Q1 wrap: three patterns that changed what 'compliant' looks like on your floor

This newsletter is a quarter old. Eleven issues since 8 May — recording obligations, the DRA Code, calling hours, the threats that don’t sound like threats, the five expensive QA misses, borrower disputes, and last week’s polite agent who fails an audit cleanly.

Read individually, they are eleven separate rules. Read together, they are three patterns, and all three point the same way. Here is what shifted this quarter, and what it tells you to watch on your own floor.

Pattern one: the burden moved from conduct to evidence

The old compliance question was behavioural — did the agent harass, threaten, abuse, call at an unreasonable hour? The question a supervisor now has to answer is different: can you show me?

Look at what the rules actually ask for across Issues 2, 3 and 9. Identify yourself and the entity you call on behalf of. Do it on a recorded line. Stay inside 8 a.m. to 7 p.m. Keep the trail. Every one of those is an evidentiary requirement wearing the clothes of a conduct requirement. A floor can be genuinely well-behaved and still fail, because the finding isn’t “your agent was rude” — it’s “you cannot produce the call, or the disclosure inside it.”

This is the pattern with the most unpleasant arithmetic behind it. Sampling five per cent of calls was defensible when the test was whether your agents were broadly decent. It stops being defensible when the test is whether a specific call, on a specific date, contains a specific disclosure. You cannot sample your way to that answer.

Watch item for Q2: the revised recovery-conduct framework published in draft on 20 May 2026 carries a proposed effective date of 1 October 2026. It is still a draft and should be treated as one — but if it lands as proposed, the gap between “we behave well” and “we can evidence it” closes on a date that is now inside your planning horizon.

Pattern two: language became a coverage number, not an HR preference

Issue 4 was the quietest issue of the quarter and, on reflection, the one with the largest hidden exposure.

A collections call in Mumbai or Lucknow rarely stays in one language. If your audit process reads that call in English — or reads a translation of it — then whatever coverage percentage sits on your dashboard is describing a different call than the one the borrower was on. Tone, threat, implied consequence and half the compliance signal live in the code-switch. They do not survive translation, and they certainly do not survive an auditor who wasn’t listening for them.

What makes this pattern dangerous is that it produces no failures. A missed violation in an unaudited language doesn’t show up as a red flag; it shows up as nothing at all. Your scorecard looks clean because the risky calls were never really read.

The Q2 version of this question isn’t “do we have Hindi-speaking QA staff.” It’s: what proportion of our audited minutes were audited in the language they were actually spoken, and what proportion were audited in translation or skipped?

Pattern three: the grievance channel became the enforcement channel

Issue 10 set out the machinery — a nodal grievance redressal officer, a 30-day window, and escalation to the Complaint Management System and the RBI’s Integrated Ombudsman Scheme if the borrower isn’t satisfied. That framework is in force today.

Put it beside Issue 11’s polite-agent failures and the consequence becomes clear. The violations that never sound wrong — the skipped disclosure, the accommodating agent adjudicating a debt he has no mandate over, the settlement figure nobody authorised — don’t surface in your QA sample. They surface downstream, as a complaint, in a channel that is now instrumented end to end and has a clock running on it.

That has a commercial edge to it that agencies feel before they understand it. The lender owns the complaint number. So the lender pushes the audit obligation onto the agency, and the agency finds itself asked for evidence on a timeline set by someone else’s regulator. Grievance-side pressure travels down the chain, and it arrives as an audit request.

The same movement, three times

These aren’t three trends. They’re one.

Supervision is shifting from judging behaviour to reading records. Evidence, language coverage and grievance trails are simply the three places where that shift shows up on a collections floor. And in all three, the failure mode is identical: the floor is fine, the records are thin, and nobody notices until someone external asks a question the records can’t answer.

Three things to track in Q2

Not compliance resolutions — measurements. Each one is a number you can actually produce.

1. Your evidence rate, not your compliance rate. Pick twenty calls from last month at random. For how many can you produce the recording, with the identification disclosure audibly inside it, within an hour? That fraction is your real exposure number. Most floors have never calculated it.

2. Your language-adjusted coverage. Take your coverage percentage and multiply it by the share of audited minutes reviewed in the language actually spoken. The result is usually a good deal smaller than the number on the dashboard, and it is the honest one.

3. Your complaint-to-audit loop. When a grievance is escalated, does that call get pulled back into the QA sample and scored? If escalated complaints never re-enter your audit process, you are discarding the highest-signal calls you have — the ones an outsider has already told you went wrong.

The floors that will pass a 2027 audit aren’t the ones with the politest agents or the best intentions. They’re the ones who can answer a question about a single call from four months ago in under an hour, in the language it was spoken.