are complaining about deliveries, and the managers are blaming each other for it. The business feels harder to run than it did three months ago.
When that happens, it's worth looking at the KPIs before you look at the people. Sometimes the employee has done exactly what the scorecard asked, and the scorecard asked for the wrong thing.
A KPI changes how people work
Once you hold someone to a number, they pay attention to it. They prioritise it, defend it and make trade-offs around it. So a KPI does two jobs at once. It measures the work, and it shapes the work.
Economists have a name for one side of this. In 1975 the British economist Charles Goodhart observed that a statistical relationship tends to break down once it's used as a target for control. In 1997 the anthropologist Marilyn Strathern summed up the general idea in the line most people now quote: when a measure becomes a target, it ceases to be a good measure.
At the extreme, this ends in scandal. In 2016 the US Consumer Financial Protection Bureau fined Wells Fargo $100 million after finding that employees had opened roughly 1.5 million deposit accounts that may not have been authorised by customers, to hit sales targets and earn bonuses. Few businesses will see anything like that. The everyday version is quieter: good people get very efficient at the wrong thing.
What a sales target can do
Take a distribution business in Mombasa whose sales manager has one target: KSh 12 million in sales a month. (This is an illustrative example, not a client.)
At month-end the figure is KSh 12.7 million. Green. Then finance looks closer. Some of the sales went on long credit to customers who already owed money. Several orders were discounted heavily to close before the 30th. Operations paid overtime to meet delivery dates sales had promised without checking capacity. Procurement bought stock urgently at higher prices to fill the orders.
Revenue went up. Margin, cash and the customer experience all got worse.
Was the sales manager wrong? He did what the scorecard told him mattered. What the business actually wanted was revenue it could deliver at a margin and collect on time. The target only described part of that.
Every KPI carries an assumption: if this person improves this number, something the business cares about will improve too. Assumptions can be wrong. If someone hits their KPI and the business result doesn't move, it may be the measure that needs to change.
Too far from the person, or too close
Individual KPIs tend to go wrong in two directions.
Some are too far from the person's control. Tell an operations officer to grow company revenue by 20% and you've given them something important that they can't do much about. They don't set prices, find customers or run the sales team.
Others are too close to the activity. Tell customer service to close 90% of complaints within 24 hours and they can control it completely. They may also start closing tickets to protect the number when the customer's problem hasn't been fixed.
The most useful measures sit in between. They track something the person can genuinely influence, with a believable link to the result the business wants. For the customer service team, that might be complaints that stay resolved, checked by how many customers come back with the same issue within a month.
Ask how the number could be hit badly
Before a KPI goes live, ask one uncomfortable question: how could a capable person hit this number and leave the business worse off?
If procurement is asked to cut purchasing costs by 8%, one way to get there is to switch to cheaper suppliers who deliver late or send poor stock. So the cost target needs a second measure beside it, such as on-time delivery from suppliers or the share of stock rejected on arrival.
The person who'll work to the KPI is usually the best one to ask. Sit down with them and say: "If you wanted to hit this number without actually helping the business, how would you do it?" You're testing the measure before they have to live with it. Most people will find the weak points quickly, and it shows them you're interested in the result behind the number.
When departments' targets pull against each other
A KPI can make perfect sense inside one department and still cause damage across the business.
Say sales wins a large, urgent order. Sales wants it shipped to protect revenue. Operations wants stock bought today to protect delivery times. Procurement won't use the fast supplier because it costs more and would hurt the savings target. Finance wants the order held because the customer has an overdue balance.
Each manager is being reasonable by their own scorecard. Between them they want four different decisions, and in many founder-led businesses the founder ends up settling it. If that happens every week, the KPIs are sending unresolved decisions to your desk.
The fix is usually to give departments a shared definition of a good sale. If sales is measured on revenue collected within agreed credit terms, sales and finance are working toward the same thing. Where one department depends on another, measure the handover too. If operations can't dispatch because materials arrive late, procurement can be measured on materials delivered to the floor on time and to specification. Then a miss can be traced to where it actually happened.
Questions to ask before a number goes on a scorecard
- If this number improves, why should the business care?
- Does this person have the authority and resources to move it?
- Will it tell us something early enough to act, or only confirm a problem after it's happened?
- How could someone hit it while missing the real objective? If that's easy, add a second measure.
- What other department or behaviour could it damage?
If a KPI fails one of these, a tougher target won't help. Change the measure.
Why good KPIs drift
Choosing better KPIs is only half the job. In many businesses the KPIs are set at the start of the year and nobody checks whether they're still driving the right behaviour. Managers report the number each month without discussing what's behind it. When two targets collide, the decision goes to the founder, and the targets stay as they were.
What makes KPIs work is the routine around them: expectations agreed with each person, monthly conversations about the result and how it was reached, and managers who raise a bad measure early. In our work with founder-led businesses, that routine is often what's missing, even where the scorecards look well designed. Talentos helps founders build a way of managing performance where this happens as part of normal management, without the founder refereeing it. The Performance Picture takes ten working days. We look at how performance is actually managed across your business, from leadership and staff perspectives and from your records, and show you where it breaks down and what to fix first. You can start with the free three-minute Quick Picture.
Some roles don't have a natural number at all. Our guide on measuring staff performance in roles without clear KPIs covers those. If your problem is deciding what a red month on the scorecard means, read how to read department KPIs.