Collection Agency Performance Management: Compare Agencies
Last reviewed: November 2026
Collection agency performance management is the process of measuring, comparing and managing external collection agencies on recovery, cost and conduct, using common data and fair allocation. It tells a lender which agencies should get more accounts, which need correction and which should exit.
This guide is for heads of collections and COOs at NBFCs, banks and fintech lenders that use external agencies. It covers why raw rankings mislead, how to build a fair scorecard and how RBI's 2027 recovery rules change agency oversight.
Key Takeaways
- From 1 January 2027, RBI requires NBFC recovery policies to cover agency due diligence, performance evaluation, audit and penal action.
- Agents must hold the IIBF debt recovery agent certificate, and calls must be recorded and kept for six months.
- Raw recovery rankings reward the agency with the easiest accounts, so compare on a mix-adjusted basis.
- Conduct belongs in the scorecard, because RBI says targets and incentives must not induce harsh practices.
Why Is Agency Performance a Leadership Problem?
Agency performance is a leadership problem because agencies control a large share of recoveries and almost all of the lender's conduct risk in the field. A poor agency costs money twice: in missed collections and in complaints, penalties and reputational damage.
RBI has made this explicit. The NBFC Responsible Business Conduct amendment of 6 August 2026, effective 1 January 2027, requires a recovery policy covering agency eligibility, due diligence, performance evaluation, inspection or audit, and penal action against non-compliant agencies.
The same rules require compensation for borrowers and guarantors who suffer losses from non-compliant recovery actions. Agency oversight is now a board-level policy matter, not a vendor management task.
Why Do Agency Comparisons Usually Mislead?
Agency comparisons usually mislead for five reasons: uneven allocation, different definitions, missing conduct data, payouts on gross collections and stale agent records. Each makes a weak agency look strong or a good agency look weak.
| Problem | What Happens | Where It Shows Up |
|---|---|---|
| Uneven allocation | One agency gets fresher, easier accounts | High raw recovery that does not hold up |
| Different definitions | Agencies count "resolved" differently | Reports that cannot be compared |
| Missing conduct data | Complaints and call quality not linked to agencies | Conduct problems found only after escalation |
| Gross-collection payouts | Agencies paid on all money received | Payouts for accounts that would have self-cured |
| Stale agent records | Uncertified or unverified agents in the field | Audit findings and regulatory risk |
Fixing these is mainly a data and allocation problem. Once allocation is fair and data is common, the scorecard becomes credible to agencies as well.
How Do Lenders Manage Collection Agency Performance?
Lenders manage agency performance in five layers, from allocation to payouts. Each layer feeds the next, and conduct runs through all of them.
- Fair allocation: distribute comparable accounts across agencies.
- Common data capture: record every contact, promise and payment the same way.
- Balanced scorecard: measure recovery, cost and conduct together.
- Conduct monitoring: audit calls, visits and complaints by agency.
- Payouts and action: link incentives, warnings and exits to the scorecard.
Layer 1: Are Agencies Getting Comparable Accounts?
Allocate accounts randomly within segments defined by days past due (DPD) bucket, ticket size, product and geography. This champion-challenger design lets you compare agencies on like-for-like books.
Digiqt's collections prioritization AI agent shows how accounts can be scored and placed with external agencies based on expected recovery.
Layer 2: Is Every Contact Recorded the Same Way?
Give every agency the same disposition codes, promise-to-pay (PTP) fields and payment receipt process. RBI's 2027 rules also require lenders to document the time and number of calls and to issue proper receipts for any collection.
A shared field collection app makes this easier than collecting spreadsheets from each agency.
Layer 3: Does the Scorecard Balance Recovery, Cost and Conduct?
Measure recovery against the expected rate for each agency's account mix, not in absolute terms. Then add cost per ₹100 recovered and conduct measures such as complaints and call audit results.
Weight conduct heavily enough that it can change an agency's ranking. Otherwise agencies will learn that only collections matter.
Bucket averages are a starting point, not the best benchmark. Digiqt's recovery rate prediction AI agent shows how expected recovery can be estimated account by account, which makes the comparison fairer still.
Layer 4: How Is Conduct Monitored?
From 1 January 2027, calls must be recorded, the borrower told about the recording, and recordings kept for six months, or longer if the matter is in court. Contact is allowed only between 08:00 and 19:00 unless the borrower requests otherwise.
Sample recordings by agency every week, and track complaints per 1,000 accounts. Digiqt's conduct risk surveillance AI agent shows how conduct signals can be monitored at scale.
Layer 5: Do Payouts Reward the Right Behaviour?
Pay on mix-adjusted performance and roll-backs, not on gross money received. RBI's amendment states that recovery targets and incentives must not induce harsh practices.
Define penal actions in the contract, such as payout holds, reduced allocation and termination. If an agency is terminated, RBI's rules require borrowers to be told and the website list updated promptly.
Is your best-ranked agency actually your best, or just the one with the easiest accounts?
Talk to Our Lending Technology Specialists
Visit Digiqt to see how SalesPro tracks follow-ups and performance across field teams.
How Does Collection Agency Performance Management Work on Real Portfolios?
On real portfolios, a mix-adjusted scorecard often reverses the raw ranking. In this illustrative scenario, an NBFC gives two agencies 5,000 accounts each, with an average overdue of ₹24,000, or ₹12 crore per agency.
Agency A received mostly 31–60 DPD accounts, and Agency B mostly 61–90 DPD accounts. The lender's historical recovery benchmark is 38% for 31–60 DPD and 26% for 61–90 DPD.
| Measure | Agency A | Agency B |
|---|---|---|
| Account mix | 70% in 31–60, 30% in 61–90 | 30% in 31–60, 70% in 61–90 |
| Expected recovery | 34.4% | 29.6% |
| Actual recovery | 35% (₹4.20 crore) | 32% (₹3.84 crore) |
| Mix-adjusted index | 1.02 | 1.08 |
| Commission paid | 8%, ₹33.6 lakh | 7%, ₹26.88 lakh |
| Complaints per 1,000 accounts | 3.6 | 0.8 |
On raw recovery, Agency A looks better by ₹36 lakh. On a mix-adjusted basis, Agency B outperforms, costs less per rupee and generates less than a quarter of the complaints.
The lesson for leadership: raw rankings were rewarding easier accounts and riskier conduct. Fair allocation and a balanced scorecard moved volume to the agency that deserved it.
Which Approach Fits Your Collections Model?
The right approach depends on how many agencies you use, your DPD mix and how much of collections is in the field. Larger books need formal champion-challenger allocation, while smaller lenders can start with a common scorecard.
| Situation | Recommended Approach |
|---|---|
| Two or three agencies, small book | Common data format and monthly scorecard with conduct metrics |
| Many agencies across states | Randomised allocation within segments and mix-adjusted indices |
| Heavy field collections | Visit logs, receipts and geotagged activity in one app |
| Digital lender with LSP-led collections | Same scorecard for LSPs acting as recovery agents |
What Changes for Digital Lenders?
Digital lenders must also tell the borrower, by email or SMS, the details of any recovery agent before that agent makes contact. Our guide to RBI digital lending guidelines compliance covers the wider framework for lending service providers (LSPs).
Where Does SalesPro Fit in Agency Management?
SalesPro fits as the enablement and visibility layer for field teams. Digiqt built SalesPro as a sales acceleration platform, and its stated features map closely onto how collections leaders manage agents.
The platform lists lead and follow-up tracking with automated reminders, and performance dashboards that highlight top and underperformers. It also offers gamified training through videos and quizzes, role-play practice for objection handling, instant Q&A support and centralised access to product information.
For agencies, that means one place to train agents on scripts and conduct rules, practise difficult conversations and see who follows up. Allocation logic and recovery data would still come from your collections system.
Onboarding should follow the same discipline. Verify agent antecedents before engagement, check certificates, obtain the agency's written undertaking on your code of conduct, and start new agencies on a probation allocation before full volume.
Earlier in the cycle, Digiqt's early-stage collections AI agent shows how accounts can be resolved before they ever reach an agency.
Which Metrics Show Agency Performance Clearly?
Six metrics show agency performance clearly: mix-adjusted recovery, roll-back rate, cost per ₹100 recovered, complaints per 1,000 accounts, PTP kept rate and certification coverage. Review them together, because any one alone can be gamed.
| Metric | Formula | What a Weak Number Signals |
|---|---|---|
| Mix-adjusted recovery | Actual recovery ÷ expected recovery for the mix | Agency underperforms on comparable accounts |
| Roll-back rate | Accounts moved to a lower DPD bucket ÷ accounts allocated | Collections that do not cure accounts |
| Cost per ₹100 recovered | Agency payout ÷ amount recovered × 100 | Payout structure too generous |
| Complaints per 1,000 accounts | Complaints ÷ accounts allocated × 1,000 | Conduct risk in the field |
| PTP kept rate | Promises kept ÷ promises recorded | Pressure tactics or fake promises |
| Certification coverage | Certified active agents ÷ active agents | Regulatory breach from 2027 |
Share each agency's scorecard with that agency every month. Transparent scorecards change behaviour faster than penalties.
Which Mistakes Weaken Agency Oversight?
Four mistakes weaken agency oversight, and each one becomes more costly under the 2027 rules.
- Ranking on raw recovery. It rewards easy allocations, not skill.
- Ignoring conduct until a complaint escalates. Call audits and complaint data must feed the scorecard every month.
- Losing track of individual agents. Certification, background checks and ID cards apply to people, not just agency firms.
- Paying on gross collections. Self-cures and borrower-initiated payments should not earn full commission.
What Is Changing Next?
The biggest change is RBI's recovery framework for NBFCs and banks, effective 1 January 2027. It adds published agency lists updated within seven calendar days of any change, prior intimation of agency details at least one day before the first visit, and strict limits on locking financed devices.
Lenders will need agency data that is complete, current and auditable. For a wider view of recovery operations, see our guide to collection efficiency, or build these controls into a digital lending platform from the start.
Ready to compare every agency on recovery, cost and conduct before RBI's 2027 rules apply?
Talk to Our Lending Technology Specialists
Visit Digiqt to see how SalesPro and our lending platform give collections leaders full visibility of field teams.
