Castles in the Air

An operator with three cars doesn't really need to understand his margin. At that size he is the whole system. He eats the small losses himself, drives when a driver fails to show, knows which of his cars is burning too much fuel, and the business survives on vibes and tiny margins with no prospects for scale.

The same man with fifty cars is losing money he can't see. A fleet carries its value in tiny bits, daily, hundreds of small cash settlements that all have to be funneled properly. At three vehicles he is the funnel. At fifty he cannot walk the lot fast enough, and what he needs by then is throughput, management capacity at every turn, and some way of expanding both while still running a tight ship.

So capital arrives looking for the operator who already runs at scale, with systems to inspect and financials worth underwriting. In most emerging markets that operator has never existed. Even globally the successful fleet and rental businesses come down to a handful of companies, and those are markets where the capital and the operating history have both been sitting there for decades. Here there is no audited record, and no owner who can state his own unit economics with much confidence. So how are you supposed to choose your winners? You end up choosing them before the market has produced any.

The vector

Bogotá is where I had to actually do that. Our kind of business wasn't really known there when we showed up, and what existed instead was something older and more disjointed. A driver took his licence from one owner and his car from somewhere else, and then worked under a rotation that decided which days of the week he could be on the road at all.

The licence is called a cupo. Bogotá caps them at 50,007, and they trade between owners at prices nobody publishes. Both the transport ministry and the city's mobility office hold that the word has no standing in Colombian regulation, so no rule creates a cupo and none governs what it costs.¹ It is worth real money, it is capped in supply, and legally it does not exist, which is more or less what the informal stack looks like everywhere. Then there is pico y placa, which takes two plate digits off the road each day, Monday to Saturday, from half past five in the morning until nine at night.² So every taxi in the city sits idle one working day out of every week.

The temptation when you arrive somewhere like that is to try to change the kind of business being done. That was never the search. Nobody in that system would have called what they were doing strategy, and it's really just what you improvise when the arrangement you inherited stops fitting what you want to build. So we went looking for whoever was already at the far edge of stretching the system as it stood, and for the younger entrepreneurs pushing at that same edge. Some were reworking the payout to licence owners, some were pushing on driver financing, some were aggregating cupos, or aggregating cars to pool resources for scale.

The hunch, and this part is fairly personal to me, was that you can nudge someone much further when he is already pointed in the vector direction you see the opportunity in, and where your technology can do the most to expand what he is already doing. Call it the vector: which way he is already moving, rather than how big he is today. So you back whatever is already moving, with technology and capital and operating support, working backwards from the future you think is coming.

Never one bet

The first one we found that way was one of the oldest taxi companies in the country, with new blood in management pointed at building something large. They had already written their own app, already started down a direct rent-to-own path, and already held a big share of the city's licences. All of that was there before we ever walked in, which to be honest was a lucky streak. They were ready to co-create with us while keeping their identity distinct and sharpening their own operations.

But it's always important to foster some competition, and one partner on his own is a single point of failure for the whole market and not just for the deal. So alongside them we backed some very pumped young entrepreneurs working bikes, pushing the limits on driver financing, on collections, and on giving drivers a good enough daily experience that they stayed.

Most markets won't hand you an incumbent with new blood like that. For those I tend to use a formula of four hustlers, two outsiders, and two adjacent business people. The hustlers give you founder mode and they learn quickest. The outsiders have their own reason to win, and they bring contrarian ideas about building operations from the ground up, the kind nobody inside the industry would think to try. The adjacent ones bring the acumen of proper business structure, which the other six mostly don't have.

Eight isn't really about insurance. They all have different origin stories, so they come at the business differently, and they learn off each other watching what the others try. The market gets experimented on eight ways at once, which pushes the rate of experimentation up and keeps any single failed experiment from ever getting proportionally large.

The trust layer

Everyone will tell you the outsiders are the slot to cut first. The most disciplined fleet operation I've ever seen came out of that slot, an all-female-led transport operator with instincts about management slightly more unorthodox than a typical hustler's.

She focused on what actually mattered to the drivers on her fleet. Most of them were in their forties, and beyond the job they cared about their families and their kids, so her whole employee management was built around helping them get better life outcomes, financially and relationally, for their families. It started with helping them save a portion of their income, then helping them schedule their shifts better so they could be around when it mattered. She then essentially used the float from those same savings to create a credit system the drivers could tap into.

Imagine you drive for her. You put aside a portion of what you earn each week, she holds it, and when something lands that you cannot cover, you borrow against it instead of going to a lender. She built that float herself, out of the drivers' own money, in a market where no float is expected at all. Very quickly the trust layer was insanely robust, and everyone wanted to become a driver on her fleet. Only then could she put in the quality control, the expectations and the standards that every other operator tries to impose in week one. She never really had to impose the discipline. She had more drivers wanting in than she had places, and the reputation she had built with them did most of the work for her.

The two ceilings

And then she stopped. Three hundred cars, and not because she was beaten. She was comfortable. She had all she needed, and for an emerging market it was an outcome she was terribly satisfied with.

That is the first of the two ceilings. Scale here gets limited at a level of comfort long before it gets limited at a level of discipline, and I think that's because comfort is worth a great deal more to you when you started with much less. It is the same ladder in a soft-currency economy, where every rung you climb counts for more than it would somewhere else.

None of it was for want of a method. On a personal level I believe most of what she did could have been written down and repeated. Savings can be opted into, details on the kids can be collected, school fees automated out of the float. It just takes an operator who wants that scale and is willing to stay longer in the arena to play that game. What she ran out of was appetite, and appetite is the one thing you can't put into someone with capital.

The other ceiling is competence, and it breaks in one fairly specific place. Management span stops being able to sustain daily operations, and the teams are not able to train the next crop of managers. It is less that the owner runs out of ability and more that the business cannot produce its next set of managers.

Of the eight, seven worked.³ The one that did not was a hustler who failed in transport and then made it later in a completely unrelated field, which for portfolio building is not a bad outcome. So selection is more or less a solved problem. Getting past both ceilings isn't. Of the seven, one did.

The one who broke through had appetite from the start and was willing to do the work of stretching. For him that meant setting up operations in a new country the moment he felt confident enough in the first one, and then doing it again. He was exercising that muscle faster and faster, so the feedback loops got shorter and he learned quicker for it, and more and more people went through his school of management until the school itself got codified into a repeatable playbook. That is what carried him past the second ceiling. He did not get better at managing more people himself. He built a business that produced its own managers.

Castles in the air

So the question of which operator is allowed to grow turns into a question about reading appetite in a room, before anything has happened to prove it.

The one who broke through had it from the beginning, and you could hear it. He already had a story for global expansion from day one, even though he didn't yet know how to build it. I call it building castles in the air, and then rolling your sleeves to do the work as if you never thought of a castle in the air. You need both halves. Someone with only the story is a dreamer who never actually builds anything. Someone with only the work is a very good operator whose story is only as big as the life he already has, which is roughly what every ceiling in the portfolio looked like.

With no track record, talk is mostly what you have, so you end up reading different things from talk pointed in different directions. The forward story you check for size, whether it is larger than anything this person could currently build. The backward story you check for detail, because the sleeves are in the stories of what he has built before. The more detail he can get into about a business he has already run, the numbers, the specific mess, the month something went wrong, the more credible he is as a builder and not just air puffing. Anyone can pass the first one. The second is harder to fake if you haven't actually been there.

Within proximity

The room tells you one more thing, which is how he is building his current team. If you can't look at each of them and think this one could just as well be a founder, the team will not go far. Problems in emerging markets come at you from outside and inside at once, and to move fast you need people who think on their feet and make decisions like the business belongs to them, with alignment that is implicit and a great deal of accountability behind it.

You live with the risk that they can walk, but the hallmark of a great operator is that he can convince these otherwise founder-type people to build a large story with him instead of building their own individual small ones. I've seen one break in Accra. I think what happened along the line was management falling out with the founder, and the founder not having the grace to seed those managers into their new venture, which could have kept the compounding within proximity. Instead it turned into a rat race to the bottom.

Seeding is investing in them, or giving them equity in operations, or both, to build an adjacency that is large. The split itself is survivable. What does the damage is refusing to finance it. Compounding concentrates in a place, and I think it does much the same in a network of people, so you keep them within proximity.

Which leaves the accounting. Eight bets, seven working businesses, one operator past the ceiling. What you buy with all of it is a higher ceiling rather than a scaled operator. Each of those seven ended up running something several times larger than anything they had ever actually run before, and then stopped, satisfied, at a number that would have looked impossible to them at the start. Their castles were real. They were just built to the size each of them was already comfortable with.