Calling the Network Dokitami Had Already Signed

The starting condition

By late August, Dokitami's partnerships department had signed a large number of businesses across Lagos: pharmacies, gyms, salons, supermarkets, estates, workplaces. Field officers had visited them, explained the offer, registered them on the portal, and in many cases delivered banners and flyers. On paper, that is a distribution network: points of contact with customers who might want a telehealth weight-loss service, each one theoretically ready to mention it to the right person.

In practice, the network was almost entirely silent. Officers kept signing new locations because that was the target they were given and the number that was easy to report every evening. Nobody had built a mechanism that asked an already-signed partner to actually produce a referral, or that would have surfaced it if they weren't. A partner could be visited, signed, equipped with materials, and photographed for evidence, and still never once tell a customer about the service — and the reporting system had no way to distinguish that partner from one that was working.

The question that mattered

The obvious move, three weeks into a slow-starting channel, is to sign faster and harder: more officers, more visits, more locations. That instinct was wrong here, and abandoning it took a specific piece of evidence to force: three weeks of acquisition, and a large signed-partner base, had produced almost nothing in the way of paying customers. Adding more signed partners to a pipeline that wasn't converting the ones it already had would only have made the same problem bigger and more expensive.

The real question was narrower: could the existing partner base be made to produce customer leads at all, without adding a single new signature, just by asking directly instead of waiting passively for referrals that weren't coming? That is a different question from "are these good partners" — it assumes the partners are fine and tests whether the missing piece was ever the ask.

What was tried, and what the evidence showed

The decision, made the evening before it took effect, was total: stop new business acquisition entirely and redirect the whole field team, cut down from around 30 officers to under 25, toward one job — call every partner they had personally signed and ask whether the business knew anyone who might be interested in the service. No new visits. No new signings. The stated reasoning at the time was explicit: partner referrals, follow-up, sales, and tracking were not connected to each other, leads were being lost to a poor handoff, and signed partners simply weren't using whatever tools they'd been given to submit a referral. Rather than build a better tool first, the department tested a simpler hypothesis: that the handoff problem was partly a tool problem and partly a motivation problem, and that a phone call closes the motivation gap faster than a portal redesign does.

The mechanics were kept close to what officers already had. Calls were run in blocks through the morning, using ordinary phone calls or WhatsApp calls when a line was hard to reach, with a follow-up text if a call didn't land. Every officer logged how many partners they attempted, how many they actually reached, and how many produced a usable lead — three separate numbers, not one blended count, because attempted and reached and lead are different facts and collapsing them hides which stage is actually failing. Every lead that came out of a call carried the partner's code, so a name and a phone number could be traced back to the specific business that produced it rather than landing in the system unattributed.

It surfaced a real friction the department hadn't had to deal with when the interaction was a one-time signing conversation: partners were reluctant to hand over a customer's phone number without that customer's knowledge. Some wanted to ask first and call back. Some didn't want to be identified as the source of the referral at all. Some would only offer up people who had already asked them about weight-loss treatment, not their ordinary customers. None of that showed up when a business was simply agreeing, in principle, to refer people someday. It only showed up the moment the department asked for an actual name on an actual day, which is exactly why waiting for referrals to arrive passively had never worked: the practical, human friction in producing one was invisible until someone forced the moment where it had to happen.

The system built

What came out of that one day was a specific, repeatable motion rather than a one-off experiment: a scheduled morning calling block, run against the existing signed-partner list rather than a cold-call list; a required three-number report per officer — attempted, reached, leads — logged the same day; a partner code attached to every lead at the point of capture, not reconstructed afterward; and an explicit instruction to get the referred person's consent before their number was submitted, both to protect the business relationship and to avoid handing the sales team a contact who didn't know they'd be called.

The threshold for what counted as a "lead" was also tightened in the same motion. A number with no context wasn't a lead. A name, a number, the partner it came from, and some signal of actual interest was. That distinction mattered downstream, because the whole point of the exercise was to feed a sales team that would immediately try to convert these contacts, and a sales team working a list padded with uninterested numbers converts worse than one working a shorter, real one.

What changed, honestly stated

Against three weeks of new-business acquisition that had produced almost nothing, the first single day of calling the existing partner base produced 135 potential customer contacts, including 65 that came directly from a partner referral rather than a cold approach. The gap between those two numbers is the whole finding: the constraint had never been how many businesses were willing to sign up. It was that being signed up and being asked to actually do something were two entirely different states, and the department had never built a mechanism that moved a partner from one to the other.

What the day did not resolve, and what should not be overstated, is conversion. A hundred and thirty-five contacts is not a hundred and thirty-five paying customers, and the record from that day shows real resistance from partners around consent and privacy that a single successful call doesn't erase — it manages it, case by case, and some partners will keep saying no. The method also depends on an officer having a real relationship with the partner they're calling; it is not obviously repeatable at a scale where officers are calling businesses they signed weeks ago and barely remember.

What this generalises to

A partner agreeing to refer customers is not the same fact as a partner referring customers, and a system that only measures the first will look healthy for as long as nobody checks the second. The fix here wasn't a better portal or a bigger incentive; it was a person picking up the phone and asking, on a specific day, for a specific name. That is a cheap, low-technology intervention, and it worked precisely because the bottleneck had never been technical.

The broader lesson is about where to look when a channel with real reach isn't producing. It is tempting to conclude the audience is wrong or the offer is weak. Here, redirecting the exact same team toward the exact same partners, with nothing added except a direct ask and a way to track the answer, outproduced three weeks of expansion. Before adding more of anything, it is worth testing whether what's already been built has actually been asked to work.