Succession Planning With Data, Not Gut Feel

Open the file. The one called “Succession 2026 v4 FINAL”.

You’ll find an org chart with color-coded boxes, three names under every director-level role, and a column that says “ready in 1-2 years” repeated in almost every row. It was presented eight months ago. Someone said “great work.” Nobody opened it again.

The uncomfortable question: if your Director of Operations resigns tomorrow, is that file worth anything to you?

The honest answer is usually no. What you have isn’t a succession plan. It’s a list of names someone wrote from memory, with no data behind it. And a list of names is not coverage.

A succession plan isn’t a list. It’s a metric

A real succession plan answers a single question: what percentage of my critical roles has someone with a validated profile who is ready to step in?

If you can’t answer that with a number, you don’t have a plan. You have good intentions in slide format.

That question breaks down into four decisions, and each one has its own way of going wrong: which roles are critical (you almost always pick them badly), how much coverage you have in each (you almost never measure it), how you classify people (almost always with a misused 9-box), and what you do about the uncovered roles (almost always nothing).

Let’s take them one at a time.

Step 1: identify critical roles, not senior roles

The founding mistake in nearly every succession plan is confusing “critical” with “hierarchical.” If your plan covers the executive committee and stops there, you’re protecting the org chart, not the business.

A role is critical when its vacancy does real damage. And real damage isn’t distributed by seniority level. These are the three questions that define criticality; a role makes your list if the answer is yes to at least two of them.

What breaks if this person doesn’t show up tomorrow? Not “which tasks go undone” — what stops. Revenue, deliveries, regulatory compliance, the client relationship that carries a third of your billing. There are individual contributors with nobody reporting to them whose exit stalls the monthly close. There’s one specialist who knows how to operate the bank integration. Neither of them appears anywhere near the top of the org chart.

How long does it take to replace them from outside? A role the market can fill in four weeks is inconvenient. One that takes six months to fill and another six to become productive is critical, even if the title sounds modest.

How much knowledge lives only in their head? Undocumented processes, client relationships, accumulated judgment. When none of that is written down anywhere, the person doesn’t occupy the role. They are the role.

Run that filter and the final list won’t match the org chart. Operational and technical roles you never considered will make it in, and management positions that already have three possible replacements will drop out. That mismatch is information in itself: it’s the first time you’re looking at risk instead of status.

Step 2: bench strength, or how deep your bench really is

In sports, the bench is the people sitting there waiting to come in. Bench strength is how well those substitutes play when you have to pull a starter. Translated: for each critical role, how many people on your current payroll could take it, and how soon?

The base metric is the coverage ratio:

Coverage = validated successors / critical roles

A ratio of 1.0 means you have, on average, one successor per critical role. Sounds fine, and it’s fragile: the average hides the fact that some roles have three candidates and others have zero. The number that actually matters is the count of roles with zero successors. That’s your real risk, and it fits in an email.

The word doing all the work here is validated. A validated successor isn’t someone their boss happens to like. It’s someone about whom you can state three things with evidence.

• They perform well today. You have assessments, not impressions.
• They have the profile of the destination role, which is not the profile of the role they’re in now. How that gets calculated — the psychometric profile of each level, and why the best specialist is rarely the best manager — we cover in full in Potential Mapping.
• They want the job. Yes, you have to ask them. A surprising number of “successors” have zero intention of taking the role you assigned them in your deck.

If any of the three is missing, that person isn’t coverage. They’re a hope with a first and last name.

Step 3: the 9-box, and the three traps everyone falls into

The 9-box is the most used tool in succession planning and also the most misused. It comes from the matrix McKinsey built for General Electric in the seventies, designed to decide which business units to invest in. Then someone swapped “business unit” for “person,” and here we are.

It’s a three-by-three grid. One axis is performance (how they’re doing today), the other is potential (how far they could go). Everyone lands in one of nine cells. Used well, it separates “are they good at what they do?” from “would they be good at something else?” Those aren’t the same question, and confusing them is the origin of half the bad promotions out there.

Now, the traps.

Trap 1: both axes are measured with the same instrument — the manager’s opinion

This is the original defect. If the manager rates performance and the manager also rates potential, you don’t have two axes. You have one, drawn twice. The matrix collapses onto the diagonal: whoever performs well gets labeled high potential, end of story.

The 9-box only works when each axis comes from a different source. Performance: defined criteria, results, evidence. Potential: a psychometric measurement of the profile, independent of who you find likeable. That’s what you solve by measuring the entire talent lifecycle with the same engine: without a common, comparable instrument, the second axis is smoke.

Trap 2: “potential” without saying for what

“High potential” is an empty label if you don’t finish the sentence: potential for which role? Potential isn’t a property of a person, like height. It’s the relationship between a profile and a destination. Someone can have enormous potential for a deeply specialized technical role and mediocre potential for managing people. With a single generic potential axis, you’re averaging two things that point in opposite directions.

The fix: don’t build one 9-box, build one per destination family. One for the move into management, one for the technical track. You’ll watch the same person change cells depending on where you’re pointing them, and that’s exactly the data you need.

Trap 3: the matrix is a snapshot and you treat it as a verdict

The 9-box tells you where someone is today, with the work they do today and the boss they have today. It’s not a permanent diagnosis.

The symptom: you use the cell to label instead of to decide. “So-and-so is a 2-2” stops being a reading and becomes an identity, and then a self-fulfilling prophecy. Nobody invests in developing someone who has already been filed away. Practical rule: the cell doesn’t leave the calibration room. What leaves the room is the action.

Step 4: readiness, or the only column your CEO cares about

You have your critical roles and your candidates classified. What’s missing is the question your CEO will ask the day the Director of Operations resigns: how soon?

That’s readiness, and it only takes three answers. Three, not five: the temptation to add nuance is what turns the tool into decoration.

Ready now. Could take the role in under 90 days. Has the profile, the performance, and the context. The acid test: would you put them in tomorrow without calling a recruiter? If your answer has a “but” in it, they’re not ready.

Ready in 1-2 years. Has the profile, is missing experiences: exposure to the executive committee, owning a budget, managing someone more senior than them. And here’s the part almost nobody does: if you write “1-2 years,” you also have to write which specific experiences close that gap and who is accountable for making them happen. Readiness with no development plan attached is an elegant way of saying “I don’t know.”

Not a successor for this role. That’s not a judgment about the person, it’s a judgment about the fit between a profile and a destination. Someone who can’t succeed the commercial director may be an excellent successor for the technical one. That distinction is what separates a development system from a blacklist.

The bias to fight is optimism: everyone wants to write “ready in 1-2 years” because it sounds like there’s a plan without obliging anyone to do anything. If more than half your rows say that, your plan isn’t calibrated. It’s being polite.

What you do about the uncovered roles

You finish the exercise, you count, and there are seven critical roles with zero ready successors. This is where the plan earns its keep, because so far all you’ve done is describe a problem. There are four plays, and they’re mutually exclusive: pick one per role.

Build. You take the best internal candidate with a fitting profile and design the path: which experiences they need, in what order, with which sponsor, with what review date. The cheapest play and the slowest. Works when the incumbent isn’t leaving tomorrow.

Buy. You go to the market now, before you have the vacancy. Cultivating a relationship with two or three external candidates for a role that isn’t open yet buys you months of advantage the day it opens. And if you’ve written down the target profile, you can assess them with the same instrument you use on your internal people and compare them on the same scale.

Redesign. If a single role concentrates three functions nobody else has mastered, the problem isn’t a missing successor. It’s that the role is badly designed. Split it. A role that’s impossible to succeed is an architecture defect, not a talent one.

Shield. When there’s no successor, no market, and no way to redesign — the classic case of the founder CTO — you reduce the damage: document what lives only in their head, distribute their key relationships, retain them with a clear-eyed view of what’s at stake. That isn’t a succession plan. It’s a containment plan, and it beats pretending you have the former.

What isn’t a play: leaving the cell blank and looking at it again next year.

The science: why intuition fails precisely here

There’s a reason promotions go wrong systematically, and it isn’t that managers are stupid.

Laurence Peter stated it in 1969 and it stuck as the Peter principle: in a hierarchy, every employee tends to rise to their level of incompetence. It sounds like an office joke, but for fifty years it was folklore with no hard evidence behind it.

In 2019, Benson, Li and Shue put it to the test with real data on promotions from salespeople to sales managers and published the result in the Quarterly Journal of Economics. The finding: companies promote based on performance in the current role, and that performance does not predict performance as a manager. To gain a mediocre manager, the organization loses its best salesperson.

Translated to your plan: when the criterion is “whoever performs best today in the role below,” you’re not using an imperfect shortcut. You’re using a signal that points at the wrong place. The second axis of the 9-box exists to correct that, and it only corrects it if you measure it with something other than the same opinion you already used for the first. Why a personality profile predicts what prior performance doesn’t is in the OCEAN model explained.

What you can do this week

Without buying anything, without a committee, in a few hours.

One. Build the list of critical roles using the three questions above. Start with a blank sheet, not with the org chart. If you end up with more than 15 roles in a mid-sized company, you didn’t filter. You made a wish list.

Two. Next to each role, write the names of the successors. Now cross out everyone for whom you can’t show evidence of all three conditions — performance, destination profile, genuine interest. Count the empty rows. That number is your current succession plan, measured for the first time.

Three. Take your most exposed critical role and write the profile of the destination, not of the current incumbent. Which traits make someone perform there. You’ll discover you had never written it down, and that half the debate about who succeeds them was people comparing against different definitions.

---

Succession doesn’t fail the day someone resigns. It fails two years earlier, in the meeting where three names got written into a file and nobody asked what evidence was behind them.

A plan that works isn’t longer or prettier than the one you have. It’s more honest: fewer names, better measured, with a coverage number you can say out loud without looking away.

Meet Talento Index, the assessment of your current team with the same engine you use to hire. Start by measuring the people you already have: your bench already exists, you just don’t know yet who’s on it.