The Metric That Was Measuring the Wrong Thing

The starting condition

By the fourth week of building Dokitami's field team, the numbers looked like momentum. A fast-growing count of signed partner locations across Lagos. Around 30 field officers active, with more in onboarding. A distribution network of flyers and banners stretched across five hubs. Every internal report showed growth, because every internal report was counting the same thing: how many stores had said yes.

Revenue was not following. Signed partners were not producing referrals, and referrals were not producing orders. Officers were making dozens of visits a day, signing real businesses, and doing exactly what they had been told to do. The problem was upstream of effort. The department had built an entire operating rhythm, targets, reporting, a leaderboard, pay structure, around a number that had no reliable relationship to money coming in.

This had been flagged before it became a crisis. In the very first week of planning the department, before a single field officer had been hired, the recommendation on record was that the department should be measured by activated locations that had produced a verified order, not by signings, specifically to avoid optimising for the wrong metric. The warning existed. It did not change behaviour, because nothing in the daily reporting cadence forced anyone to act on it. A target of new stores per week is easy to report against every evening. A target of verified revenue is harder to report against daily, especially early, when the number is often zero. The easier metric won by default, for weeks, until the gap between it and revenue became too large to explain away.

The question that mattered

The question that eventually got asked was not how to get more stores. It was what the department was actually worth, in money, and whether it could justify what it cost. A field team that size, on full pay, has a real monthly cost, and that cost was running well ahead of anything the sales side had produced. The question forced a second one: if store count is not the number that matters, what is the smallest, most honest number that is, and how do you rebuild pay and targets around it without simply stalling the whole operation.

What was tried, and what the evidence showed

Before the pivot, there had already been a smaller, earlier version of the same correction. In the second week of field operations, an internal leaderboard ranking officers by lead count was quietly removed, with the reasoning stated plainly at the time: the only sound metric was orders and payment, because a lead count is not useful unless it leads to something. That was a partial fix. It corrected the leaderboard but not the underlying targets, which still ran on visits and signings. The department kept scaling on the metric it already knew was wrong, because replacing it meant admitting that weeks of reported "progress" hadn't produced anything a bank account would recognise.

A week before the full pivot, a warning was raised again: hiring should pause until the team had actually finished putting materials in the locations already signed, rather than continuing to add headcount against revenue that had not moved. Recruitment paused. It did not solve the underlying measurement problem; it only slowed the rate at which the problem was getting bigger.

The evidence that forced the actual rebuild was simple and undeniable: three weeks and thousands of visits had produced a large signed-partner base and almost nothing behind it. No amount of reframing made that look like a functioning commercial channel. The diagnosis, once someone sat with it honestly, was that the whole chain between a signed partner and a paid customer had never been built. Partners weren't referring anyone, because nothing in the system asked them to or tracked whether they had. Store count had been standing in for a channel that didn't yet exist.

The system built

The rebuild had two parts, decided and communicated within about a day of each other.

The first was operational: instead of continuing to sign new businesses, the existing team was redirected entirely toward the partners already on the books. The field team was cut to under 25 officers and expansion was paused. Every remaining officer's job for the following weeks became calling the partners they had already signed, asking directly whether they knew anyone who might want the product, rather than knocking on new doors. Every referral was tied to a partner code sent over WhatsApp, so that a lead and a sale could be traced back to a specific partner and a specific officer without ambiguity. The instruction that had existed on paper since before the department was hired — that a location should only count once it had produced a verified order — finally became the thing people were measured against day to day, not just the thing written in a planning document.

The second part was structural: pay was rebuilt around sales rather than activity, replacing a fixed structure tied to visits and signings with a new contract that tied earnings to what an officer actually closed. Anyone who wanted to remain on the team had to sign the new terms for the following month; everyone else was offboarded. The same restructuring that made the department affordable to run also made it possible to expand it again, into a second city, because for the first time growth wasn't automatically growth in cost.

What changed, honestly stated

The most immediate, measurable change came from the first full day of the new calling motion: officers working their existing partner list generated 135 potential customer contacts in a single day, including 65 direct referrals from partners who had been signed for weeks and had never referred anyone until they were asked to, on the record, by phone. That is not proof the whole department was fixed. It is proof that the constraint had never been partner supply. The company already had a large signed base to work with; nothing in the system had ever asked a signed partner to actually produce a customer, or measured whether they had.

What the record does not show, because it ends before the following month started, is whether the new pay terms held up once officers who had signed under the old, easier terms actually worked under them for a full cycle, or what the retention and output looked like once the transition settled. That is a real limit on how much credit this pivot can honestly claim, and it is worth stating rather than smoothing over.

What this generalises to

A growth number that is easy to report daily will out-compete an accurate number that is hard to report daily, unless something in the operating rhythm forces the accurate number to be reported anyway — and that has to be a structural decision, not a one-time correction. Removing a leaderboard fixes a leaderboard. It does not fix a compensation structure still built on the same flawed premise underneath it. The two have to be rebuilt together, because leaving pay tied to the old metric while quietly discouraging people from chasing it is the fastest way to produce reports that look right and revenue that doesn't move.

The other generalisable point is about where the constraint actually was. It is tempting, watching a channel underperform, to assume the fix is more of whatever built the channel in the first place — more visits, more signings, more officers. Here the actual constraint was on the other side of the funnel entirely: a large asset, the signed partner base, sitting unused because no one had ever asked it to produce anything and measured whether it had. Redirecting existing capacity toward an asset already owned turned out to be faster and cheaper than acquiring more of the thing that wasn't the bottleneck.