Thirty Days After the Signature

I have watched a CEO sign a partnership agreement with real enthusiasm — the handshake, the photo, the LinkedIn post — and then watched that same partnership produce nothing for four months. Not because the partner backed out. Not because the market rejected the offer. Because the person standing behind the counter who would actually have to say something to a customer never received a single sentence telling her what to say.

Picture a pharmacy in Surulere that signs on with a telehealth provider. The owner is enthusiastic on the call, sees the logic immediately, signs the paperwork inside a week. Three weeks later a field officer visits to check progress and asks the attendant working the counter what happens when a customer wants to use the service. The attendant looks blank. Nobody told her anything happens at all. The signature exists in a drawer somewhere, and the counter runs exactly as it did before the ink dried.

This is not a story about a lazy pharmacy owner. It is the default outcome of how partnerships get built, and understanding why requires taking apart a word that hides more than it reveals.

The word doing too much work

"Partnership" in Nigerian healthcare distribution covers at least four distinct arrangements, and most agreements never specify which one they are. In a referral arrangement, Partner A tells its customers about Provider B, and B does the clinical work. In a distribution arrangement, Partner A's physical footprint becomes a channel through which B's product reaches customers, with A doing real operational lifting rather than merely pointing customers toward B. A procurement arrangement runs the other direction: B supplies something A needs to run its own business, and the money flows from A to B. Co-selling asks both parties to bring something to a joint offer and to sell it actively, which means both need a reason to keep showing up.

Two people can sit across a table, both nodding, both using the word "partnership," and be describing two of these four things without either noticing the mismatch. The MOU gets drafted in language elastic enough to cover both readings, because elastic language is what lawyers reach for when the business side hasn't done the harder work of choosing. Everyone leaves the room satisfied. Nobody has actually agreed to anything a receptionist could act on.

The test I have come to trust is brutally simple: can you write one sentence, in plain language, naming the actual transaction? Not the relationship — the transaction. Something like: a customer enrolled by Partner A receives a consultation at Clinic B within seven days, recorded against a shared referral code. If you cannot produce that sentence, you have not built a partnership. You have built an understanding, which is a different and much cheaper thing to hold, and a much more expensive thing to discover you were holding when the revenue doesn't arrive.

Five ways the sentence goes unwritten

Even when the sentence gets written, it dies in one of five fairly predictable ways, and I want to be honest that I resisted this list for a while. My first instinct, running a field team of fifty-plus people signing partners across pharmacy, retail, fitness, wellness and workplace channels, was to treat every stalled partnership as a single failure with a single cause — usually whichever cause was easiest to blame that week. Bad partner, bad timing, bad luck. It took watching enough of these die to notice they were not one failure wearing different faces. They were five separate and mostly independent points of collapse, and a partnership that survived four of them could still be killed by the fifth.

The first is ownerlessness. The CEO signs, the CEO believes the deal is done, and the CEO moves on to the next signature — because signing feels like progress and closing meetings feels like output. Nobody hands the receptionist or the field officer a script. The person who actually touches the customer at the moment of decision was never in the room where the deal got made, and information about what was agreed rarely survives the trip down the org chart intact, if it survives at all.

The second is bad timing. An introduction to a new service has to land where a customer already naturally pauses — during registration, at checkout, at a staff town hall, at the moment someone is already holding out a form. Bolt the invitation onto a moment that doesn't already carry a pause in it, and it competes with everything else demanding the customer's attention at a moment they were trying to leave, not linger.

The third is an offer nobody can restate. Ask five staff at a partner location what the customer actually gets, and if you receive five different answers, you have an offer that exists in a slide deck and nowhere else. A customer will not act on an offer the person offering it can't describe in one sentence, because hesitation from the person selling reads as a warning, not a nudge.

The fourth is broken attribution, and this one kills partnerships slowly rather than quickly, which makes it more dangerous. A partner introduces ten customers over a month. None of it gets tracked back to the partner in any visible way. From where the partner sits, they made an effort and got no proof it did anything. There is no reason to repeat behavior that appears, from the inside, to have vanished into a void. The revenue may in fact be flowing — but if the partner can't see their fingerprints on it, they stop investing energy in producing more of it, and the channel starves from disuse rather than from failure.

The fifth is unresolved risk around data. A partner half-wonders whether sharing information about their customers or employees creates a liability they haven't thought through, and rather than ask and risk an awkward conversation, they just quietly do less. Under the Nigeria Data Protection Act, health-related information sits in the sensitive personal data category, carrying tighter handling obligations than an ordinary customer list — the NDPC has been explicit that any health-related detail about a person, including something as specific as a disability status, gets treated as sensitive because it is health-related, with real discrimination risk if it's mishandled or fed into anything algorithmic. If nobody on either side of the partnership has actually worked out who is allowed to hold what, the safest move for a nervous partner is to do nothing, and nothing is what you get.

What actually needs to exist

None of these five failure modes require a lawyer to fix. They require a document nobody usually bothers to write: a short operating sheet, a page, attached to the legal agreement rather than buried inside it, that survives the first thirty days.

It names an owner on each side — not a department, a person, with a phone number. It states the channel in one line: where, physically or digitally, does the introduction happen. It sets a weekly target measured in introductions made, not revenue booked, because revenue lags introductions by weeks and a target you can't hit for a month is a target nobody tracks in week one, when tracking would still catch the failure while it's cheap to fix. It defines the attribution fields — what code, what form field, what identifier travels with the customer so both sides can later agree on what happened. And it sets a review date, something as unglamorous as fourteen or thirty days out, at which someone actually looks at whether any of this happened.

This is not a complicated document. It is complicated only in the sense that writing it forces two organizations to admit, in specific and falsifiable language, what they actually agreed to — which is precisely the discomfort that elastic MOU language exists to postpone.

There's a version of this argument that says the operating sheet is really a management tool dressed up as a partnership tool, and that version is half right. It is a management tool. But the reason it belongs stapled to the legal agreement rather than living in an internal tracker somewhere is that the partner needs to see it too. A partner who has never been shown the weekly target, the review date, or the attribution field has no way to distinguish a channel that is quietly working from one that has already gone silent. Sharing the operating sheet is what turns "we'll check in sometime" into a relationship where both sides know, on a specific date, whether the thing they signed is alive.

It's worth adding here that Nigeria's health financing landscape makes the "who does what" question sharper than it looks on paper. The National Health Insurance Authority Act of 2022 set up several distinct coverage tracks — the Formal Sector Health Insurance Programme, the Organised Private Sector programme, coverage for vulnerable groups through the Basic Healthcare Provision Fund, and GIFSHIP for the informal sector and groups outside formal employment, alongside private plans. A partnership that assumes "the customer pays" or "the employer pays" without specifying which financing track applies is assuming away exactly the kind of ambiguity that stalls a transaction at the front desk, when the person at the counter discovers the code doesn't map to a scheme anyone actually recognizes.

Back to the counter

Go back to that attendant in Surulere. The fix for her situation is a card taped near the register with four lines on it: what to say, what code to write down, who to call if something doesn't work, and a date circled on a calendar when someone checks whether any of it happened. That card is worth more than the signature that preceded it, because the signature only proves two organizations agreed to try. The card is the only evidence, thirty days later, that either of them actually did.

Notes on sources

• National Health Insurance Authority Act, 2022 and its programme categories (Formal Sector, Organised Private Sector, Vulnerable Group via the Basic Healthcare Provision Fund, GIFSHIP, private plans): nhia.gov.ng/nhia-act.
• Health-related data, including disability status, classified as sensitive personal data under the Nigeria Data Protection Act 2023: NDPC, "Disability Details are Sensitive Health Data".