Organization: Karibu — Provecta Group engagement Role: Go-to-Market & Revenue Operations Lead Location: Nairobi and Mombasa, Kenya — diaspora corridors in the US, UK and EU Outcome: A three-segment revenue architecture, a sequenced launch plan, and a five-layer monetization model — plus a readiness gate that stopped the company from marketing ahead of what it could actually deliver
Karibu — Swahili for welcome — is a Kenya-first property platform. It runs three connected portals: one for landlords and property managers, one for tenants, and one for agents, with an AI assistant for pricing and market intelligence sitting across them. The ambition was large and correctly stated: become the connective tissue of Kenya's residential real-estate economy — the rent rail landlords run on, the deal engine agents sell through, and the data layer banks plug into.
The product was largely built. What did not exist was a commercial answer to three questions: which customer do we sell first, what exactly do we charge for, and how do we know a segment is ready to receive demand before we spend money creating it.
Three candidate segments were on the table:
The instinct in the room was to lead with the banks. It was the biggest number, the most prestigious logo, and the pitch — sole data provider to Kenya's banks — sounded like a company worth funding.
The most attractive segment was the one that could not be sold yet.
A data API is only worth buying if the data is current, dense and predictive. Karibu's data would come from landlords collecting rent and agents closing sales on the platform. On day one there was almost none of it. Selling a bank a data product before the data exists is not an aggressive plan; it is a plan with no product in it.
This was not a theoretical risk. Okra — widely described as the Plaid for Nigeria — raised $16.5M and shut down in 2025, in significant part because the data-API-to-bank model was sold ahead of distribution. That is the same shape of mistake, in an adjacent market, inside the same decade.
The second problem was factual discipline. The pitch in circulation referenced selling to 42 banks. Kenya has roughly 39 licensed commercial banks per the Central Bank of Kenya directory; the 42 figure conflates that with Kenya Bankers Association membership, which is 46 institutions and includes non-banks. A number that is wrong in a first meeting with a credit or risk audience is not a rounding error — it is a signal about how everything else in the deck was assembled.
The third problem was appetite outrunning readiness. There was pressure to start driving demand. But payment rails were not live in production, which meant owner payouts could only book as pending — no money actually moved. There was no public marketing page; the root domain redirected into the demo portal, so there was no top-of-funnel surface at all. The WhatsApp assistant was not visible or instrumented. The portals could be reached with empty states. Marketing into that is not growth; it is buying first impressions you cannot honour.
Before writing a go-to-market plan, I ran the market-discovery work that had to come first: not how big is the market, but why is the problem still unsolved, and who is keeping it that way.
The finding was that opacity in the Kenyan land record is not a defect awaiting a technology fix — it is an asset that produces rent for an identifiable set of beneficiaries, and it is actively defended. The evidence for that is on the public record: an official state audit found that five years after the national digitisation platform launched, only 2 of 83 land registries were fully digitised. Members of the National Assembly's Public Accounts Committee stated on the parliamentary record that digitisation is frustrated internally by those who benefit from the process remaining manual. Kenya's grey-listing by the Financial Action Task Force in February 2024 cited opaque beneficial ownership and Nairobi real estate as a laundering conduit.
I wrote that up as an internal intelligence brief with every material claim tagged by evidentiary weight — official finding, court-adjudicated, credible journalism, academic analysis, or allegation — with charged individuals presumed innocent and estimates never presented as audited fact. The discipline mattered because the brief named real institutions and, in places, real people.
The commercial conclusion changed the strategy. If opacity is defended, the venture cannot win by trying to become or replace the registry — that fight is against entrenched interests and a two-decade track record of reform failure. It wins by becoming the bureau, not the registry: an independent verification layer sitting beside the official record. And the paying customer is not the state and not the incumbent — it is the diaspora buyer who has personally lost money to a forged title, because that is where the demand-pull already exists.
That single finding is what made anti-fraud the diaspora message rather than a feature bullet, and it is what made the trust layer the product rather than the marketing.
I reframed the segments from three parallel bets into one sequence, where each stage funds and feeds the next:
Landlords first. They generate the transaction volume and, therefore, the data. Shortest path to value, shortest path to a live payment event.
Agents and diaspora second. They monetize the highest-ticket events — property sales — and manufacture the trust narrative that pulls landlords in behind them, since many diaspora buyers become absentee landlords who then need management.
Banks third. The data harvest, not the seed. Groundwork — compliance, relationships, association membership — starts in month zero and runs in parallel, because SOC 2 Type II and data-protection registration are long-pole items. But no API marketing until the data and the public developer portal exist.
The sequencing rule was written down as non-negotiable, because the pull toward the prestigious segment is constant and needs something to push back against.
The temptation in a property platform is to charge agents to list or to be seen. I ruled that out. It inflates prices for the buyer — which is precisely the thing the brand promises to fix — and it monetizes the wrong side of a market where trust is the scarce good.
Instead I designed a five-layer revenue stack, deliberately additive over time and engagement-positive at every layer, with the consumer side free at the point of use permanently:
Three things were explicitly excluded and written into the document as exclusions: buyer-side fees of any kind, ever; pay-to-list for ordinary agents; and own-account buying and brokerage, which is capital-intensive and historically failure-prone.
Underneath the stack sits the economic thesis: the software is the wedge, the financial flows are the business. Embedding payments raises revenue per customer several times over. The money is in the payment throughput take, the cross-border owner-payout FX spread, and value-added services — which is roughly two-thirds of revenue at the mature comparable in this category. The core software can be cheap or free, because its job is to win density.
Landlords — lead with the reconciliation pain. The rail already exists in Kenya; M-Pesa has more than 51 million users. What does not exist is an automated way to match payment confirmations to units and ledgers, so property managers do it by hand every month. Automated reconciliation plus receipting plus payment reminders wired to the ledger is the wedge. Distribution runs through channel partners who each bring dozens to hundreds of units at once — the estate agents' registration board directory, the surveyors' institution with 6,500+ members, the developers' association with 200+ members — plus housing cooperatives, which are the single largest source of housing and land finance in Kenya at roughly 31.8%, ahead of banks. I flagged that cooperative partners must be vetted individually following the 2024–25 sector scandal.
Diaspora — lead with anti-fraud, on WhatsApp, US corridor first. The United States accounts for 51% of remittance inflows, so that corridor gets the first push. The channel is WhatsApp rather than landing pages: click-to-WhatsApp advertising has been measured at roughly 94% higher conversion and 92% lower cost per lead than landing-page equivalents, and Kenyan messaging costs around $0.022 per marketing message. Every listing carries a deep link into a conversation, a qualification bot triages, and a named human owns the relationship from there.
Banks — account-based, not lead generation. All ~39 banks are treated as one tiered named-account list, reached primarily through the Kenya Bankers Association and the sector conference circuit. The play is to sign 3–5 design-partner banks that the founder personally owns, on free or discounted access with an explicit exit ramp that converts to a paid contract. The door-opener is a scoped AI-enablement engagement rather than the API itself — a board-level easy yes that lands the relationship, then expands into the recurring data product. I capped services at 15–20% of revenue so the blended margin stays software-like and the company does not quietly become a consultancy.
Anti-fraud only works as a message if the product can actually deliver it. I specified the trust stack as a sequence of checkpoints, each of which doubles as a conversion step that moves a buyer down the funnel: verified agent status tied to real registration; a remote title search through the state e-citizen platform at KES 500; milestone-released escrow held in an independent lawyer's client account; systematic video site visits; and filmed title handovers used as social proof for the next buyer.
That last one matters commercially. Every completed handover becomes the marketing asset that closes the following deal, which is how a trust product compounds instead of resetting.
The cheapest acquisition channel turned out not to be paid media, SEO, or outbound. Every real-estate professional in Kenya operates under a permanent credibility deficit with diaspora buyers, and cannot fix it alone — their own marketing is exactly what the diaspora distrusts. A third-party verifier can fix it for them.
So verification was designed to be something members want to broadcast. Karibu does not certify skill; it certifies structural facts — registration status, no open complaints, transaction volume verifiable through platform telemetry, checkable references. The verified agent then publicises their own verification to their existing buyer book, because it neutralises their single biggest objection. The platform gets distribution it does not pay for, and the member gets credibility they could not manufacture themselves.
Signups are a vanity number in this category. I defined activation per segment as the first moment real value moves, and wrote the instrumentation to match:
Each event fires into one analytics store and feeds the funnel dashboard. The retention case for getting this right is blunt: customers who do not reach first value in month one churn at roughly three times the rate, and about 43% of small-business losses happen in the first quarter.
The stated ambition was 70–90% of single-family residences. That number is not reachable through thin national coverage; it is reachable through local saturation and tipping. So the operating model is: pick one Nairobi estate cluster with high single-family rental density and active local property managers; sign the property managers, agents and landlords who work that specific cluster; seed properties so nothing shows an empty state; drive it to the point where everyone here pays rent through Karibu; bank the profit and take the next adjacent cluster.
The governing KPI is the share of the target cluster's rental units on the platform — not national users. And the framing I put on the goal is that the gap to the second-place player in a tight geography matters more than the absolute percentage.
The discipline I am most confident about is the one that prevented activity rather than created it. I wrote a single rule into the top of the playbook:
Never spend on brand exposure faster than a segment can onboard and retain the users it attracts.
Then I audited the live platform segment by segment against a written definition of market-ready, and gave each a status. Landlords: roughly 80% ready, gated on live payment rails. Agents and diaspora: roughly 70%, gated on surfacing the WhatsApp assistant and seeding supply. Banks: not public-facing at all — the developer routes returned 404, so a prospective bank could not read documentation, request access, get sandbox keys, or make a test call unaided.
A segment does not open for marketing until its definition of market-ready is met. That converted an open-ended launch debate into a short, ordered blocker list — payment rails, public landing page, developer portal, WhatsApp instrumentation, seeded supply, compliance started — and made it obvious that the fastest route to revenue ran through fixing six things, not through spending more.
The engagement produced the commercial architecture the platform was missing:
A sequencing decision with a documented reason. Landlords, then agents and diaspora, then banks — with the counter-example that justifies it named explicitly, so the decision survives the next person who wants to lead with the bank logo.
A revenue model that does not tax the buyer. Five layers, consumer-free permanently, monetizing the payment rail, the FX spread, verification and data rather than listing fees.
An account-based bank motion with an honest number in it. ~39 licensed commercial banks per the central bank directory, 3–5 design partners, services capped at 15–20% of revenue, a 12–18 month cycle assumed rather than wished away, and compliance started in month zero because it is the long pole.
Activation and retention defined as measurable events, with the instrumentation specified per segment and the density KPI — share of the beachhead cluster — installed as the number that governs the land-grab.
A readiness gate that throttled spend to capability, with a six-item blocker list that made the critical path visible.
A 90-day plan and a channel map covering payment rails, public surface, brand system, the first estate cluster, the first property-management partners, the first verified agents, the diaspora advertising pilot, and the bank groundwork running in parallel.
The work that mattered most was not the plan. It was the two refusals.
The first was refusing to sell the bank segment first, when that was the segment everyone wanted to sell. The argument that won was not opinion — it was a named company, in an adjacent African market, that raised $16.5M and died doing exactly that.
The second was refusing to start marketing before the payment rails were live. The gap between the product works and the product can take money and give it back out is invisible from the inside and total from the outside. A landlord who signs up, adds a property, and then watches an owner payout sit as pending does not come back — and no amount of acquisition spend recovers that.
Both refusals cost the engagement momentum in the short term. Both were the reason there was anything to accelerate later.
The discovery research is the piece I would point to first, though. It is easy to look at a broken market and see a technology gap. The harder and more useful question is why is it still broken — and when the answer is that someone profits from the breakage and defends it, the entire strategy changes. It moved the company off a fight it could not win, onto a position it could hold, and told it exactly which customer was already motivated enough to pay.