It is unusual for a senior practitioner to describe their own venture as a failure. Most polish the language. A learning experience. A pivot. A sunset.
Geoffrey Coombs — 15 years in emerging-markets finance, Oxford-trained in African Studies, and a French DFI Proparco alumnus — does not. « I think it’s perhaps more interesting and a little more straightforward to say. It was a failure. It didn’t work in the end. »
The candor is the doorway. In the third episode of Transformation Compass, the podcast of ISC Paris’s DBA program, host Nabil Ghantous invites Coombs to walk through the origin and the unraveling of his own entrepreneurial project in Ghana. The lessons that emerge are not Africa-specific. They are useful for anyone leading or evaluating ventures in environments where the local logic is not the imported logic.
The project started where most cross-border ventures start: with a visible inefficiency.
Coombs noticed that artisanal fish in Ghana — caught by small fishermen, not industrial trawlers — passed through three or four hands before reaching the consumer. The price tripled or quadrupled along the way. Traceability was lost. Quality degraded.
The scale was not marginal. Coombs estimates that around 10% of Ghana’s population is directly or indirectly affected by artisanal fishing activity — making this anything but a niche.
The proposed solution looked elegant. Connect producers to consumers directly. Use the post-iPhone technological revolution to reduce friction. Share the captured efficiency between fishermen and end buyers.
« The idea was to connect the beginning of the chain directly with the end of the chain, address some of these issues of traceability, and hopefully share the economic benefit of reducing the inefficiency in the chain, both in terms of lower end prices for the end consumer and a higher initial payment for the fisherman. »
The model was not naïve. Similar approaches had succeeded in other verticals — bananas, coffee. The macro narrative was tail-wind: Africa rising. Emerging middle-class consumer. The seed had been planted.
The execution is where the lessons begin.
The first error was structural. Coombs and his team underestimated how long it would take for the model to take root and for the local market to adapt around it.
« One thing that I think we maybe underestimated when we were working on our project was the length of time that is required for new ideas to take root, to iron out difficulties, and potentially look at ways of nudging towards behavioral change. »
He references a friend currently building a food-transformation venture in Central Africa. The friend has built a full year of runway just to figure out the downstream distribution problem before launching.
« I think that’s a really smart thing for him to be doing. »
The implicit critique of his own earlier project is honest: he did not give himself that year.
The lesson generalizes. In any environment where data is thin and behaviors are less documented, the timeline of a project is not the timeline of a spreadsheet. It is the timeline of a relationship — and relationships are not assets that can be compressed with capital. Promising too much too early to investors and partners — what Coombs calls « counterparties that may be kind of waiting over your shoulder for brilliance to be delivered » — is one of the most reliable ways to corner a founder into a bad decision.
The second lesson is the one Coombs treats with the most care, because it is the one that took him longest to see.
In the early phase of the project, the team observed what looked like irrational behavior. A fisherman on the beach would sell his catch to a small local buyer at a suboptimal price, even when a larger buyer was visibly available with cash in hand for the entire load.
Coombs’s first response was the response of every outsider: this is illogical. It must be that no one has come in with enough capital, or the right technology, to fix it.
The actual explanation was different.
« There is a collective sense of responsibility for one another. If that person who took a little bit of produce isn’t able to sell, they don’t have economic activity. So that person’s going to be in a very difficult spot. The person selling recognizes that they need to be supportive of their linkages around them. »
The behavior was not irrational. It was rational under a different set of weighted variables — variables that the imported business model had not seen.
« Trying to force behavioral change for your elegant new solution may not always be the right way to go. You might need to think about what are the dynamics here that are really at play. Take the time to understand that, which is almost like a sort of sociology project in itself, and that may get you to better solutions in the end. »
The point is not Africa-specific. It is universal. Every cross-border venture meets local behavior that looks suboptimal until the social fabric beneath it is mapped. The investors and operators who succeed are the ones who treat that mapping as part of the diligence — not as a footnote to it.
What unites the two lessons is a single executive discipline. Coombs frames it modestly when Nabil asks him how an outsider develops local fluency.
« A degree of openness to new experience. Knowing where my strengths are, but also where my weaknesses are. Try to really listen and call upon external advice when I can see that there’s something that I’m not getting. »
It sounds like generic advice. It is not. It is a specific cognitive posture with practical consequences: do not pre-design the solution before mapping the actual decision logic of the people the solution is meant to serve.
The venture in Ghana failed in part because the design preceded the listening. The friend in Central Africa is succeeding so far in part because the order has been reversed — a year of listening, then design.
The reversal is what most cross-border M&A misses, and what every senior leader operating outside their home certainty has to learn either before, or the hard way. It is also the kind of evidence-before-conclusion discipline that research training installs.
The full episode goes further — into the hierarchy of risks across macro, sector and project levels; into the data paucity of less-studied environments; and into why « African risk » is the wrong frame for real investment decisions on the continent.
For senior leaders weighing ventures, partnerships, or acquisitions in markets that do not yet read like home, this conversation is a useful corrective. The most expensive mistakes in cross-border strategy are usually made before the model is built — in the moment when the designer assumes the local logic is the imported logic.
