Your compliance officer had a quiet quarter. The KRA returns went in, the county permits were renewed, and no inspector turned up with a problem. So what did they achieve?
Your personal assistant had a good month too, as far as you can tell. Your diary mostly worked. You walked into meetings with the papers you needed. Two clashes were sorted out before you heard about them. How would you put that on an appraisal form?
Growing businesses run into this once they have people whose work doesn't end in a sale or a finished unit. Some roles create value through judgement, or by preventing problems, or by making other people's work run better. Their best month may look like nothing happened. When the appraisal form asks for a score, the temptation is to find something countable and measure that.
Counting activity gives you numbers, but little else
To measure the PA, a manager might count emails answered, meetings scheduled or tasks closed. Now there's data. It doesn't tell you much. A PA could schedule 180 meetings in a month and still let your time fill up with conversations that didn't need you. They could answer every message quickly and miss the three that mattered.
Ratings have the same weakness. A manager scores someone 4 out of 5 for communication, 3 for initiative and 3.5 for strategic thinking, and the sheet works out an average of 3.5. It looks exact. Then ask why initiative was a 3 and not a 4, and which situations the score is based on. Often there's no clear answer. The opinion is still there. It's just been given a decimal point.
So the useful question changes. "Can I put a number on this?" matters less than "Does this evidence tell me whether the role is working?"
The person who prevents the fire gets less credit
Picture two operations managers at an agribusiness supplier in Eldoret. (This is an illustrative example, not a client.)
On Thursday afternoon the first one discovers that the wrong stock has been loaded for a large order to a supermarket chain. The buyer is angry. He finds a truck, reloads it, and the delivery arrives at 8pm. The founder thanks him in the WhatsApp group for real ownership.
The second manager noticed on Tuesday that the stock allocation for her order was wrong. She fixed it, confirmed the transport on Wednesday, and Thursday passed without anyone noticing. Nobody thanked her.
Rescues are visible. Prevention mostly isn't. If your appraisals reward what people remember, you may end up teaching staff that the dramatic recovery counts and the quiet Tuesday check doesn't. The fair test is to trace the problem back. A manager who saves a customer after a supplier collapses overnight deserves the credit. A manager who works late to fix a loading error he made himself on Tuesday has a Tuesday problem to discuss.
Where the value of a role shows up
Different roles show their value in different places, so they need different kinds of evidence.
Some roles produce a piece of work: an analyst's report, a designer's artwork, an engineer's drawing. The mistake is to count volume. The evidence is whether the work was accurate, on time, and good enough for someone to make a decision with.
Some roles make other people more effective: a PA, IT support, an office administrator. Counting their tasks tells you little. Look at whether important work moves more smoothly because they're there. Do clashes get caught early? Do small matters get handled before they reach the founder?
Some roles protect the business: compliance, credit control, financial control. A quiet month could be skill or luck. Check whether the required controls were actually carried out, whether problems were raised early, and whether known risks sat untouched until someone else spotted them.
Some roles are mostly judgement: managers and senior specialists. Counting decisions is meaningless. Look at how they reached them. Did they get the facts, weigh the trade-offs and consult the people affected? Did they bring a recommendation, or pass the problem up to you?
Judgement can be specific
Many founders assume that if a measure isn't a number, it must be vague. It doesn't have to be.
Telling a finance manager "I expect good judgement" is vague. Agreeing in advance how an unusual payment request should be handled isn't. Say a supplier emails asking you to pay this month's invoice to a new bank account or a new M-Pesa till. You might agree that the finance manager will:
- Confirm the request by calling the supplier on a number already on file, not one in the email.
- Work out what it costs to delay the payment and what it risks to make it.
- Check with whoever manages that supplier.
- Bring you a clear recommendation before any money moves.
When a request like that comes in, you don't score "problem-solving" out of five. You check what they did against what was agreed. That's a judgement, but it's one another manager could follow.
Look at ordinary weeks and difficult moments
For these roles, look at two kinds of evidence across the year, so that one rescue or one mistake doesn't decide the rating.
The first is ordinary work. Does it arrive when promised? Are handovers clean? Do colleagues have to keep chasing this person? A brilliant response to one emergency doesn't make up for friction every week.
The second is the moments that test someone: a supplier fails, a deadline moves, a senior person makes an unusual request. Ordinary weeks tell you whether the work is reliable. Those moments show how the person thinks under pressure.
Questions to ask before you write a measure
Before you invent a number for a hard-to-measure role, work through these:
- What should be different in the business because this role exists? If you can't answer this, the role itself needs defining first.
- Where does the value show up: a piece of work, other people's effectiveness, protection, or judgement?
- What evidence would already exist if the role were working? Reports, a log of decisions, fewer matters escalated to you. Use what the work produces where you can.
- Which situations would show whether this person can really do the job?
- Could another sensible manager look at the same evidence and see why you rated them as you did?
Our article on job descriptions and role scorecards covers the first question in more detail.
Where it usually slips
The standards above only help if they're agreed before the work happens, and if someone notes what happened against them during the year. Without that, the appraisal comes round and the manager rates from memory, and memory favours the rescue.
That's the part we often find missing in founder-led businesses. The founder knows what a good PA or a good compliance officer looks like. The managers mean to write things down. Nobody has made it part of how the business runs, so the evidence isn't there when it's needed. Talentos works with founders to build that: clear expectations for each role, short monthly check-ins with notes, and managers who keep it going without being chased. The Performance Picture is a ten-working-day assessment of how performance is actually managed across your business, drawing on leadership and staff perspectives and your records, and it shows where things break down and what to fix first. If you'd like a quick first view, try the free three-minute Quick Picture.
For roles that do have numbers, our piece on why teams can hit their KPIs and the business still struggles looks at how to choose them.
This article gives general information on Kenyan employment law and performance management. It isn't legal advice for a specific case.