Collections Training Budget 2027-28: Churn-Cost Worksheet

Quick answer
No published benchmark exists for collections training budgets, so build your own. A worksheet for 2027-28 covering churn, certificates and release time.
Your finance partner asks a simple question: how much should we be spending on collections training next year? You look for a benchmark. There isn't one.
We searched. We found no published data on training budgets for collection agents in India, no cost-per-agent figure we'd trust, and no study that links agent training to recovery. There are numbers circulating on attrition and hiring costs, but when we traced some of the most repeated ones back to their sources, they fell apart, so we don't use them.
That's actually useful news for you. It means a collections training budget for 2027-28 can't be argued from an industry average. It has to come from your own numbers: how many people you have, how fast they leave, what the certificate costs, and how much field time you lose. Those numbers will convince a CFO better than any benchmark, because they're yours.
This post gives you a worksheet. The financial year runs from April to March, and January to March is, on our inference, the likely window for planning. Every figure in the worked example is a placeholder. Replace them all. I run a company that sells training software, so bear that in mind when you read the sections on what to buy.
What's different about next year
A budget that's simply last year's plus 10% will miss this. From 1 January 2027, RBI's recovery Directions change what you've got to cover.
| Change | Budget effect |
|---|---|
| Agencies engage only certificate holders; existing agents at NBFCs and most others get one year; commercial banks and small finance banks get none (RBI) | New cost: training, exam fees and time off the field, for existing agents and for every replacement |
| Code of conduct covers your own employees too | Bigger audience for conduct lessons |
| Calls logged and recorded, kept for six months | Technology and audit cost, and a source of training data |
| Monitoring structure and penal action for agencies | Vendor management, audit, trace tests |
| Policy on death and distress | Specialised team, lessons, system changes |
| Certificate window ends around 31 December 2027 for many lenders | The whole of FY2027-28 is the catch-up year |
The last row is important. If you're an NBFC, nearly all of your transition sits within this one budget year. And on our reading, every replacement hired from January needs the certificate at engagement. Our certificate plan sets this out, so have your counsel confirm it.
Numbers we won't use, and why
Before the worksheet, a warning on what you might find in search results.
You'll see very high attrition percentages quoted for NBFC frontline roles and for collection agents. We traced the most repeated ones. They didn't lead to a checkable source, so we won't repeat them here. You'll also find claims that it costs some specific amount to replace an agent, on blogs that don't say where the figure comes from.
What exists that's a little firmer:
- Sa-Dhan's microfinance data: in one quarter, 38,253 staff left and 33,593 joined, on 2.15 lakh staff, partly because of rationalisation. A chart note puts field attrition at about 35% to 40% (Sa-Dhan). This covers microfinance, not agency collectors.
- A TeamLease executive said frontline BFSI attrition was persisting at 30% to 40% in FY24, covering all frontline BFSI roles (MarketScreener, via Reuters). It's dated, and it isn't specific to collections.
Use those for context if you must. Then measure your own. A finance team will trust a number from your payroll and agency lists much more than one from a blog, including this one.
The worksheet
There are seven blocks. Fill them in for each group: in-house field, in-house tele-calling, agency field, agency tele-calling. You can do it in a spreadsheet in an afternoon.
Block A: Headcount and churn
| Line | Your number | Where to get it |
|---|---|---|
| A1. People at the point of customer interface today | Allocation file, call logs, agency lists | |
| A2. Monthly exits, as a % of headcount | Your register. Agencies may need asking | |
| A3. Replacements needed per year (A1 × A2 × 12) | Calculated | |
| A4. Share of exits in the first 90 days | Your register | |
| A5. New territories or growth hires | Business plan |
If you don't have A2 for agencies, that's your first finding. Ask each agency for the past six months. Getting that number will tell you something about how each one's run.
Block B: The certificate
These are IIBF's figures, not ours: exam fee of ₹1,200 plus GST per attempt, plus charges; a suggested consolidated training fee for agencies of no more than ₹4,000 online and ₹6,500 offline; 100 hours below graduate level (about 17 to 30 days) and 50 for graduates (about 8 to 15 days); and a FY2025-26 pass rate of 46.84% (IIBF FAQ, annual report). Check them again before you finalise, since fees change.
| Line | Your number | Note |
|---|---|---|
| B1. Agents without the certificate today | Excludes those who'll leave before they need it | |
| B2. Replacements needing it (from A3) | Unless you hire certified people | |
| B3. Training fee per person | Use IIBF's suggested ceiling as a starting point; get real quotes | |
| B4. Exam attempts per pass | The pass rate doesn't split first and repeat attempts. Try 1.5 and 2.5 | |
| B5. Exam fee per attempt | IIBF's figure, plus GST | |
| B6. Who pays? | Lender, agency or agent. Nothing we read settles this |
That last line might be your most important negotiation of the year. If the agency pays, it'll want a higher commission. If the agent pays, you'll find it harder to hire. If you pay, you've got a line in your budget. Decide knowingly.
Block C: Time off the field
This is usually the largest real cost, and it's rarely in a training budget at all.
| Line | Your number | Note |
|---|---|---|
| C1. Days of certificate training per person | 8 to 30, depending on qualification | |
| C2. Is the person paid during it? | And by whom | |
| C3. Daily pay or stipend during training | Your number | |
| C4. Accounts not worked while they train | Is the territory covered by someone else? | |
| C5. First-week onboarding days | See our first 7 days plan | |
| C6. Mentor days lost | A mentor loses collection time too |
Block D: Conduct and code lessons
| Line | Your number | Note |
|---|---|---|
| D1. Roles and languages | Each language adds review work | |
| D2. Content creation and review | Internal time or a vendor; include compliance and legal review | |
| D3. Delivery | Platform or trainer. Ask vendors; don't assume per-seat pricing suits a high-churn workforce | |
| D4. Team leader time | One huddle question a week adds up | |
| D5. Agent time | Short lessons mean little field time lost; classroom means a lot |
Block E: Oversight
| Line | Your number | Note |
|---|---|---|
| E1. Vendor managers and auditors | Duty under the monitoring paragraph | |
| E2. Call audit: sample size and cost | Now better supported by recordings | |
| E3. Agency visits and trace tests | Our agency checklist has a method | |
| E4. Register upkeep | Someone owns this |
Block F: The cost of not doing it
This is the hard block, and you should be honest about what you don't know.
| Line | What you can say | What you can't |
|---|---|---|
| F1. Cost per complaint | Time spent on your complaint process, from your data | That training will prevent a given number |
| F2. Compensation | Your policy must provide compensation for loss from non-compliant actions; use past cases, if any | A forecast |
| F3. Penalties | RBI's public penalties are small in rupees: Navi Finserv ₹3.80 lakh; Krazybee ₹42.48 lakh; RBL Bank ₹2.27 crore, across five failures | That a fine is the main risk |
| F4. Business restriction | One case found: Mahindra Finance in 2022, which expected 3,000 to 4,000 fewer repossessions a month on a base of 4,000 to 5,000, by company estimate (Business Today) | That it would happen to you |
| F5. Cost per replacement hire | Recruiting, onboarding and lost collections, from your own books | An outside benchmark |
A tip for this conversation: don't build a business case on fear of fines. They're small, and a CFO can look them up. The stronger argument is that this is a cost you're going to carry anyway, through the certificate and through churn, and the only question is whether you carry it in an organised way.
Block G: Options
Give your finance partner three options rather than one number.
| Option | What it includes | Who it suits |
|---|---|---|
| Minimum | Certificate compliance, code acknowledged, basic register | Lenders with small agency exposure |
| Standard | Minimum, plus short conduct lessons by role and language, first-week plan, call audits | Most lenders |
| Strong | Standard, plus specialised hardship team, mentor pay, comparison pilots to test what works | Lenders with large field books or past findings |
A worked example
Here's how Blocks A to C might look. It's an illustrative lender, not a real one, and every number's a placeholder.
| Item | Assumption | Result |
|---|---|---|
| Agents at the point of customer interface | 1,000 | |
| Already certified | 40% | 400 |
| Without the certificate | 60% | 600 |
| Monthly exits | 3% | 30 a month; 360 a year |
| Existing agents who stay long enough to need it | About 70% of 600 | 420 |
| People needing certification in the year | 420 + 360 | 780 |
| Training fee, using IIBF's online ceiling | ₹4,000 | ₹31.2 lakh |
| Exam attempts at 2 per pass | 2 × ₹1,200, before GST and charges | ₹18.7 lakh |
| Training days, assuming an average of 15 | 15 × 780 | 11,700 person-days |
Look at that last line. Fees come to about ₹50 lakh. But 11,700 person-days is a lot of people not collecting, and whichever way you price a day, it's likely to be the bigger number. Agencies know this, which is why they may resist. So when someone says the certificate is cheap, they're looking at the wrong line.
Now look at what the assumption about churn does.
| Monthly exits | Replacements a year | People needing certification | Person-days at 15 days each |
|---|---|---|---|
| 2% | 240 | 660 | 9,900 |
| 3% | 360 | 780 | 11,700 |
| 4% | 480 | 900 | 13,500 |
Each percentage point of monthly exits adds about 1,800 person-days of certificate training, in this illustration. That's the churn-cost argument in one number, and it suggests that keeping people may be worth more than any training efficiency you can find. It also suggests you should look at why people leave in their first 90 days, if that's where your exits cluster.
Where to save money
A few places where you might reasonably cut.
Don't pay for classroom time for conduct content. Bringing a dispersed field force to a room for rules they could learn in 20 minutes on a phone is hard to justify, unless you've got a specific need.
Don't send everyone everything. A tele-caller doesn't need the repossession lesson. Splitting by role cuts time, and it improves the content.
Hire certified people where you can. The certificate has no expiry at present, according to IIBF, so experienced agents who've moved on are worth a call.
Fix the first 90 days. If early exits are your largest source of churn, a modest onboarding investment could reduce the certificate bill too.
Be careful about per-seat pricing. If 30 people leave and 30 join each month, a named-seat licence could mean you're paying for people who've gone. Ask vendors how they treat this. Pay-for-active-use models exist, ours among them, and you should compare them on your own churn.
Where not to save
Compliance review of content. A wrongly translated rule, sent to 1,000 people, is worse than no lesson.
Team leaders. They need their lessons first. If the budget's tight, cut elsewhere.
Measurement. Without a comparison group, you'll never know whether your spending worked, and next year's budget conversation will be the same as this one.
How to talk to your CFO about returns
Here's the honest position. No Indian study we could find links agent training to recovery. Vendor claims exist, but they usually have no method or control behind them.
So don't promise a recovery uplift. Instead, present it in three parts.
- Required spending. The certificate, the code and the monitoring. It's not optional, and you're choosing how to deliver it most efficiently.
- Risk reduction. Fewer complaints and better call audit scores. These can be measured directly, and they're easier to attribute to training than recovery.
- A test. A pilot in two matched territories, measuring promise-kept rates and cost to collect, with a comparison group. If it shows a return, you'll have the first such evidence in the market, and you can expand it. If it doesn't, you've spent little.
CFOs generally respect that framing. It doesn't ask them to believe anything you can't show.
Our frontline training ROI framework and ROI calculator can help you structure this, and our post on measuring collections training goes through the metrics.
A timetable for the budget
This is our inference about timing. Only the April to March financial year is firm.
| When | What to do |
|---|---|
| November to December | Get headcount and churn from each agency. Take legal advice on who needs the certificate |
| January | Fill in the worksheet. Get quotes for training and delivery |
| February | Agree who pays with agencies. Agree the option with finance |
| March | Sign contracts. Book the first cohorts for April and May |
| April onward | Cohorts start. Report monthly on certified count and spending |
If you're an NBFC, don't leave the certificate cohorts until the second half of the year. Three attempts within nine months of training, with exams only on set dates, means you'll want plenty of slack before 31 December 2027.
Where Leap10x fits
I've said we sell in this area, so here's where we'd show up in your worksheet. Leap10x MicroLearning belongs in Block D: short conduct and process lessons on WhatsApp, with no app or login, in languages including Hindi, Tamil, Telugu, Marathi, Bengali and Kannada, with tracking by agency and person. Our pricing is a free Lite tier for up to 500 users with a 14-day trial, and custom Enterprise pricing, on a pay-for-active-use model. We won't quote a number in a blog post, because it depends on your scale.
HDFC Bank has had "32,000 collection agents trained" on the platform, with "95%+ learner engagement". We don't appear in Block B at all, because we don't award or replace the IIBF certificate. And we can't tell you we'll raise your recovery rate, because no one has shown that yet.
FAQ
How much should a lender budget for collections training? No reliable benchmark exists. Build it from your own headcount, churn, certificate costs and time off the field, using a worksheet like this one.
What does the IIBF certificate cost? IIBF lists an exam fee of ₹1,200 plus GST per attempt, and suggests a consolidated training fee of no more than ₹4,000 online or ₹6,500 offline. Check for current figures. Time off the field is usually the bigger cost.
Who pays for agents' certificates? Nothing we read settles this. It's a negotiation between lenders, agencies and agents.
Can we promise a recovery uplift from training? We wouldn't. Promise measures you can attribute, such as complaints and audit scores, and test recovery in a controlled pilot.
When is the budget decided? The financial year runs April to March. January to March is the likely window, on our inference, but check your own cycle.
Your numbers are the benchmark
Nobody's going to hand you a figure for what this should cost. That's fine. Get your own headcount and churn this month, and fill in the blocks. The first time you see person-days of certificate training written down, you'll have most of the budget conversation in front of you.
CTA: Want help filling in Block D with real numbers for your languages and roles? Book a Demo.
Frontline training
Training software built for deskless teams
Onboarding, SOPs, safety refreshers, and compliance evidence for workers with no corporate email and no company device — tracked by site, role, and shift.


