The Nigerian startup ecosystem map

We are building a public map of the organisations building Nigeria’s startup ecosystem. Who funds, who teaches, who gives you a desk, who writes cheques, city by city. Free to use, free to be listed in, and maintained by the organisations themselves rather than by us.

This note is how we got there: the question that started it, the five things we considered building, what already exists, and what would make us stop.

Where it started

Last month we put out a technical partnership offer: we rebuild an ecosystem organisation’s website for nothing, and in exchange we get access, referrals and our name on the work. It got more response in a fortnight than anything else we have published.

Which raised a better question than the one the offer answers. A website is a small thing. What could we build for this ecosystem that it cannot otherwise get?

That is a partnerships question before it is a product question, and our partnership page already commits us to an answer: we give real value first, and we take access rather than money. So the honest version of the question is what value we could give that nobody else in the room is positioned to give.

What we found first

Reading these organisations’ sites properly, rather than glancing at them, the same shape kept appearing. They convene real people and move real money, and almost none of them can produce evidence of either.

Not through negligence. Their value is a graph, and nobody holds the graph anywhere. Who they know, who they convened, who they introduced to whom, which founder got which cheque after which room. It lives in spreadsheets, WhatsApp threads and one person’s memory.

I recognised it immediately, because it is our own story. Two thousand people in our community, and when a founder asked me for a frontend developer I could vouch for about twenty. The talent existed. The index did not. I have written about that here before.

Every one of these organisations is in that position about its own network. Most have not noticed yet.

Five things we could build

Working from that, five candidates came out, and it is worth listing all of them because four are still alive and only one became this.

An index that scores each city against a published model of how ecosystems develop, so a bottleneck is locatable rather than argued about.

A convening record. These organisations run hundreds of events and nothing records who was in the room. A check-in at the door and one question afterwards turns “we connected forty million dollars” from a claim into a trace.

A registry of the organisations themselves, with the connections recorded. Not a list, a graph: this hub introduced that founder to that fund.

Matching gaps to the people who could close them. Once you have the first and the third, you can answer “this city has talent forming and no local capital, and here are the four organisations positioned to help”.

A view for funders, because nobody deploying money into Nigeria has per-city data, and the people with budgets are the ones who would pay for it.

The convening record and the registry are the ones I still want. The funder view is where the money probably is. But none of them was a product yet, because every one of them required somebody to sit down and do data entry as a favour to us. That is the same wall our own contact register keeps hitting.

The thing that turned it

One network publishes a proper framework for all this, and it is better than anything I would have written. It sorts the ecosystem into six kinds of enabler: capital, policy, talent, media, programmes and place.

Read that list again. Nobody in it builds anything.

Talent, in that model, means supplying developers and designers to startups. It does not mean the organisations that design, build and operate the systems the ecosystem itself runs on. There is no category for that. Which is why a fund can deploy a million dollars through a Google Form and nobody experiences it as strange: in the sector’s own map of itself, the technical half does not exist.

That is the whole Senpai argument, drawn by somebody else, in their own diagram, without us in it.

And then the reframe that made it a product. Stop trying to collect the data. Let the organisations add themselves.

The moment the ecosystem is doing the recording, the registry, the gaps and the city view all fall out of it for free. The census is the diagnostic. A city with six co-working spaces, three programmes and no capital shows you that without anybody answering a subjective question.

It is not a list, it is a set of questions

Collecting the data is not the point. Being able to ask it things is the point, and it is the part I have been describing badly.

These are the questions it has to answer, each as a filter rather than a search:

• Who is funding companies in Abuja?
• How many active programmes are running in Lagos right now?
• Who serves women founders in Kano?
• Which organisations run open programmes rather than invite-only ones?
• Which cities have nobody working on capital at all?
• Who should a founder in Kaduna talk to first?
• Which organisations work in more than one city?

None of those is answerable anywhere today, and none of them is answerable from free text. That is why the form is mostly fixed choices rather than boxes: seven categories, fourteen problems, ten communities, four ways of engaging, and a plain statement of whether an organisation writes cheques, connects you to capital, or neither.

It is also why the taxonomy has to be right before the data arrives. Adding an option later is cheap. Renaming one after two hundred organisations have picked it turns every historical answer into a small lie.

The paragraph each organisation writes in its own words is what makes the page worth reading. The fixed fields are what make the map worth using.

What already exists, because we are not first

Before building anything, the honest question is whether this has been done. It mostly has.

AfriLabs runs the largest network of innovation hubs in Africa, with hundreds of member hubs across dozens of countries, and it has profile pages. It already has a page for at least one of the Abuja hubs I was researching. Briter, formerly Briter Bridges, publishes an Africa innovation ecosystem map and sells market intelligence to funders and researchers. GSMA has mapped active tech hubs across the continent with them. Startup Genome ranks ecosystems globally.

These are serious efforts by serious people and I am not going to pretend otherwise.

What none of them is, as far as I can tell:

• Free and open. Briter is a commercial intelligence platform. AfriLabs is a membership network, so you are on it if you joined it, which is a different thing from a census.
• All of the ecosystem. Hub directories cover hubs, which is one of six categories. Not the funders, the policy bodies, the media, the builders.
• City-level and current. Annual reports and continental maps do not tell a founder in Uyo who to talk to this week.
• Maintained by the organisations themselves, which is the only way a record like this stays true after the launch.

That is the gap. It is narrower than “nobody has mapped this”, and I would rather state it narrowly and be right.

And if one of them already does this well enough, we should not build it. That is a real possibility and I would rather find out in a fortnight than after a year.

The four versions we rejected

One: score every city. Take that seven-phase model, write forty questions, publish a number per city. I liked this for about a day. It fails because nobody owns a city, so there is nobody to answer, and the people best placed to answer have an interest in their city looking active. Worse, it produces a league table, and a league table of Nigerian cities insults eleven of them to flatter Lagos.

Two: research it ourselves. Credible, and it does not scale past a few cities. It also makes the record ours, which means it dies the moment we get busy.

Three: build the shared platform. One system that every hub configures for its own applications and cohorts. This is genuinely the big idea and it is still there in five years. It needs a trust position we have not earned, and a density we do not have. Announcing it now would be a press release.

Four: write the report. Fast, and we would have had it out in a week. But a report is read once and then it is last year’s PDF, which is the exact failure mode we keep pointing at in other people’s footers.

What survived, and why

Let the ecosystem register itself.

Every register we have built this year has failed the same way: each entry costs the person entering, so nothing gets entered. Our own contact register is the same story.

This one reverses it, and not through goodwill, which is not a plan. Through search.

Most of these organisations have a slow site with no structured data. A fast, properly marked-up page about them, on a maintained domain, can plausibly rank for their own name. That is free distribution, and it makes being listed worth three minutes instead of being a favour they do us.

The diagnostic then falls out of the census for nothing. If organisations register by city and category, the gaps compute themselves: a city with six co-working spaces, three programmes and no capital shows you that without anybody answering a question about phases. Counted, never scored. No rankings.

The constraints, stated in advance

Capacity. One person builds and runs everything here. Every feature has to survive that, which is why there are no accounts, no messaging and no feed.

The cold start. A directory with eleven entries looks abandoned. It has to launch already populated, which means hand-seeding sixty organisations before anyone sees it. If sixty cannot be assembled, that result is the finding and we stop.

Decay. A listed hub that closed two years ago is worse than an absent one. Every entry carries a date it was last checked and an annual re-confirmation. If that lapses, this becomes the thing it was built to replace.

The conflict of interest, which is real. We sell services to the organisations in this directory. We are adding a seventh category, builders, that we happen to occupy. We run the thing. All three facts go on the about page in plain words, because a directory that hides who runs it deserves the suspicion it gets.

What it is actually for

This is the part I nearly left out, and it is the reason to build it rather than one of the other four.

Strengthen the ecosystem, rather than rank it. Our partnership page says we give real value first and take access rather than money, and this is that at a scale one relationship at a time cannot reach. Nobody has to sign anything or partner with us to benefit from it.

Make everybody’s work more credible, including ours. Right now an organisation here can say it has run forty programmes and there is no way for anyone to check, which means the ones telling the truth get no more credit than the ones rounding up. A dated public record, maintained by the organisations themselves, is worth most to the people already doing the work properly. That is the whole argument of our systems index turned outward: a number somebody can check beats a claim on a homepage, and it beats it for everyone, not just for us.

Make access easier. A founder in a city that is not Lagos currently finds help through who they happen to know. That is the quietest way an ecosystem stays small, and it is the thing this fixes most directly. If the map does nothing else, it should let somebody in Uyo find, in a minute, who can actually help them this month.

And show the gap honestly, including where it flatters us. I expect the builder category to be close to empty nationwide. If it is, that is worth far more as a number people can check than as an argument on our own site. If it is not, and there are twenty technical teams serving this ecosystem that I have not found, then I was wrong in public and the map is how I find out.

Much later, if it earns it, a record that travels between organisations, so a founder vetted in one place arrives credible in another. Not a plan. A direction.

Where it actually is, and what is next

Seven organisations, three cities, seeded by hand. Not a product.

I have also built the interface twice and got it wrong twice. The first was a table of counts with a headline announcing what we had proved, which is a report wearing a product’s clothes. The second was a map, which is the right shape and still assembled out of my defaults rather than designed.

What I keep doing is building the thing that proves my point instead of the thing somebody would use. The test written down for the next attempt: would a hub in Uyo link to their page from their own website? If not, it is not finished, whatever I think of the map.

The next goal is deliberately small. A live platform, on its own domain, carrying the first seeded set, with a form somebody can use to claim their entry or submit a new organisation. That is it. Not the funder view, not the convening record, not the graph.

A great deal about how this works at scale is still unresolved, and I would rather say so than pretend otherwise. Who checks a submission when there is one of me. What happens when a listed organisation turns out to be bad. How a record stays true in year three rather than becoming the stale directory this was built to replace. How the taxonomy survives two hundred organisations disagreeing with it. None of those are answered, and some of them cannot be answered until there is enough in it to break.

So there will be more notes about this before anybody fills in a form, and possibly before anybody sees it. The next one is likely about the design, which is the part I have most clearly got wrong so far.