CRED collected ₹0.45 Cr more with the Riverline virtual programme.

What is CRED, and what were they solving for?

CRED is a members-only fintech platform that rewards users for paying their credit-card bills on time.

Early-bucket (B0) borrowers are accounts only days past due. Reach them in that window and most pay. Miss it and they slide into buckets that are far harder to recover.

The question was simple: how do you improve recovery in this bucket without adding headcount? Riverline worked 3,513 of these borrowers in June and 12,000 in July.

Four things needed to change:

1. Payment inside the chat. A borrower who is ready to pay should not have to go looking for the payment page.

2. A plan behind the reminders. Not single messages, but a set that runs across the cycle.

3. A real conversation. Something that can work out why a borrower has not paid, and respond to it.

4. Proof. A side-by-side comparison that shows what the effort actually added.

How the program works

Riverline designs and runs the program on CRED’s early-bucket book: the messaging, segmentation, timing and measurement. It runs to the goals CRED sets. That replaces single, one-off reminders with a sequenced conversation across the collections cycle. In practice:

1. WhatsApp as the primary channel: 94.3% of messages were delivered and 75.2% were read.

2. A WhatsApp bot that understands the borrower and still pushes for payment: it works out why the borrower has not paid, while keeping the conversation pointed at clearing the dues.

3. Follow-ups that carry the last conversation forward: the next message picks up where the previous exchange ended and refers to what was already discussed, instead of restarting from a template.

4. An instant reply, with the payment link in it: when a borrower answers, the bot responds within seconds and puts a one-tap link in front of them. A borrower who says they are ready to pay is converted while the intent is still live, not hours later when it has cooled off.

5. A sequenced cadence: first reminder, final-day nudge, then follow-up broadcasts. Seven campaigns across June rather than a single touch.

6. Test-vs-control by default: changes are measured against a held-back group, so the program’s contribution stays separable from everything else.

Results

One month on CRED’s early-bucket (B0) book, measured side by side against a held-back group on the usual process:

• 3.4× program scaled month over month (3,513 → ~12,000 borrowers).

• +1.3pp pay rate lift vs usual process.

• 94.3% message delivery rate.

• 76.7% of repliers went on to pay.

Pay rate: Riverline virtual programme vs usual process

Usual process: 77.6%. Riverline virtual programme: 78.9%. Lift: +1.30 percentage points, adding ₹0.45 Cr more collected. The scale runs 77.0 to 79.5%, zoomed, not zero-based.

Across the cohort, 72.8% of the tracked outstanding book was resolved and 74.6% of the overdue amount recovered. Pay rate improved in every engagement segment month over month.

Next on the program: the same setup is built to run at around 1,00,000 cases, with AI voice calling coming online alongside WhatsApp on the same borrower record.

Looking to lift collections on your early-bucket book?

If you’re looking to reach every borrower on the channel they read, run controlled, measurable campaigns, and scale recovery without scaling headcount, we’d love to show you how Riverline can help. Book a demo at riverline.ai.

CRED × Riverline · June 2026 cohort (3,513 borrowers) · Figures pulled live 2 July 2026. Pay rates measured on principal outstanding. Rupee amounts shown as percentages by design.