Succession planning has a reputation problem: it sounds like an event. Build the plan, present the plan, file the plan. In practice it is a process, a loop the organization runs on a cadence, and the organizations that get value from it are the ones that treat stage seven as seriously as stage one. Running that loop is HR's job; owning the judgments inside it belongs to the business.
Here is the full loop, stage by stage, with the failure points marked.
Stage 1: Decide which roles are critical
Everything downstream inherits the quality of this decision, so resist the two easy answers: "all director-plus roles" (too many, plan collapses under its own weight) and "just the C-suite" (misses the specialist whose departure would actually hurt most).
Rank roles by cost of vacancy. The questions that surface it:
- What decisions stop being made while this seat is empty?
- What revenue, relationships, or filings does this role personally carry?
- How thin is the external market for this skill set, and how long would a search run?
- Is this a single point of failure regardless of level?
The output is a short list, usually 15 to 30 roles for a mid-market org, each with a one-line justification. The justification matters: it is the difference between "your role isn't critical" being an insult and being a documented, revisitable decision.
Failure point: letting the list grow to flatter people. Every added role dilutes the attention the genuinely critical ones get.
Stage 2: Define the success profile
For each critical role, define what success requires: five to eight weighted dimensions, each observable. This is the stage most processes skip, and skipping it converts every later stage into opinion-trading.
A success profile for a plant manager might weight safety leadership and union relationships heavily; the profile for a VP of product might weight portfolio judgment and executive communication. Same company, different profiles. If your dimensions are identical for every role, they are measuring vibes.
Involve the incumbent and their manager, then sanity-check with one person the role serves. Incumbents know the invisible parts of the job; the people it serves know which parts actually matter.
Failure point: unweighted dimensions. When everything counts equally, a candidate strong in the three things the role barely needs scores the same as one strong in the three things it lives on.
Stage 3: Identify candidates wider than feels comfortable
Now names. The instinct is to write down the two people everyone already expects. Do that, then force the aperture wider:
- The strongest person one level below the obvious candidates
- Peers in adjacent functions who have led through comparable complexity
- The person quietly running the hardest cross-functional program in the company
- Anyone the 9-box exercise placed high on potential but who lacks a sponsor
Wider first-pass lists exist because succession decisions made years later keep landing on someone who wasn't the original favorite. The favorite leaves, or plateaus, or turns the job down. Optionality is the point of a pipeline.
The internal-first bias here is earned, not sentimental. Wharton's Matthew Bidwell, studying external hires versus internal moves, found external hires were paid roughly 18 to 20 percent more than internal people moved into the same jobs, received worse performance evaluations for their first two years, and exited at higher rates. The pipeline you build in this stage is what makes the cheaper, better-performing option available when the vacancy comes.
Failure point: visibility bias. Candidates who present well to executives get named; candidates who deliver quietly two floors down don't. The corrective is structural, not moral: require every critical role's candidate list to include at least one name from outside the incumbent's direct line.
Stage 4: Assess against the profile
Score every candidate against every dimension of the role profile, with evidence, on a small scale (four points is enough). Then calibrate: put the raters in one room and make them defend their scores against each other's standards, because a 3 from a tough grader and a 3 from a generous one are different numbers until calibration makes them the same.
Assessment output is two things per candidate: a scored profile, and a readiness window (Ready Now / 1 year / 2 to 3 years). The window is a commitment, not a mood. Ready Now means you would hand over the role rather than run a search.
Failure point: scores without provenance. A number with no evidence, no author, and no date cannot be challenged, and what cannot be challenged cannot be trusted. Keep the receipts. This is also where an auditable scoring model beats a spreadsheet: the history of who changed what, and why, survives.
Stage 5: Build development plans that close specific gaps
Each scored gap becomes a development action with three properties: it involves doing a piece of the target role under real conditions, it has a date, and it has an owner.
Weak: "improve financial acumen." Working: "owns the FY28 budget build for the division, presents to the CFO in November, readiness rescored in December."
Exposure assignments that reliably move readiness: running a budget cycle end to end, leading an integration or reorg workstream, holding an executive-escalation rotation, presenting to the board, covering the target role during the incumbent's sabbatical. Notice the pattern: each one produces evidence for stage 4's next pass. The loop feeds itself.
Failure point: development that is all inputs (courses, coaching, books) and no exposure. Inputs are fine as supplements. Nobody becomes board-ready in a classroom.
Stage 6: Run the review cadence
The plan is now alive, which means it starts dying immediately: people resign, get promoted, disengage, or change their ambitions. The review cadence is the maintenance schedule.
- Quarterly, per top-tier role: 30 minutes. Changes, risks, action status.
- Twice yearly, whole plan: rescore on new evidence, recalibrate, refresh readiness windows, retire completed or stale actions.
- On trigger events, immediately: resignation, reorg, or a named successor departing. Rerun every plan the event touches, and stress-test the chain: if A backfills B, who backfills A? Departures cascade, and a plan that has never been stress-tested tends to fail at exactly the second link.
Failure point: reviews without calendar enforcement. A cadence that lives in intentions gets eaten by whatever is urgent that week. Put the reviews on the calendar for the year, with named owners, in January.
Stage 7: Measure whether any of it worked
The loop needs a scoreboard, or stage 1 through 6 become a ritual. The metrics that tell you the truth, each covered in depth in succession planning metrics:
- Internal fill rate for critical roles. The headline number. If critical vacancies keep going to external searches, the pipeline isn't producing.
- Bench coverage. Share of critical roles with at least one genuine Ready Now candidate, tracked over time. Watch the trend, not the snapshot.
- Time-to-fill for critical vacancies. Planned transitions should be measured in weeks.
- Successor retention. If named successors keep leaving before the promotion, they either didn't know, didn't believe it, or got tired of waiting. Each cause has a different fix; all of them show up in this number first.
- Development action completion. The leading indicator. Actions completing on schedule now predicts readiness improving two quarters out.
Report these to the executive team on the same cadence as the plan reviews. What gets measured gets maintained; what gets maintained is what a board can actually rely on.
The through-line across all seven stages: succession planning works when it produces evidence and dates, and fails when it produces names and adjectives. Build the loop, keep it honest, and the next surprise resignation is a transition instead of a crisis.