Notification and Customer Communication Platforms

Why messaging stops being a small line item

Notification platforms are rarely in anyone's plan for the year. They reach the agenda in one of two ways. Finance looks at the messaging spend and finds it cannot be broken down by type, by product or by the department that caused it. Or a customer disputes a transaction, says nobody told them, and there is no way to show what was sent.

In Tanzania a mobile number is the practical address of a customer, and each message to it costs money. That cost grows with customer numbers rather than with the balances those customers hold, so institutions growing through many small accounts feel it first. It also decays: customers change numbers and drop the SIM they registered with, so a number taken at account opening may now reach nobody. Without evidence of delivery, an institution assumes it can still contact its whole book.

Where it hurts depends on what you do. For a lender, reminders are a collection instrument: a message that does not arrive is money not collected, and one that arrives the day after a customer paid costs goodwill for nothing. For a SACCO or a microfinance institution, contacting a member every month is a real share of what that membership earns. For a bank the exposure is disputes, fraud warnings and the changes you must announce before they take effect. For a telco or a fintech, messaging is a running cost and part of the product at once.

What customers receive, and what your own people can do

Start with the messages themselves, because few institutions have seen the whole list written down. Money in and money out. Salary arrival and low balance. One-time codes at login and at payment. Loan payments falling due, and missed. Statements, and the proof of income a landlord or an employer asks for. Service notices when something is unavailable. Fee changes you are obliged to announce. Campaigns and their follow-up.

Customers should have a say in this. Which channel they prefer: a text message on a basic handset, a message inside your app, a recorded voice call where reading is a barrier, or the chat applications they use all day. Which language they read. And the ability to stop what they never asked for without losing the messages that protect their money. A customer forced to choose between offers and fraud warnings will choose silence.

The other half is for your own staff, and it is where most of the return sits. Marketing defines an audience, sees what a send will cost before committing to it, and sees what came back. Compliance approves wording before it goes out, not after a customer forwards it to somebody. A support officer on a call can see what that customer was sent and when, which settles most arguments about being told. Finance reads cost by message type, product and channel.

Which of these launch first is agreed with you. The most useful early conversation is often about what to stop sending: institutions that look properly find messages going out that nobody has owned for years.

Consent, and the difference between telling and selling

A customer with a loan expects to be told when a payment is due. That is part of the service they bought; not sending it would be the failure. The same customer has not agreed to be sold a second product because they hold the first. Tanzania has a Personal Data Protection Act, and the line between telling a customer what serving them requires and selling them something is one it takes seriously.

This is where institutions tend to be exposed. Agreement to be contacted is often one tick on a form signed at account opening, treated ever since as covering everything. When a customer objects, or a supervisor asks, that is a thin answer. What holds up is a record of what each customer agreed to, when and through what, and an opt-out that stops what they asked to stop and nothing else.

The commercial argument runs the same way. Over-messaging costs twice, once at the price per message and again in attention. Customers who hear from you constantly stop reading, and the message that mattered, a fraud warning or a final arrears notice, is ignored with the rest. How often you contact a customer, at what hours, and which messages survive a budget cut are yours to decide.

What has to be arranged before the first message goes out

Reaching customers by message is a commercial arrangement before it is anything else. You need agreements with the mobile network operators whose subscribers you serve, or with a messaging provider that already holds them. Price is per message, negotiated against volume and different by network, and the forecast it rests on is yours to produce. Where a short code is involved, it sits within the national numbering plan the Tanzania Communications Regulatory Authority administers.

Then the name customers see when a message from you arrives. Sender identities are registered with the operators or through your provider, and they belong in your institution's name rather than a supplier's. In a market where customers are targeted constantly by people posing as their bank, the name at the top of a message is part of what makes it believable.

Every message is a statement by your institution about a fee, a date, an obligation or a right, delivered to your whole customer base with nobody reading it first. That makes the wording a matter for compliance and legal, and approving a library of it takes longer than teams expect. It needs an owner afterwards, because products and charges change and the messages describing them must change too.

Language is a business decision in the same way. English and Kiswahili at a minimum, with a rule for which customers get which. Translating financial wording is not clerical work: what a term means to a customer reading it in Kiswahili is a product question, usually settled by your own staff rather than a translator. A longer message costs more to send, every month, to every customer. These decisions sit with you, and the agreements among them are usually the longest lead time.

Everything else you run is a sender

This is the platform your customers never choose. Your app, your USSD menu, your agent network, your lending book and your payments all have reason to reach a customer, and so does your marketing team. The question is whether messaging is arranged once, with an owner and a budget, or bought again inside every project.

It sits closest to your customer records, since a campaign is only as good as the segment behind it. We build those too, so this can be scoped under the channels you run or alongside your first one.

Working with us

We start with discovery: what you send today, what it costs, what your support desk is asked, and what you are obliged to tell customers. That produces a scope and a plan, and the plan is yours whether or not we 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. We will usually argue for going live on the messages that carry risk before the ones that carry campaigns. Afterwards we support the platform and train your people. There is no lock-in, by design and by contract.