The app was never the prize

Africa keeps building social apps that fail, while its payment systems quietly succeed. When I went looking for why, this is what I found. First of three parts.

Why does Africa not have its own social media platforms? This question has come into my mind from time to time over the years, usually triggered by some piece of news about the social media giants that I felt disadvantaged African users. It came up again recently, set off by two pieces of information I came across. The first was a clip from a podcast where the speaker, a content creator in Nigeria, was complaining that he needed millions of views to earn what a content creator in a first-world country earns from only a few thousand. The second was a news article about Meta raising the cost of messaging on the WhatsApp Business API in South Africa. From the 1st of October 2026, every reply to a customer costs twelve South African cents (12c). This time, when the question came up, I decided to find an answer. That decision led me down a rabbit hole, and ultimately to three articles, of which this is the first.

At first the two pieces of news looked like separate complaints. One was a creator not earning enough. The other was a fee on business messages. But the more I looked, the more they seemed like the same problem wearing two faces. In both cases Africans were doing their work, and their talking, and their selling, on ground owned by someone else, and the owner sets the terms. The creator keeps a fraction of what his work is worth. The business pays to reach a customer it already has. Meta says the twelve cents reflects the value the message carries, and it does. That is exactly the point. The value is real, and it is being collected by the owner of the space, not by the people creating it.

It is worth noticing who pays that new fee. Not only small businesses. The banks. The big retailers. Even the mobile networks, MTN and Vodacom among them. The operators now pay Meta to reach their own subscribers. And it is not only a South African story. Nigeria's competition regulator has upheld a 220-million-dollar fine against Meta over how it handles people's data, and the COMESA authority, which covers twenty-one African countries, has opened its own investigation into Meta's WhatsApp Business terms. Different countries, same shape.

My first instinct was to build our own

I started where I think most people start. If a foreign platform is charging us to reach each other, then surely the answer is to build our own. An African WhatsApp. An African Facebook. A home-grown super app that keeps the value here. It seems obvious. Then I looked at what has actually happened every time someone tried, and the idea got a lot less comfortable, because the attempts are many and almost none of them worked out.

Exhibit A is MTN's ayoba, because it had everything the theory says you need. It launched in 2019, took its design from China's WeChat, and MTN made it free to use by not charging data for it. It grew to around thirty-five million users by 2023. Then, in 2026, MTN switched it off. Exhibit B, WeChat itself, came into Nigeria and South Africa with the full weight of Tencent behind it, and left after not getting the traction it hoped for. Exhibit C and D came long before either of them in the form of two home-grown apps that had already risen and fallen: Mxit in South Africa and 2go in Nigeria. Both grew huge, and both are now gone. I will come back to these two later because their demise highlights an interesting point my research uncovered.

I believe these attempts failed because they failed a simple test. People in Silicon Valley call it the "Toothbrush Test". Larry Page, one of Google's founders, used to judge a product by asking whether it was something you would use once or twice a day, like a toothbrush, and whether it made your life better. When applied to social platforms, the question is not how many people sign up. It is how many reach for it, out of real need, every single day.

In my opinion, the home-grown platforms failing the test came down to three things.

The first is that they never gave people a reason to come back every day. Getting someone to sign up is easy, especially when they are already your subscribers and you can force it in their faces. A daily habit is not. A chat app only becomes a habit when the people you actually want to talk to are all on it and using it, and most of these never got that far.

The second is that the one daily slot they were fighting for was already taken. WhatsApp passes the toothbrush test easily, and Facebook did before it, and they pass it for the same reason: everyone you know is already there. That is the thing a slightly better app cannot beat. You do not switch messaging apps on your own. To move, you would have to bring your whole family, your customers and your friends across with you, all at once, and you can't. So you stay, and the challenger dies.

The third is that many of them tried to buy the habit instead of earning it. They spent heavily to pull users in, often by making the app free to use. That buys downloads. It does not buy a habit. When the spending stopped, the users left.

Now, back to Mxit and 2go, because they are the exception that proves the rule, and they died the same death. Both of them did pass the toothbrush test. In the late 2000s Mxit was the app young South Africans opened all day, and 2go was the same for young Nigerians, both of them years before WhatsApp reached this part of the world. So why are they both gone? Not because they failed the test at the start, but because they stopped passing it. When phones moved from cheap feature phones to smartphones, neither app moved with them. They did not rebuild the product, or the infrastructure underneath it, for the new kind of phone. Once WhatsApp arrived on those smartphones it took the daily habit and never gave it back. Passing the toothbrush test once is not enough. You have to keep passing it by keeping up with developments in innovation and technology. This turned out to be an important takeaway that I will ventilate later in the series.

It was always about the community

Step back from those three reasons and one thing sits underneath all of them. The platforms that failed either never turned their users into a community, or could not hold on to the one they had. And by community I mean something specific: a group of connected people who have real reasons to interact with each other. Notice that this is two things, not one. There are the connections, the map of who is linked to whom. And there are the reasons, what each person is into, that gives them something to interact about. Hold on to that split, because it comes back later. That pair, far more than downloads or sign-ups, is what the toothbrush test is actually measuring. You reach for something every day when the people you need are on the other end of it and you have a reason to reach them.

This point about the community is important because I think it is the gateway to the whole thing. I'll go as far as saying that this is what MTN did not get with ayoba. It is why Mxit and 2go worked while they did. For a few years each of them held a real community, young South Africans and young Nigerians who had someone to talk to and a reason to talk, and while they held it they were unbeatable. When they lost that community to WhatsApp, they lost everything. Get the community and you have a platform. Miss it and no amount of free data or marketing spend will save you.

And that points to a trap MTN fell into. A list of subscribers is not a community. MTN had tens of millions of people on its network that it pushed ayoba directly to, and a community still did not form. The same is true of any organisation sitting on a big database of members. The numbers are not the thing. Unless the platform meets a real, existing need for those people to interact with each other, the connections never switch on. What you tend to get instead is closer to a media channel: an owner who has to keep producing content interesting enough to hold an audience that would otherwise wander off. That is a harder job than it looks, and it is a different job from getting people to talk to each other.

Enabling money interactions passed the test

So I turned the question around. Instead of asking why the apps failed, I asked what home-grown platforms had actually passed the toothbrush test. Something people here reach for every day, out of real need. Everyone should know the answer - it was of course the payments platforms.

Think about what a payment app does. It meets a need people already had, and felt sharply: to spend and move money without the cost and the risk of drawing out cash and carrying it around. And it meets that need everywhere. On the street, in the shopping centre, at the taxi rank, in the back of a minibus. There is a reason to open the app every day, because money changes hands every day.

If you think about it, payments are also an interaction. Money moving between people is people doing something with and for each other. The payment platforms did the thing the chat apps could not: they met a standing, everyday need for people to interact, and so people kept coming back.

The winners all ran on shared ground

But there was a second thing about the payment platforms, and it took a closer look to see it. They did not each build their own separate, sealed-off system. They all ran on shared ground. Wherever your money was saved, at whichever bank or wallet, it could be reached and spent for whatever you needed. I think this is what people mean by interoperability: your money is not trapped where it happens to sit, it can move across the whole system.

Nigeria paints a clear picture. Underneath all the fintech noise is a single shared platform that every bank and wallet plugs into. It carried around eleven billion transactions in 2024, more than double two years before. OPay, PalmPay, Moniepoint and the banks all compete on top of that one shared platform. They are rivals standing on the same ground, and none of them had to create a closed community to win.

Let me be clearer on this point - it would be easy to conclude that closed communities do not scale, this is obviously false. Facebook, WhatsApp, Instagram and Twitter all scaled as closed communities. Being closed alone is not the thing that holds you back.

The banks' problem was different. Each bank was its own closed system, and those systems did not connect to one another. Instead bank customers faced significant friction when attempting interbank transactions. Their walls were not around the whole community, the way Facebook's is. Each wall was around a single bank, cutting its customers off from everyone else's. That is fragmentation, not scale, and it left every customer stuck inside whichever bank they happened to be with. Interoperability is what pulled those walls down, and once the walls were down a payment platform could reach the whole market at once instead of one institution's account holders.

There is another insight here that's worth calling out. In addition to maintaining friction around interbank transactions, they failed to evolve their own operations and customer experience. They held on to old and expensive infrastructure, and pushed inconvenient channels on their customers, a branch to queue in, a card machine, a transfer that took days and cost an arm and a leg (to be fair, maybe a finger and a toe. Either way more than customers were willing to pay). When cheaper, faster, everywhere-you-are money arrived, the banks were the incumbents standing still, and the payment platforms walked past them. It is the same story of Mxit and 2go except on the money side. The banks had the account holders. They had the head start. What they did not do was keep passing the toothbrush test as the technology moved, and so, in Nigeria especially, the payment platforms are now eating their lunch. A big base of customers, once again, turned out not to be the same thing as meeting the need.

A final but critically important takeaway from the payment platforms in Nigeria. Their successes can be greatly attributed to how they chose to identify customers. They used the individual's mobile number. Ironically, owned or at least generated by mobile operators, but the very definition of enabling interaction and forming communities. Basically everyone has one and it's used multiple times in a day to connect people to their world. Even businesses have mobile numbers - hold on to this thought because it will come up in the next article.

So why not just build our own?

Let me come back to where I started. After all the research and thinking, my answer to "why not just build our own" is that it is not that simple. Trying to build what already exists, in the way it already exists, has failed and will keep failing. It is not that it cannot be done. It is the sheer amount of money and effort it takes to create communities, the classic chicken and egg that faces any two-sided platform. Nobody joins until their people are there, and their people do not join until everybody else has. Those who tried to buy their way across that gap could not keep it up, and eventually they failed.

That said, there is a way to build communities, and the payment platforms in Nigeria demonstrated it, overtaking the banks and taking a real foothold in local markets. Understanding how they did it is what shed light on the importance of interoperability and of building on shared platforms. From here on, let us call those shared platforms rails.

Think of a train line. The rail is the shared track laid down underneath, running between and connecting everywhere. Nobody rides directly on the rail, and no single company controls it (ignore the privatised lines for a second). What you ride in is a carriage, and many different carriages run on the same track, competing for passengers. In Nigeria's payment system the shared instant-payment track is the rail, and OPay, PalmPay, Moniepoint and the banks are the carriages competing on top of it. The rail is what everyone shares so the whole thing can reach everywhere. The carriages are where companies compete and win customers. Building a home-grown social platform, or better yet a system that lets Africans benefit more from social platforms, requires knowing what should be offered as a rail and what belongs in the competing carriages.

One caution before I go on. Payments prove the principle — a shared rail underneath, competing carriages on top, all of it built and run here. They do not prove that identity, connections and content can be run the same way. Moving money is a tidy problem next to moving people's relationships and the things they say to each other, which pull in trust and moderation in a way a payment never does. So I am taking the shape from payments, not the blueprint, and the harder parts get their due in the next two articles.

How I'd build a home-grown social ecosystem

Firstly, I would not build a single social platform any more. Instead I would build a social platform ecosystem, made up of three components, each doing one of three separate jobs, each with its own rails and carriages where they apply.

The first job is connecting people. Doing that needs three things about each person: an identity, a way of recognising them; a profile, information about their interests and activities; and their connections, the map of who they are linked to. Like the payment platforms, a mobile number makes a great starting point for that ID — the way in, even if the number itself still sits with the network. Together these make up an identity rail. The rail is built on a shared platform, owned by the individual and not by a social media company. Most importantly, it can be made available to any platform that wishes to compete for that individual's attention. This is the hard layer, and it is where the real lock-in lives. Think about your own position for a moment: the reason you cannot leave WhatsApp is not the app, it is that your connections are trapped inside it. Put each person's identity on a shared rail that belongs to them, and you have taken away the thing that has defeated every African platform so far.

The second job is moving content. This layer contains both a rail and carriages. I would place making content and storing it on the rail: shared, open tools for creating quality content, and shared storage to hold it. What this means is that whatever a person generates is theirs and under their control rather than being locked inside the ecosystems of individual apps.

Right now the good creation tools are bundled into the channels on purpose, to keep people there, and I think that is about to change, because AI is making powerful creation tools cheap and universal.

The truth is that individuals already hold their content, because it often sits on their own devices, a phone or a desktop. Holding the file is a foothold the channels rarely give you, even if it stops short of owning the rights around it. The catch is that the quantity and quality of what a person can generate is limited by the processing power and storage on the device. On this continent those are exactly the things in short supply. Shared storage and processing power, plus AI-enablement, lifts the limitation. It gives the individual room to make and keep quality content that they get to own.

On top of that shared rail sit the carriages: the social channels, old and new, competing on delivery and on discovery, on how well they get the right content in front of the right person. In plain terms, content would get created and stored on shared ground, then pushed out to whichever channels compete to show it well. Because discovery stays competitive, the channels keep having to improve to win attention, each drawing on the same shared identity rail underneath. The discovery algorithms are the channels' own IP, so keeping them competitive rather than shared and open is what keeps the incentive to innovate alive. Share them and you remove that incentive, and the individual is the one who loses out. Competitive discovery, by contrast, works in the content owner's favour — but only because identity and content are portable. A channel that cannot hold a creator's audience captive has to compete for that creator, and that is what pushes the value back to them.

The third job is perhaps my own insertion. It is making offers: letting individuals put goods and services in front of a market and complete a payment transaction. I believe this job is necessary because it creates an added incentive for the individual to be part of the social ecosystem in the first place. If being on the network is also a way to reach customers, sell, and earn, then generating content and building connections is no longer just something you do for its own sake. It pays. That gives people a reason to show up and keep showing up, which is the very thing the failed platforms could never manufacture.

This third job requires payment rails. Unlike the previous two jobs, the rails already exist and have been presented earlier. In addition to being a suitable analogy for how rails work, and proof that they can be built and operated locally, they also add significant value to the social ecosystem.

That is how I would build a home-grown social ecosystem. The identity rail and content rail currently do not exist as shared infrastructure anywhere on the continent. They only exist inside closed channels that we do not own, which is the whole reason the value keeps leaving.

It comes down to ownership

In the end this is not really about technology. It is about ownership, and the terms that come with it. The same money, the same attention, the same work can move through a system and leave almost nothing behind, or stay and build up, and the difference that matters most is who owns the rail it runs on and what they are allowed to charge you for using it.

That is why I cannot get WhatsApp's twelve-cent charge out of my head. It is small and specific, and it is a reminder that we are renting, and that the landlord can raise the rent as and when they wish. The creator in that podcast clip is paying his own version of the same rent. Everyone on these social channels is probably paying it too, without ever seeing it as a line on an invoice or bank statement. The payment rails are my proof that the way out is real, and not just a wish.

In the next article I want to look at the first rail we have not built: the identity rail, the one that would let you keep your profile, connections and identity from being locked into one or more social channels. Then, in the third, the harder question of letting the people who make a network valuable share in the value they create. Both are tougher than payments. Neither is impossible. The reason to build them already exists for many creators, and arrived for many businesses on the 1st of October 2026.

---

Sources and tools: I built the evidence for this series over three rounds of deep research using Kimi's deep-research tool, covering about thirty African platform attempts, the record on individual data ownership in richer countries, and the history of reward currencies. I graded every important figure by how solid its source is. Most rest on official data; the one exception is ayoba's peak user count, which is MTN's own reported figure. I wrote and edited the piece with Claude, and built the diagrams in matplotlib in the house style. The hero image was generated with GPT Image 2, via kie.ai.