Customers Case studies Muthoot Fincorp · Collections
ORI × Muthoot Fincorp · Case study

AI voicebot collections program drives measurable recoveries

In its second month live, the AI voicebot collections program recovered payments from 5,171 of 21,796 accounts dialled across three campaigns — a blended 23.73% collection rate, with Post-Due Bandhan Plus reaching 31.56%, the strongest of the three.

SectorNBFC
Use caseCollections
GeographyIndia · pan-portfolio
LanguagesHi · Ta · Te · Kn · Ml
Last verified Aug 2026 · from DATA LAKE snapshotSee all Collections case studies →
Recovery by bucket · July 20265 LANGSLive
Post-Due Bandhan+31.56%
Bucket X (Field)23.33%
Bucket 1 & 213.92%
→ Payment Gateway LIVEReal-time collections happen inside the call itself.
5,171Paid
21,796Dialled
5Languages
23.73%
Blended collection rate
All buckets · July 2026
31.56%
Post-Due Bandhan Plus
strongest bucket
5,171
Accounts recovered
of 21,796 dialled
5
Languages live
Hi · Ta · Te · Kn · Ml

About the client

Muthoot Fincorp engaged ORI to run AI voice bot campaigns for collections outreach on its loan portfolios. Collections outreach today spans three campaign buckets — Post-Due Bandhan Plus, Bucket X (Field) and Bucket 1 & 2 — carried in five languages: Hindi, Tamil, Telugu, Kannada and Malayalam.

The challenge

With real collection outcomes now measurable across languages and buckets, the program's priorities have sharpened around recovery performance itself.

01

Collection rates vary widely by language within the same bucket and need levelling up.

02

Bucket 1 & 2 is recovering at roughly half the rate of Post-Due Bandhan Plus and needs targeted script tuning.

03

Redial cycles need to be enabled to convert more of the unpaid base after the first full round.

04

Dialer integration, needed for bot-to-agent handoff on complex disputes, is still upcoming.

05

Recovery performance needs to be proven at scale before extending to the Gold Loan portfolio.

06

Conversations need continued fine-tuning so recovery, not just connection, improves across all five languages.

Results, measured end-to-end

Paid/unpaid outcomes by collections bucket, across all five languages combined.

Collections bucketPaidTotal dialledColl. %
Post-Due — Bandhan Plus2,6028,24431.56%
Bucket X — Field1,6927,25223.33%
Bucket 1 & 2 (Seamless only)8776,30013.92%

Integrations & program building blocks

Overdue API — LIVE. Real-time overdue data feeds directly into bot conversations. Payment Gateway — LIVE. Real-time collections happen inside the bot call itself, contributing directly to the recovery numbers above.

The solution we designed

End-to-end journey

The Pre-Due Bot reminds customers ahead of their EMI due date to encourage on-time payment. The Post-Due Bot engages overdue customers, captures intent and dispositions, and nudges payment.

Multilingual experience

Collections outreach runs in five languages — Hindi, Tamil, Telugu, Kannada and Malayalam — with recovery tracked independently in each.

Integrated systems

The Payment Gateway enables real-time collections within the call itself, feeding directly into the paid/unpaid outcomes tracked per bucket.

Measurement built in

Every bucket is tracked to a paid/unpaid disposition and a collection rate, by language, so recovery performance — not just call activity — drives the program's decisions.

Collections outreach across our loan portfolios means reaching thousands of customers every month, across five different languages. Our best-performing segment now recovers close to one in three accounts we reach — a level of consistency manual calling alone couldn't get us to. That has given us real confidence to scale this across the rest of our portfolio.

MBusiness Stakeholder · Muthoot Fincorp

Program governance & cadence

Bi-weekly leadership cadence. Leadership meets every two weeks to review collection-rate trends by bucket and language, and align on improvement pointers.

Continuous call audits. Ongoing review of call recordings on lower-recovery segments to catch gaps and improve conversion.

Post-campaign analysis. Disposition data is analyzed bucket-by-bucket and language-by-language to understand who paid and who did not.

The takeaway

In its first full month of measurable recovery performance, the AI voice bot collections program converted 5,171 of 21,796 dialled accounts into payments — a blended 23.73% collection rate — with Post-Due Bandhan Plus leading at 31.56%. Levelling up the lower-recovery segments, enabling redial cycles, and completing dialer integration are next, carried forward through a bi-weekly leadership cadence, ahead of an extension to the Gold Loan portfolio.

Have a workload like this?

We'll pull the closest live production analog and walk you through how the numbers were reached.

Customers Case studies Muthoot Fincorp · Collections
ORI × Muthoot Fincorp · Case study

AI voicebot collections program drives measurable recoveries

In its second month live, the AI voicebot collections program recovered payments from 5,171 of 21,796 accounts dialled across three campaigns — a blended 23.73% collection rate, with Post-Due Bandhan Plus reaching 31.56%, the strongest of the three.

SectorNBFC
Use caseCollections
GeographyIndia · pan-portfolio
LanguagesHi · Ta · Te · Kn · Ml
Last verified Aug 2026 · from DATA LAKE snapshotSee all Collections case studies →
Recovery by bucket · July 20265 LANGSLive
Post-Due Bandhan+31.56%
Bucket X (Field)23.33%
Bucket 1 & 213.92%
→ Payment Gateway LIVEReal-time collections happen inside the call itself.
5,171Paid
21,796Dialled
5Languages
23.73%
Blended collection rate
All buckets · July 2026
31.56%
Post-Due Bandhan Plus
strongest bucket
5,171
Accounts recovered
of 21,796 dialled
5
Languages live
Hi · Ta · Te · Kn · Ml

About the client

Muthoot Fincorp engaged ORI to run AI voice bot campaigns for collections outreach on its loan portfolios. Collections outreach today spans three campaign buckets — Post-Due Bandhan Plus, Bucket X (Field) and Bucket 1 & 2 — carried in five languages: Hindi, Tamil, Telugu, Kannada and Malayalam.

The challenge

With real collection outcomes now measurable across languages and buckets, the program's priorities have sharpened around recovery performance itself.

01

Collection rates vary widely by language within the same bucket and need levelling up.

02

Bucket 1 & 2 is recovering at roughly half the rate of Post-Due Bandhan Plus and needs targeted script tuning.

03

Redial cycles need to be enabled to convert more of the unpaid base after the first full round.

04

Dialer integration, needed for bot-to-agent handoff on complex disputes, is still upcoming.

05

Recovery performance needs to be proven at scale before extending to the Gold Loan portfolio.

06

Conversations need continued fine-tuning so recovery, not just connection, improves across all five languages.

Results, measured end-to-end

Paid/unpaid outcomes by collections bucket, across all five languages combined.

Collections bucketPaidTotal dialledColl. %
Post-Due — Bandhan Plus2,6028,24431.56%
Bucket X — Field1,6927,25223.33%
Bucket 1 & 2 (Seamless only)8776,30013.92%

Integrations & program building blocks

Overdue API — LIVE. Real-time overdue data feeds directly into bot conversations. Payment Gateway — LIVE. Real-time collections happen inside the bot call itself, contributing directly to the recovery numbers above.

The solution we designed

End-to-end journey

The Pre-Due Bot reminds customers ahead of their EMI due date to encourage on-time payment. The Post-Due Bot engages overdue customers, captures intent and dispositions, and nudges payment.

Multilingual experience

Collections outreach runs in five languages — Hindi, Tamil, Telugu, Kannada and Malayalam — with recovery tracked independently in each.

Integrated systems

The Payment Gateway enables real-time collections within the call itself, feeding directly into the paid/unpaid outcomes tracked per bucket.

Measurement built in

Every bucket is tracked to a paid/unpaid disposition and a collection rate, by language, so recovery performance — not just call activity — drives the program's decisions.

Collections outreach across our loan portfolios means reaching thousands of customers every month, across five different languages. Our best-performing segment now recovers close to one in three accounts we reach — a level of consistency manual calling alone couldn't get us to. That has given us real confidence to scale this across the rest of our portfolio.

MBusiness Stakeholder · Muthoot Fincorp

Program governance & cadence

Bi-weekly leadership cadence. Leadership meets every two weeks to review collection-rate trends by bucket and language, and align on improvement pointers.

Continuous call audits. Ongoing review of call recordings on lower-recovery segments to catch gaps and improve conversion.

Post-campaign analysis. Disposition data is analyzed bucket-by-bucket and language-by-language to understand who paid and who did not.

The takeaway

In its first full month of measurable recovery performance, the AI voice bot collections program converted 5,171 of 21,796 dialled accounts into payments — a blended 23.73% collection rate — with Post-Due Bandhan Plus leading at 31.56%. Levelling up the lower-recovery segments, enabling redial cycles, and completing dialer integration are next, carried forward through a bi-weekly leadership cadence, ahead of an extension to the Gold Loan portfolio.

Have a workload like this?

We'll pull the closest live production analog and walk you through how the numbers were reached.