08 · Recognizing & Developing Talent¶
Most managers identify talent by noticing who reminds them of themselves, and then develop it by sending that person on a course. Both halves are wrong. The first produces a team that is demographically and cognitively narrow and quietly demoralised; the second produces a certificate and no change in behaviour. Recognising talent is a discipline of looking for specific observable signals — several of which look like inconvenience — and developing it is mostly about the work you assign, not the training you buy. This module gives you a performance and potential grid, a signal checklist, a development plan template built on the 70/20/10 split, and a worked development conversation.
1. Performance is not potential¶
The most expensive confusion in talent management. Performance is delivery in the current role; potential is capacity to succeed in a materially different one. They correlate weakly, and treating them as the same thing is how organisations promote their best individual contributor into a mediocre manager and lose both.
| Low potential signal | High potential signal | |
|---|---|---|
| High performance | Core strength. Excellent where they are. Deepen, reward, do not push upward against their wishes | Grow. Stretch now, or they leave within a year |
| Low performance | Address. Module 6 conversation — role fit or performance | Investigate. Almost always a fixable cause: wrong role, bad fit with you, unclear expectations, life circumstances |
Two notes on using this honestly. The "core strength" box is not a consolation prize — a team of people brilliant and content in their current roles is a functional team, and pressuring them into management because it is the only visible progression is a failure of your career ladder, not of them. And the "investigate" box is where most overlooked talent sits; low performance is a symptom with many causes, and only one of them is capability.
2. Signals of potential you can actually observe¶
Not "shows leadership." These are behaviours you can point to with a date attached.
| Signal | What it looks like concretely | Why it predicts |
|---|---|---|
| Learning agility | Changes their approach after feedback within weeks, not months | The single strongest predictor of success in an unfamiliar role |
| Asks about the why | Wants the business reason behind a task, not just the task | Will make good decisions when you're not there |
| Seeks disconfirmation | Goes and finds out where they're wrong; changes their mind out loud | Won't defend a failing plan to protect their ego |
| Scope creep, the good kind | Notices adjacent problems and fixes them unasked | Already operating a level up |
| Teaches | Others go to them, informally, and leave abler | Multiplier, not just a producer |
| Handles ambiguity | Makes progress with incomplete information rather than waiting | Senior roles are mostly incomplete information |
| Takes the unglamorous job | Volunteers for the necessary, unrewarded thing | Ownership rather than positioning |
And the signals that mislead, which is the harder half of this list:
- Confidence. Reads as potential and predicts almost nothing.
- Visibility. Being in the room with you is a function of geography and personality. Some of your strongest people are on a different site, in a different timezone, or simply quiet.
- Similarity to you. The most reliably distorting bias in this whole area. If your list of high-potentials all share your background, communication style or working pattern, the list is wrong.
- Long hours. Measures availability, not capability, and frequently measures poor delegation.
A cheap corrective: for each direct report, write down one instance of a potential signal from the table above with a date. Whoever you cannot fill in is not necessarily low-potential — they may just be someone you have not been watching.
3. The development plan¶
Development is 70% challenging work, 20% people, 10% formal learning. Most plans invert this, because the 10% is the only part with a budget code. Build the plan from the assignment backwards.
DEVELOPMENT PLAN
Person: ______________ Period: __________ Reviewed: __________
WHERE THEY WANT TO GO
In their words (not my paraphrase):
What they want that I know the organisation can't give:
STRENGTH TO DEEPEN
The thing they are unusually good at:
How the next 6 months makes them more so:
THE ONE GAP THAT MATTERS
(Only one. A plan with four gaps develops nothing)
The gap, stated behaviourally:
What "closed" would look like, observably:
70 · THE WORK
Stretch assignment that requires the gap to close:
What could genuinely go wrong, and what I'll do about it:
Decision rights (see Module 5 level 1-7): ____
20 · THE PEOPLE
Who they should be exposed to, and how I'll arrange it:
Who mentors, and on what specifically:
What I will do to make their work visible upward:
10 · FORMAL
Course/reading/certification, if any — and what it's FOR:
CHECKPOINTS
Dates: Format:
How we'll know it worked (observable, not "feels better"):
The "what could go wrong" box is the difference between a stretch assignment and a trap. A stretch that fails publicly with no cover from you is not development; it is an expensive lesson in not volunteering.
4. A worked development conversation¶
Hana manages a customer support function. Joel is her steadiest agent — high performance, three years in role, never puts himself forward, regularly ignored in talent reviews because he is quiet and works from a regional office. Hana has noticed three signals: new joiners informally go to him, he rewrote the refund guidance nobody asked him to touch, and he changed his approach to escalations within two weeks of one piece of feedback.
Hana: I want to talk about the next couple of years rather than this month's numbers. Where do you actually want to go? Not the version you think I want to hear.
Joel: Honestly, I don't know. I've been asked if I want to manage and I don't think I do. Which I assume means I'm stuck.
Hana: It doesn't, and I want to separate two things. Do you not want to manage, or do you not want to stop doing the work?
Joel: The second one. The managers here spend all day in meetings about the work and never touch it. That looks miserable to me.
Hana: That's useful, because there's a direction that isn't management. Can I tell you what I've actually observed? Three things. Every new starter in the last year has ended up asking you things rather than their buddy. You rewrote the refund guidance in March without being asked, and it's the version everyone uses now. And after we talked about escalation tone in February, your escalations changed inside two weeks — that one is rarer than you'd think.
Joel: I didn't think anyone noticed the refund thing.
Hana: It's the reason we're having this conversation. So the strength is that you make other people better at the work, and you improve the work itself. The gap — and there's only one I care about — is that all of that happens in a five-metre radius around your desk. You've never presented anything, never sat in a cross-functional forum, and nobody outside this office knows your name. That's what stops you being a principal-level specialist rather than a senior agent.
Joel: That's fair. The presenting part is genuinely unappealing.
Hana: Then let's make it as small as possible and real rather than practising. Here's what I'd propose for the next six months. The work: you own the redesign of the whole escalation playbook — that's your call at level 5, meaning you decide and I back it, and it needs sign off from Ops and Legal, which means you'll have to explain it to people who don't work here. The people: I'll introduce you to Sanjay in Ops, and I'd like Priya to mentor you on one specific thing, which is how she runs a room she doesn't chair. And I'll do this — every time the playbook comes up in my director's meeting, I'll say it's yours and I'll have you present it once, in the last month, not the first.
Joel: What happens if Legal blocks the whole thing?
Hana: Then we do it together and I say publicly it was my scoping that missed the constraint, because that will be true. What I won't do is take it back off you halfway.
Joel: What's the formal bit? There's usually a course.
Hana: Only if you want one, and only for something specific. There's a half-day on facilitating workshops that would actually be used in month four. That's it — I'm not sending you on anything to tick a box.
Note what makes this a plan rather than a chat: one gap, an assignment that cannot be completed without closing it, a named delegation level, a named mentor with a specific subject, an explicit commitment about visibility, and a stated failure position.
5. Failure modes¶
- The nine-box as a verdict. Grids are for prompting a conversation about a person, not for stamping them. Position changes with role, manager and life stage; re-examine at least annually.
- Development plans nobody re-reads. A plan with no checkpoint date is a document, not a plan. Put the review in both calendars now.
- Developing only the visible. Set yourself the test in section 2 — one dated signal per person. The blanks are your blind spots.
- Promising what you can't deliver. Never imply a promotion you do not control. Say what you will do — advocate, assign, make visible — and be explicit about what is not yours to give.
How It Actually Works¶
Performance and potential are different variables because they're estimated from different evidence with different reliability. Performance is measured from a track record of completed, comparable tasks — it's a retrospective, relatively low-noise signal because it's grounded in outcomes that already happened. Potential is a forward-looking prediction about how someone will perform in a different role or under different conditions than the ones that generated their track record — which means it's inherently an extrapolation, not an observation, and extrapolations are only as good as the signals used to make them. Conflating the two (promoting the best performer in the current role into a role requiring different skills) is the mechanism behind the well-documented "Peter Principle" pattern: performance in role A provides almost no information about performance in role B when the roles require different underlying capabilities, so using it as the sole promotion criterion selects for the wrong variable roughly as often as it selects for the right one.
Why the specific "signals of potential" (learning agility, how someone handles novel ambiguous problems, how they respond to feedback) are better predictors than raw current output. These signals target the underlying capability that generalizes across roles — the ability to build new mental models quickly from limited information — rather than the surface competence in one specific, already-mastered task. This maps onto a well-replicated finding in industrial psychology: general cognitive ability and learning agility measures predict performance across novel roles far better than performance in the current, already-optimized role does, precisely because the current role's performance has been shaped by months or years of practice effects that won't transfer to a genuinely new context.
Why a development plan needs the same specificity as the delegation contract, and fails for the same information-asymmetry reasons when it doesn't have it. A vague development goal ("build your leadership skills") gives both leader and report no shared checkpoint to calibrate against, so each independently estimates progress using their own incomplete information — the leader from occasional visible moments, the report from their own felt sense of growth — and the two estimates routinely diverge without either party noticing until a promotion decision forces the comparison, at which point the gap reads as a surprise disagreement rather than what it actually was: an unaddressed measurement problem from the start.
Exercise¶
Place every direct report on the grid in section 1, then justify each placement with one dated observation from the table in section 2. Any placement you cannot evidence is an impression, not an assessment — mark those, because they are where your biases live.
Then pick one person, ideally from the "investigate" box or someone you rarely see, and build the plan in section 3 with them — in a conversation, not beforehand. Hold yourself to two constraints: exactly one gap, and the 70% assignment must be work that already needs doing, not an exercise invented for their benefit. Book the first checkpoint before you leave the room.