The dormancy problem: why 34% of mobile money agents go inactive within 90 days

In every mobile money network across East Africa, there is a dormancy problem. Operators recruit agents, onboard them, equip them — and then watch 30-40% go inactive within their first 90 days. The standard response is to recruit more. The smarter response is to understand why.

The three predictors

After analysing field visit data and transaction records from mobile money networks across Tanzania, three factors emerge consistently as the strongest predictors of agent dormancy within 90 days.

1. Float management failure

An agent who runs out of float cannot transact. An agent who cannot transact for more than 48 hours starts to disengage. After 72 hours without a transaction, the probability of full dormancy within 30 days rises sharply.

The critical metric is not the float balance at any given moment — it is the trend. An agent whose float is declining over three consecutive measurement points is at risk, even if the current balance is above the threshold. TradeMetriq's float monitoring uses a linear projection to identify declining trajectories before the balance hits critical.

2. KYC gaps

Incomplete KYC (Know Your Customer) documentation correlates strongly with early dormancy. The mechanism is not regulatory — it is motivational. Agents with incomplete KYC know they are operating in a grey area. They are more likely to disengage when volume is low, because they have less confidence in their legitimacy.

Tracking KYC completion rates per agent, per outlet, and per region gives operations teams a proactive lever. Closing a KYC gap takes a single field visit. Recovering a dormant agent takes weeks.

3. Visit frequency decay

The most consistent predictor of dormancy is visit frequency. An agent who was visited weekly in their first month, then biweekly in month two, then monthly in month three, is following a decay curve that ends in dormancy.

The field team's own visit schedule is, in a feedback loop, one of the strongest drivers of agent activity. When the supervisor shows up, the agent activates. When the supervisor stops coming, the agent drifts.

The mathematics of prediction

TradeMetriq's churn risk score for mobile money agent outlets uses a weighted composite:

float_risk(35%) + kyc_gap(30%) + visit_frequency_decay(35%)

Each component is calculated independently, then combined with fixed weights. The result is a 0–100 risk score, updated daily. Agents scoring above 70 are flagged for intervention.

The weights reflect empirical findings from field data. They can be adjusted for specific network conditions — but the default weights are calibrated for the East African mobile money environment.

What operations teams can do

The dormancy problem is not unsolvable. It is a data problem masquerading as a people problem.

When operations teams have daily visibility into float trends, KYC completion rates, and visit frequency decay — and when that visibility triggers early alerts rather than month-end reports — dormancy rates fall significantly. Haraka Pesa Mobile Ltd reduced agent dormancy by 34% in their first 90 days on TradeMetriq. The intervention was not a new training programme or an incentive scheme. It was earlier information, acted on earlier.

That is all early warning systems are: earlier information. The mathematics tells you where to look. Your field team does the rest.