Marisol Transfers rebuilt bidding around first send and cut cost a third
Paid Growth
A worked example, not a client result. The organisation, the figures and the quotation below show how an engagement of this shape is structured and what it sets out to move. They are not outcomes achieved for a named client.
Sahel TransferCROPaid Growth
How Sahel Transfer recovered 18 points of KYC completion in 11 weeks
Document capture was failing on older Android devices and nobody owned the number. Fixing it changed what every unit of acquisition spend was worth.
+18ptskyc completion
−34%cost per first send
+18ptsKYC COMPLETION
−34%COST PER FIRST SEND
2.1xREPEAT SEND RATE
11 wksTO FIRST RESULT
Measured over 11 weeks against a matched prior period. Attribution and confidence notes available on request.
The situation
Sahel Transfer had spent eighteen months building a clean, fast app for the UK to Nigeria corridor, and by most measures it was working. Installs were growing month over month, cost per install had come down twice, and the paid team had a dashboard that looked healthy in every meeting.
Revenue was flat. Not falling, which would have triggered alarm, but flat while spend rose, which triggered arguments instead. The growth team argued the funnel was leaking. The product team argued acquisition was buying the wrong people. Neither could prove it, because the analytics stopped at signup.
By the time they came to us they had a specific and unhelpful theory: that Nigerian senders in the UK were simply harder to verify than the market average. It was the kind of explanation that ends investigation rather than starting it.
What we found
The first thing we did was split the funnel by device tier, which had never been done. On devices released within three years, verification completion sat at a respectable rate. On older Android handsets, which represented a large share of their actual sender base, it collapsed at the document capture screen.
The cause was mechanical rather than regulatory. The capture component required a file size and image quality that older cameras produced inconsistently, and the failure state returned a generic error and sent the sender back to the start of the flow with no explanation of what had gone wrong.
Two further findings compounded it. Manual review at the ninetieth percentile took just over four days with no status communication of any kind, so senders who had done everything correctly assumed they had been rejected. And the rejection messages themselves had been written for an auditor rather than for a person holding a phone.
Funnel and diagnostic table
| Stage or area | Position | Cumulative or detail | Read |
|---|---|---|---|
| Account created | 100% | 100% | Baseline |
| Verification started | 74% | 74% | Intent held up |
| Documents submitted | 51% | 38% | The leak, device-driven |
| Liveness passed | 86% | 33% | Lighting and older cameras |
| Approved | 91% | 30% | Manual review delay |
| First transfer completed | 58% | 17% | Momentum lost in the wait |
What changed
1. Instrumented the eight funnel events properly, split by corridor, device tier and document type, so the number had an owner and a weekly review.
2. Rebuilt document capture with progressive compression, on-device guidance and a retry that returns to the failed step rather than to the start.
3. Rewrote every rejection message in plain language, naming the specific problem and the specific fix.
4. Added a pending-verification status experience with a stated expectation, which changed nothing about review speed and a great deal about abandonment.
5. Fed post-KYC and first-send events back to Google and Meta, so bidding stopped optimising toward people who would never complete.
The sequence mattered. Had we fed post-KYC events back to the ad platforms before fixing capture, the algorithms would have learned to avoid exactly the sender segment the business depends on. Measurement first, then the funnel, then the spend.
Cost per install rose by roughly a fifth after the bidding change, which required explaining internally more than once. Cost per completed sender fell by a third over the same period.
Cost per install went up and everyone panicked for a week. Then the transfer numbers came in and nobody mentioned installs again.
— Amara Bello, Head of Growth, Sahel Transfer
How the work ran
Event map rebuilt around first send
Eight events instrumented, split three ways, replacing a two-event setup that stopped at signup.
Document capture flow, before and after
Progressive compression, on-device guidance, and a retry that returns to the failed step.
Bidding moved from install to first transfer
Post-KYC signals fed back to both platforms once the funnel could support the traffic.
What moved
| Metric | Before | After | Change |
|---|---|---|---|
| KYC completion rate | 30% | 48% | +18pts |
| Document submission rate | 51% | 79% | +28pts |
| Cost per first completed send | Baseline | −34% | Improved |
| Cost per install | Baseline | +19% | Deliberate |
| Repeat send rate, 90-day cohort | Baseline | 2.1x | Improved |
| Support contacts about verification | Baseline | −41% | Improved |
Mandatory line beneath: “Measured over 11 weeks against a matched prior period. Attribution and confidence notes available on request.”
The engagement in brief
Scope and shape
- Digital-first remittance app
- UK → Nigeria
- 11 weeks
- CRO
- Paid Growth
They found three things in the first two weeks that we had been staring at for six months. The device split alone would have paid for the engagement.
— Amara Bello, Head of Growth, Sahel Transfer
Same funnel, different corridor?
A fixed-fee growth audit covering acquisition, KYC completion, retention and tracking across your corridors. You keep the roadmap whether or not we work together.