Walk into any accelerator cohort in Kampala and ask the room what they measure, and the answers arrive quickly and at length: followers, footfall, orders, likes, deliveries, sometimes a revenue figure recalled to the nearest hundred thousand. Ask a second question, what decision changes if that number moves, and the room goes quiet. That silence is the actual problem, and it is not a numeracy problem. These founders can calculate. What they lack is the link between a measurement and a choice, which is the only thing that turns data into management, and its absence explains why so much reporting effort produces so little operating difference. A metric with no attached decision is a chore. It gets collected because someone was told to collect it, reviewed because a meeting exists, and abandoned the moment the week gets hard, which is precisely when it would have earned its keep.
Key Takeaways
- The test for any metric: name the decision it would change and the threshold at which it changes. A metric that fails this test is a chore, not a measurement.
- Founders are not data-poor. Most hold more information than they use, in WhatsApp order histories and mobile-money ledgers they never read as data.
- Thresholds must be set before the number arrives. Deciding what counts as bad after seeing the result is how a dashboard becomes decoration.
- Five numbers reviewed weekly beat thirty reviewed never, and the five-number discipline is a cadence question before it is an analytics question.
- Human capital is the asset most firms never measure and most often waste, and it is measurable with the crudest instruments.
- Self-reported founder data has a known bias direction: confident guesses beat honest uncertainty in the telling, and are worse in the deciding.
What makes a metric worth collecting?
An attached decision, stated in advance, with a number on it. “We track repeat purchase rate” is a chore. “If repeat purchase falls below forty percent for two consecutive months, we stop the new-customer spend and call the last thirty buyers” is a metric, because it names the threshold, the action, and the window. The discipline is not statistical, it is grammatical: every number in the review deserves an if-then sentence written beside it before the number is ever seen.
Writing the threshold in advance is the part founders skip, and skipping it is what makes dashboards decorative. If the acceptable value is decided after the result arrives, the result can never be bad, because the standard moves to meet it. Every founder does this and almost none notice, which is why the practice has to be structural rather than intentional: the threshold goes into the sheet in the same session that creates the row.
The corollary is that most firms should measure far less. A five-line weekly review with thresholds beats a thirty-line monthly report without them, and the cadence carries more of the benefit than the sophistication does.
Why do founders think they have no data?
Because they are looking for a dashboard and their data is in a phone. The Ugandan founder who says she has no numbers has a WhatsApp Business account with every order she has taken, a mobile-money statement with every payment received and its date, an airtime record, a supplier chat thread with prices in it, and a delivery rider she pays per trip. That is a transaction log, a customer list, a price history and a variable-cost record, sitting unread because none of it looks like a spreadsheet.
The unlock is almost never a tool. It is one hour spent transcribing thirty transactions into six columns, which produces the average order value, the repeat rate, the busiest day, the true delivery cost, and the two customers who account for a suspicious share of revenue. Those five facts change decisions immediately and cost nothing but attention. The founder who reaches for accounting software before doing that hour will have an unused subscription and the same fog.
The frontier-market caution matters here too. Analytics advice written for firms with card payments and web funnels assumes instrumentation that does not exist, and its prescriptions land as guilt. The instrumentation that does exist is the mobile-money rail and the chat thread, and it is genuinely rich. The task is reading it, not replacing it.
What is the most under-measured asset?
The people. Firms count stock and cash and almost never count the thing that determines whether either is handled well, and the measures are cruder and more available than founders assume: who can run a shift unsupervised, how long a new hire takes to reach that point, which tasks only the founder can do, and what proportion of the week those tasks consume. Four numbers, none of them requiring a system, all of them predicting whether the business can grow without breaking the owner.
The last of those is the one to start with. A founder who is the only person able to price a job, approve a discount, or close a sale has built a firm with a single point of failure, and the percentage of the week consumed by founder-only tasks is the most honest capacity metric a small firm has. It also has an attached decision, which is what makes it a real metric: when the figure crosses a threshold, the next hire is not a salesperson, it is whoever removes the bottleneck, and the delegation work becomes the growth work.
FAQ
How many metrics should a small firm track?
Five, reviewed weekly, each with a threshold written before the first reading. Add a sixth only by removing one.
What if the data is messy or incomplete?
Use it. A rough average from thirty transcribed transactions beats an exact figure that does not exist. Precision is worth less than the decision it enables.
Is this an argument against accounting software?
No, an argument against buying it first. Do the manual hour, learn which six columns you actually consult, then automate those. Tools bought before that hour tend to encode the fog.
Does self-reported data have a bias?
A predictable one. In assessments, founders who write “I do not know yet, here is what I would measure first” are usually closer to the truth than founders who supply confident round numbers, and they make better decisions later.
Related Reading
- The five-number dashboard for the small firm
- Founder operating cadence: the weekly rhythm that holds
- SOPs and delegation in a low-formality business
- Who actually finishes: what 99 alumni say about selection
Sources and Evidence
- Blacksmith, N. and McCusker, M. E., Data-Driven Decision Making in Entrepreneurship: Tools for Maximizing Human Capital, on decision-linked measurement and human capital as a measurable asset.
- Internal capability-assessment corpus, 2026, in which numerate but hedged profitability answers proved more diagnostic than confident round figures.
- Baig, P., Montoya-Martinez, E. et al., Entrepreneurial Ecosystems: Resource Acquisition, Flow and Management, on absorptive capacity and why resources without measurement discipline do not convert.