Implementation

Coaching That Doesn't Stick

25 September 2026

One agent's probing skill across four weeks after coaching, staying flat between 21% and 24%

“We’ve spoken about this before.”

Every team leader has said it. The same agent, the same issue, the third or fourth coaching session in a row. The agent nods, agrees, promises to work on it. A few weeks later, the same feedback comes up again.

It’s easy to conclude that the agent isn’t trying. Sometimes that’s true. But far more often, the coaching itself has no way of telling whether it worked.

Why coaching doesn’t stick

The feedback is too broad. “Be more empathetic” or “improve your call handling” sounds clear in a coaching room. On the next call, the agent has no idea what to do differently.

There’s no follow-up. The session happens, a note goes in a file, and the next time anyone looks at that agent’s calls is the next scheduled review — often a month later, on a different handful of calls.

Nobody checks the same thing twice. One week the review is about the opening. Next time it’s about the close. The skill that was coached is never measured again, so nobody knows if it moved.

Progress depends on which calls are heard. If the follow-up review happens to catch two good calls, the coaching “worked”. If it catches two bad ones, it “didn’t”. Neither is true.

What a coaching loop needs

Coaching that sticks has the same three parts every time:

  1. One specific skill. Not “call handling”, but “asking one question to find the real problem before answering”.
  2. A clear starting point. How often does that skill appear on the agent’s calls today?
  3. The same measure, next week. Has it moved, on all their calls, not just the two that got reviewed?

Without the third part, coaching is a conversation. With it, coaching is something you can manage.

Find the one skill to work on

The useful view of an agent is not their overall score. It’s which specific skills are missing from their calls.

The Customer Interaction Indicator for one agent across 50 calls in The Heartbeat app: My Name 92%, Welcome 90%, Courtesy 90%, Name Verification 50%, Probing 18%, Cust Education 14%, Assistance 8%, Cushioning 6%, Active Listening 0%

From The Heartbeat app: Agent 4’s skills across 50 calls in the week before coaching. The opening is strong. Probing appears on 9 calls out of 50, and active listening on none. That’s where the coaching session should start.

A view like this makes the coaching conversation specific. Instead of “your handling needs work”, it’s “on 41 of your last 50 calls, you didn’t ask a question before answering. Here’s one to try.”

Then check the same skill again

The most important step is the one most teams skip: looking at the same skill, for the same agent, a week later.

The Customer Interaction Indicator for the same agent the week after coaching in The Heartbeat app: probing on 21 of 52 calls (40%), up from 9 of 50, and active listening on 9 of 52

From The Heartbeat app: the same agent, the week after coaching, 52 calls. Probing is up from 9 calls in 50 (18%) to 21 in 52 (40%), and active listening has started to appear. The opening hasn’t moved, because it didn’t need to.

If probing has gone from 18% to 40% of calls, the coaching landed. Say so — agents rarely hear that their effort showed up in the numbers.

If it hasn’t moved, that’s useful too. It means the next session needs to be different: a better example, a phrase they can actually use, or a closer look at whether something else is getting in the way.

Making coaching stick

1. Coach one skill at a time. An agent can change one habit in a week. They can’t change five.

2. Use their own calls. Play the moment from their own call where the skill was missing. It lands far better than a general reminder.

3. Give them the words. “Before I check that, can I ask — is the issue the amount, or the date?” A specific line gets used. A principle doesn’t.

4. Look again, same window, same skill. A week later, check the same thing. Movement in the weak skill is the only real proof the coaching worked.

5. Change approach when nothing moves. If the number is flat after two sessions, repeating the same feedback a third time won’t help.

How The Heartbeat handles this

The Heartbeat reviews 100% of your agents’ calls, in any department. Analysts + AI go through every recording, in Hindi, English or Hinglish, and every AI flag is checked by a trained analyst before it reaches you.

Each agent’s weakest stage and skill are tracked over any date range you choose, so you can pick the skill to coach this week and check the same skill again the next. Because every call is scored, the comparison covers all their calls, not whichever ones happened to be reviewed. Your data stays in India, hosted on AWS ap-south-1 (Mumbai).

Coaching doesn’t fail because agents don’t care. It fails when nobody checks whether it worked.