Most performance management still works the same way. Set a target, wait for the result, react to the number. That is a reasonable system for finding out what happened. It is a poor system for finding out why, because it almost never looks at the behaviour that produced the result in the first place.
Across contact centres, quality teams typically review somewhere between one and five per cent of customer conversations, often lower once the review queue backs up. In sales organisations, most reps go weeks or months without a structured conversation about how they are actually selling, as opposed to what they closed. The number gets watched closely. The hundreds of moments that built it mostly go unseen.
What the scoreboard can't tell you.
An outcome metric (a closed deal, a resolved ticket, a retention rate) is what statisticians would call a lagging indicator. It confirms what already happened. It cannot, on its own, explain why it happened, and it cannot be coached directly, because by the time it exists, the behaviour that produced it is finished.
Economist Charles Goodhart made the sharpest version of this point in 1975, in an observation about monetary policy that escaped into management thinking almost immediately. Once a measure becomes the target, people start optimising for the measure rather than the thing it was meant to represent. It isn't a claim that people act in bad faith. It's a claim about incentives: if the only thing anyone is watching is the number, the number is the only thing that will reliably move.
“When a measure becomes a target, it ceases to be a good measure.”The idea now commonly attributed to Goodhart's 1975 observation
A sales team judged purely on quarter-end revenue will hit the number in ways that don't always serve the business: more discounting, more urgency-based closing, less time spent on the discovery conversations that build a durable pipeline for next quarter. None of that shows up in this quarter's revenue line. Most of it shows up in next year's retention numbers, usually after the manager who could have coached it has moved on.
The gap this creates is measurable. Organisations that run a formal, structured coaching process report around 91% quota attainment, against under 85% where coaching happens ad hoc and unstructured. The target isn't the difference. Whether anyone was watching, and shaping, the behaviour behind it is.
The behaviour hiding inside every number.
Every outcome metric has a set of leading indicators sitting underneath it: specific, observable things people do that make the result more or less likely. A closed deal is a lagging indicator. The discovery questions asked in week one, the objection handled cleanly in week three, and the follow-up sent within a day of the final call are the leading indicators that quietly decided it.
In a sales conversation
- Asking open questions before proposing a solution
- Naming the customer's constraint back to them accurately
- Handling the first objection without reaching for a discount
- Confirming a concrete next step before the call ends
In a complaint or service conversation
- Acknowledging the emotion in the room before addressing the facts
- Dropping the script once the customer is visibly frustrated
- Offering a specific next step rather than a vague reassurance
- Closing the loop, not just the ticket
In a difficult feedback conversation
- Leading with the specific behaviour, not a character judgement
- Making space for the other person's account before responding to it
- Agreeing what changes, and by when, out loud
- Checking in on it before the next scheduled review
None of this shows up on a dashboard on the day it happens. All of it shows up, eventually, in retention, in complaint rates, and in whether a promising hire stays past their first genuinely difficult year.
Roughly 97% of conversations are never reviewed by anyone. With manual review capped at a few per cent of interactions, the behaviour driving most outcomes is invisible until it has already shaped the result, usually weeks or months before a dashboard catches up.
Why so few of these moments get coached.
The behaviour is knowable. The problem is capacity. Most frontline managers are promoted for being good at the job themselves, not for being good at watching someone else do it. Only around one in nine sales managers has had any formal training in how to coach, and most say they simply don't have the time to do it consistently, even when they know they should.
It is the same story across customer-facing teams generally. Quality assurance programmes exist in the vast majority of contact centres, but manual review typically covers a low single-digit percentage of interactions. The rest only ever shows up in aggregate: the satisfaction score, the churn rate, the escalation count. By the time a pattern is visible there, it has usually been happening for months.
Rehearsal changes what shows up in the real conversation.
If behaviour is what predicts the result, the highest-leverage place to intervene is before the real conversation happens, not after it has already been scored. This is where structured practice has a genuine evidence base behind it, not just intuitive appeal.
A meta-analysis covering 65 studies of adult workplace learners found that people who trained through realistic, computer-based simulation showed measurable gains over comparison groups: roughly 11% higher declarative knowledge, 14% higher procedural knowledge, and a 20% lift in self-efficacy, meaning people believed, correctly, that they could actually do the thing under pressure. The same body of research points to why: specific, well-timed feedback is one of the strongest levers available, with simulation-based feedback showing a notably large effect on how performance improves afterwards.
That is the premise Perceptence is built on. Rather than waiting for the real conversation to happen and reviewing it after the fact, people rehearse the specific, high-stakes conversation first, against a realistic AI counterpart, and receive structured feedback on the actual behaviours involved before the score is on the line. We believe humans are the loop: the AI creates the rehearsal room, but it's the person who walks into the real conversation better prepared. Practice becomes a leading indicator a manager can act on, weeks before an outcome metric would ever have flagged a problem.
Three questions worth asking before the next review.
None of this means outcome metrics are wrong. A business still has to hit its numbers. But a target on its own only tells a leader whether they succeeded once. The behaviour behind it tells them whether they are likely to succeed again. Before the next round of performance reviews, it is worth asking three narrower questions instead of the usual one:
- What are the three or four behaviours that most reliably predict this particular outcome?
- How many of those behaviours can anyone in the business currently see?
- Are we coaching the behaviour itself, or only reviewing the result it eventually produced?
The organisations that answer all three honestly tend to find the same thing. The number was never the problem. It was just the last place they were looking.