The Audience You Already Own

Somewhere in Soweto this morning, a man opened his telco app, bought airtime, and closed it again. The whole thing took about forty seconds. He will do it again in ten days.

That is the entire relationship. Forty seconds, three times a month, for a company that spent real money acquiring him and spends more every month keeping him.

In the same month, that man will make somewhere north of twenty commercial decisions. Where to buy maize meal. Which pharmacy. Whether the trip across town to the bigger shop is worth the fare. Which of the two hardware stores near the junction has what he needs today.

None of those decisions touch his telco. All of them are worth more than the airtime.

The premise, in two sentences

You already pay for the audience. You monetise one thing they do with it.

Everything else they do — deciding where to shop, what to buy, where to walk, who to pay — happens somewhere else, on somebody else’s surface. That gap is not a marketing problem. It is the largest unexploited asset on the balance sheet of almost every large-audience business in an emerging market.

I have watched this fail from the inside

I wrote the original VodaPay thesis. I have spent a decade in emerging-market commerce, and I have watched the same mistake get made in four different shapes.

Ayoba reached north of thirty-five million monthly active users and was switched off this year. Not because the product was bad. Because there was no payment platform attached to the chat client — no mechanism to commercialise the engagement it had genuinely built. It created attention it could not convert.

M-Pesa in South Africa ran for six years and closed with a rounding error of active users. It was used for money-send, and almost nothing else. The thing that made M-Pesa work in Kenya was never the send — it was transactional use, every day, for ordinary things. South Africa got the wallet without the reason to open it.

VodaPay today is close to a pure e-commerce play. And I do not believe pure e-commerce works in emerging markets — not at the basket sizes that actually exist. When fulfilment costs more than the average order, no amount of product quality saves you.

The pattern underneath all three: each one built reach, or built a wallet, and none of them built the thing in between. A reason to transact. Locally. Repeatedly. For something the customer was going to buy anyway.

What it looks like when somebody gets it right

Kaspi was a conventional Kazakh bank. In 2014 it opened a marketplace to any merchant with a phone — no website, no integration project, no technical skill required. A listing and a location.

The numbers it reports now are not banking numbers:

Two out of three monthly users open the app every single day. The average active consumer transacts more than seventy times a month — over twice a day. Seven hundred and thirty-two thousand active merchants. And payments plus marketplace together produce roughly two-thirds of net income.

That last one is the number that should keep a bank executive awake. Marketplace was zero in 2013. A decade later the "non-core" business is the core, and the market values the company as a platform rather than as a bank. That re-rating is worth more than any operational improvement available in the same window.

Why the transaction never followed the attention online

Roughly ninety-five percent of retail in emerging markets still happens in a physical store. That is not a lag that is about to close. It is a structural fact about basket sizes, fulfilment economics and population density.

So the money is not in moving the goods. The money is in moving the shopper — and in owning the surface where they decide where to go.

There is a number I use to explain this to people who have never had to think about it. If a return taxi fare costs thirty rand, then a basket at the distant store has to be about thirty rand cheaper before it is worth the trip. Proximity is not a convenience feature in this market. It is a savings feature, and it is denominated in currency the shopper can feel.

A nearby store with a mediocre price regularly beats a distant store with a great price. Almost no digital surface in emerging markets tells a shopper that. Mapping apps know where the shop is but not what is on promotion inside it. Retailer apps know the promotion but only for their own estate. The gap between those two is the business.

The funnel: six stages, and most organisations stop at the first

1. Audience. The customer base you already pay to acquire and retain. You have it. That is the point — this is a sunk asset, not a new investment.

2. Attention. A recurring, high-intent moment inside your app. This is the scarcest input in the whole model, and the one people get wrong most often. A tab requires a habit that does not exist yet. A moment inherits one that does — the recharge confirmation, the balance check, the notification that money has landed.

3. Intent. A customer signalling what they want, where, and when. This is the first genuinely new data asset in the chain. Nobody in these markets measures demand before the till.

4. Footfall. A verified walk-in to a specific merchant. This is the product you actually sell.

5. Transaction. Payment, wallet, loyalty. Cashback returns to the wallet inside your app, because the point of a reward is not to make this purchase cheaper — it is to make the next purchase start with you.

6. Compounding. Merchant turnover data and consumer demand data. This is the moat: underwrite merchants on observed turnover, and sell verified demand signal to brands.

Stage four is where this is won or lost

If you cannot prove the walk-in, you have built a banner ad on your own property. No brand funds a second campaign. No merchant believes the first one worked. And you cannot underwrite a shopkeeper on anything better than a bank statement.

If you can prove it, you own the first offline attribution layer in your market — and you can price in-store promotion on verified incremental spend rather than on faith. Build the measurement before the merchandising.

This is also where the four partner types diverge sharply. A bank gets attribution almost free: the card is the receipt. A telco has to build it — but can zero-rate the layer so a customer with an empty bundle can still see the offer, which no retailer or marketplace can copy. An MVNO wins on density rather than breadth; fifty merchants in one corridor outperform five thousand scattered nationally. A retailer already has supplier trade budgets flowing and unmeasured, and can be first to price them on verified basket.

The part where most in-house builds die

Every partner asks whether they should just build it. The honest answer is that the software is not the hard part, and I would not pretend otherwise.

The hard part is that within two quarters, the organisation discovers it has become a field sales operation for small businesses. Signing, verifying and activating tens of thousands of merchants, most of them informal and cash-based and with no interest in a portal. Keeping the in-app price identical to the price at the till, across thousands of independent operators, every week. Merchandising the thing so it is alive rather than a stale directory.

None of that is technically difficult. All of it is operationally relentless, and it is a permanent running cost rather than a project with a completion date. That is a different company to the one you are running.

Where the intelligence layer changes the arithmetic

Kaspi built its merchant side by hand, over a decade, in one country. That is not a timeline most people have, and it is not a model that survives four markets at once.

What compresses it is AI — not as a feature, but as the thing that makes the merchant side economically possible at all. A shopkeeper will not read a dashboard. Every company building for informal retail eventually ships one, and almost none of them get used.

So the layer has to remove the dashboard: replace "here is your data" with "here is what to do", delivered on WhatsApp, in her own language, before she thought to ask. What to stock this week. Why sales dropped yesterday. Whether she can afford to restock tomorrow. Which customers to message today.

And there is a commercial reason this matters more than it sounds. A merchant who uses that daily generates the transaction, inventory and repayment data that makes her underwritable. A merchant who ignores a dashboard generates nothing. The AI is not decoration on the marketplace — it is the mechanism that turns a listing into a data relationship, and the data relationship is what eventually becomes a credit book.

Back to the forty seconds

The man in Soweto is not going to download a new app. He is not going to attend a workshop, or learn a platform, or care about anybody’s digital transformation. He is going to open the app where his airtime lives, because that is where his money already is.

That is the whole opportunity and the whole constraint at once. The audience is already yours. The merchants are reachable. The shopper has already moved the decision onto the phone — the transaction simply never followed, and it is not going to.

The only real question is who builds the surface where that decision happens, and whether you would rather own it or rent access to it from whoever does.

This is the thesis we build on at Flood: white-label marketplaces embedded inside telco and bank apps, with an AI layer that becomes the merchant’s operating system. Live in West Africa today. No new app to learn, and no new company for the partner to become.

What would your business look like if the audience you already pay for had a reason to open the app every day?

Originally published on LinkedIn on 14 October 2026. Read the original post on LinkedIn →