Every team I've ever led has had a version of the same conversation: a manager telling a rep, "you need to hit your numbers," followed by silence, followed by the rep hitting the same numbers the next month for the same reasons they missed them this month. The manager walks away thinking they coached someone. What they actually did was supervise an outcome.
I didn't understand the difference until I'd been doing both for years without separating them. At Comcast, carrying a sales number every month, I noticed that the reps who consistently improved weren't the ones whose managers reviewed their numbers most closely — they were the ones whose managers had figured out why a particular call went sideways and built a specific habit to fix that one thing. Everyone else was getting performance management. They were getting development. Those look identical from the outside and produce completely different results over six months.
The test that actually separates them
Here's the test I use now, with my own team at IKEA and with anyone I'm coaching: after the conversation ends, did the person walk away with a number, or did they walk away with a skill?
"Your CSAT was 84% this month, target is 90%" is a number. It's accurate, it's necessary to say out loud at some point, and it changes nothing on its own. "In your last three escalations, you offered a solution before you confirmed you understood the complaint — here's what that sounds like instead" is a skill. It's specific enough to practice, and the person can actually go do something different tomorrow because of it.
Most performance conversations stop at the number because the number is easy to produce — it's already sitting in a dashboard — and the skill requires the manager to have actually watched or listened to the work closely enough to diagnose what's underneath it. That's the real cost of developing people instead of managing their performance: it takes more attention, not more authority.
Managing performance asks "are you hitting the target?" Developing people asks "what, specifically, would make hitting the target inevitable?" The second question is harder to ask and far more useful to answer.
Why this matters more once you've built something yourself
Running The JetLyfe alongside a full-time career forced this lesson on me from a different angle. There was no one to manage my performance — no dashboard, no manager reviewing my numbers. If I wanted to get better at sourcing manufacturers, or writing product copy that actually converted, I had to do my own diagnosis: what specifically am I doing wrong, and what's the smallest change that fixes it. That habit — diagnose the skill gap, not just the outcome gap — is the same habit I now try to build into every coaching conversation I have with my team.
It also taught me how much faster development compounds compared to supervision. A rep who's been told their number every month for a year has heard the same information twelve times and likely hasn't changed much. A rep who's been given one specific, practiced skill every month for a year has twelve new tools they didn't have before. The math isn't close.
What this looks like in practice
On my team, this shows up as role-play before it shows up as review. Before a rep handles a difficult account live, we've already run the actual conversation in a low-stakes setting, found the specific moment it tends to break down, and practiced the fix. By the time the real version happens, it's not the first time they've navigated that exact pattern — it's the fifth. That's a deliberate design choice, not an accident, and it's one of the places I've leaned hardest on AI-assisted training tools: building realistic role-play scenarios and reinforcement scripts at a volume no single manager could generate by hand for every rep, every month.
None of this means performance numbers don't matter. They're the scoreboard, and ignoring the scoreboard is its own kind of leadership failure. The mistake is treating the scoreboard as the conversation, instead of treating it as the reason you're having one.