Agency Banking Platforms

Why an agent reaches where a branch cannot pay for itself

A branch has to pay for premises, a safe, staff, security, power and the movement of cash. In a district town, all of that comes out of a book of small balances and small transactions. The arithmetic rarely works, which is why the branch map thins out beyond the regional centres. Customers experience that as distance: a bus fare and half a day to make a deposit worth less than the journey.

An agent changes the arithmetic. The shop is already there, already open in the evening, already holding cash for its own trade. You pay commission on transactions that happen instead of fixed costs on a building that is quiet most of the week. For what puts one branch in one town, you can be present in dozens of trading centres.

Mobile money proved the model here before banks used it, so customers in every district already know how to hand cash to a shopkeeper and expect it to land somewhere. That familiarity is worth a great deal to a bank, a SACCO or a microfinance institution arriving in the same format. It also carries a warning: your customers are already transacting at somebody's agent, and the commercial question is whose.

What a customer can do at the counter

Cash is why people come. Deposits into their own account, deposits into someone else's, and withdrawals carry the volume, and are what the network is judged on.

Around that sits the rest of the counter: transfers to other customers, other banks and mobile money wallets, bill and utility payments, school fees, loan repayment where you lend, balance enquiry, and a record the customer keeps. Account opening at the counter changes what the network is for, because it turns every agent into a recruitment point and brings identity and customer data obligations with it. Take that one deliberately rather than because it is available.

Which journeys launch first is agreed with you, against what your customers travel to a branch to do today. What they notice afterwards is narrow: whether the agent could tell them at once that the transaction had gone through, whether they left with a record of it, and whether the amount, the fee and the balance were beyond argument. A dispute at an agent counter costs more than a dispute anywhere else, because it happens in public, in front of the next customer in the queue.

What the agent needs, which is a different list

An agent is not a member of your staff. They are a trader who has put their own working capital into a float and expects a return on it. Their questions are immediate and commercial: how much have I earned today, when am I paid, how much float is left, why did that transaction fail, and who do I call about it.

So the agent's side of the service is a business tool rather than a teller screen. Float top-up and the record of every top-up. Commission earned, commission settled and what is still owed. A transaction history they can search when a customer returns three days later disputing an amount. More than one till operator in a busy shop, each of them identifiable. A support route that answers the same day. And, on your side, the ability to suspend an agent immediately.

Agents leave quietly. They rarely resign; they stop offering the service, or tell customers the system is down when they are out of cash or out of patience. Attrition shows up in your coverage map a year later, and recruiting replacements is expensive.

Where agents sit beside your branches and your other channels

Agency banking extends the branch network rather than replacing it. Routine cash moves out to the trading centre, and the branch keeps the business that needs a banker: lending conversations, high value and complex transactions, disputes, and customers who want your own staff.

The agent also serves many of the same customers who use your USSD service and your app. Someone withdraws cash at a shop in the morning and checks the balance on a feature phone in the afternoon, and expects one account, one set of limits and one fee table. Channels commissioned separately at different times drift, and here the drift lands face to face. An agent quoting one fee while the customer's handset shows another is not a technical inconsistency to anyone. It is an argument at a counter.

We build across these channels, so agency banking can fit into what you already run or form part of a wider programme.

What standing up an agent network involves

More of this project sits on your side than on any other channel, and very little of it is software. Start with the regulator. The Bank of Tanzania supervises banking in Tanzania, and serving customers through third parties happens under that supervision. What you must satisfy to appoint agents, and what you remain answerable for once you have, is a conversation for your compliance team and your supervisor, better held before recruitment than after it. Tanzania also has a Personal Data Protection Act, and an agent taking a customer's details is handling personal data on your behalf.

Then the network itself. Who qualifies as an agent and how you check them: trading history, premises, standing in the community, the capital to hold a float. What the contract says about liability for cash, exclusivity and termination. What commission pays per transaction type, who bears the fee, and whether an agent in a small trading centre can earn enough on the volume that location will realistically produce. Set commission too thin and agents quietly stop serving customers. Set it too generously and the channel loses money as it grows. That number decides more about the network's survival than anything on a screen.

Then liquidity, which is where live networks fail. An agent who is out of cash is a closed branch with your name on the wall. Where agents top up, how far they travel to do it, who serves each district, and what happens on salary dates and market days are arrangements between you, your treasury and your branch network. Software can show you the position. It cannot put cash in a shop in a district.

Then the field. Agent training, the branding of outlets so a customer can recognise a genuine agent, and the support model: who visits agents, how often, and who answers the telephone when a transaction fails with a customer standing there. An agent network is a field operation with software attached, and the institutions that do well with it resource it that way.

How we work with you

We start with discovery: your book, the districts you want to reach, the products you want at the counter, and the operational realities you work inside. That produces a scope and a delivery plan, and the plan is yours whether or not we go on to build it.

Through delivery you see working software in an environment you control rather than status reports, and the source code and documentation are yours from the first commit. Most networks are worth proving with a few agents in one district before recruitment goes wide, and we plan for that with you. Afterwards we support the service in production and train your people to take on as much of it as they want. There is no lock-in, by design and by contract.