From Journey Analytics to Behavioral Hypotheses: A Worked Example

Journey analytics can feel abstract until you walk through one real funnel. This article uses a simplified repayment journey—numbers are illustrative; the method is what you reuse.

The data snapshot

• 10,000 borrowers due this week.

• SMS delivered: 98%.

• Link clicked: 22%.

• Payment within 48 hours: 11%.

Delivery is fine. Action is not. Where is the leak?

Stage-by-stage read

• Delivered → clicked (22%) — many never engage the link. Possible barriers: weak salience in line one, unfamiliar sender, bad timing (reminder before salary).

• Clicked → paid (roughly half of clickers, far fewer overall) — friction after click: slow page, broken M-Pesa handoff, unclear amount.

The biggest opportunity depends on context. If sender trust is weak, fix Sender ID and wording before UI. If clicks are strong but payment weak, fix effort on the pay path. SMS enablement may be step zero.

Behavioral hypotheses (pick one to test first)

• Salience: "If amount and due date move to line one, click rate rises."

• Timing: "If send moves to two days before salary day for this segment, 48-hour payment rises."

• Friction: "If we replace three-step pay with one M-Pesa link, conversion among clickers rises."

Write one sentence. Design A vs B. Agree on 48-hour payment rate as primary metric.

Qualitative cross-check (do not skip)

Ask collections or support: What do customers say when they delay? If they say "I forgot until late fees," present bias and timing dominate. If they say "I did not trust the link," trust dominates. Data and stories should converge before you scale.

Close the loop

Run two weeks. Readout with data + frontline. Scale, iterate, or revisit diagnosis via Pulse. Execute the next variant in Hub when trials fit.

Related: A/B testing SMS · Behavioral economics for SMEs