Every growing firm has a handful of stars: the technical lead who unblocks everything, the client whisperer who never loses a renewal, the founder who “just sees around corners.” You celebrate them (you should). Then one day you notice the side effects:

• Every tough call routes through one inbox.
• “How we do it” lives in someone’s head.
• The rest of the team is executing, not learning.

That’s keyperson dependence: when performance, decisions or knowledge concentrate in too few hands. It feels efficient—until it isn’t. The day that person slows down, leaves, or just goes on holiday, momentum stalls. In governance terms, it’s a textbook risk that drags valuation and resilience. 

There’s a harsher framing engineers use: your bus factor—the number of people who’d need to disappear before a project collapses. If that number is “1,” you don’t have a team; you have a single point of failure. 

When the visionary leaves: cautionary (and current) examples

• Apple — the 1990s drift. After Jobs’ 1985 exit, Apple’s “innovation premium” collapsed as products meandered and investors lost faith. His 1997 return—and a tighter leadership model—rebooted discipline and invention. The lesson isn’t “find a genius.” It’s that innovation fell when the organisation couldn’t institutionalise judgement without him. 
• Apple — the open question today. Under Tim Cook, Apple has mastered scale and operational excellence, but debate continues around breakthrough innovation vs. relentless iteration (Vision Pro and services push notwithstanding). Even bulls concede Apple’s narrative now hinges on how it systematises AI-era invention without a Jobslike centre of gravity. 
• Disney — postWalt stall. After Walt Disney’s death in 1966, animation shrank, creative risk dried up, and the company lost momentum until a later leadership reboot. Vision concentrated in one person, and the pipeline atrophied without him. 
• Starbucks — the boomerang CEO. Schultz stepped down; expansion diluted the brand; performance sagged. He returned to refocus the core before handing off again—another case where systems hadn’t been strong enough to outlive the founder’s judgement. 
• NVIDIA — keyperson risk acknowledged. Jensen Huang’s fingerprints are on strategy, culture, and product bets (CUDA, AI acceleration). Even NVIDIA’s own filings flag risks tied to dependence on key personnel—standard legal language, but a useful reminder: if your CEO is the operating system, succession is an existential design problem. 

None of these argue against visionary leadership. They argue for baking the vision into systems so the company compounds—rather than pauses—when the visionary steps back.

How to avoid turning stars into chokepoints

1. Make the  work, not the person, the centre of truth

• Move critical knowhow into living playbooks (product config, pricing logic, recovery steps).
• Record judgement, not just steps: “Why we choose A over B when X is true.”
• Keep it close to the work—inside the tools people actually use (CRM/CPQ, ticketing, repo READMEs)—not in some dusty wiki.

(Internal link: “Sales Ops Isn’t Sexy—But It Might Save Your Sales Team” for how to embed commercial rules in CPQ, so margin protection isn’t optional.)

2) Raise your bus factor—deliberately

• Pairing & rotation on critical flows (quotes, migrations, major client runbooks).
• Backups for approvals—no single approver for pricing, exceptions, or releases.
• Crosstraining targets: aim for a bus factor ≥3 on anything revenuecritical. 

3. Design succession before you need it

• Treat succession like product: clear readiness milestones, shadowing plans, and “acting” drills (run a quarter without the star in the loop).
• Use interim charters when roles are vacant to avoid leadership vacuums postreorg.
• Boards/owners: measure leaders on bench strength, not just this quarter’s number. (HBR has years of evidence that poor succession is a top reason transitions fail.) 

(Internal link: “The Silence That Follows Organisational Restructure” on preventing the decision vacuum that stalls teams.)*

4. Institutionalise judgement

• Don’t just codify steps; codify decision criteria (risk thresholds, margin floors, “when to escalate”).
• Run premortems for big accounts and launches: If X person weren’t here, what would fail? Build that into the runbook.

5. Incentivise diffusion, not hoarding

• Reward enablement (docs shipped, teammates unblocked, oncall reductions)—not just heroics.
• Track dependency metrics (how many tickets/decisions hinge on one person). If the number’s rising, you’re building fragility, not performance.

(Internal link: “How Misused Metrics Kill Motivation” for picking measures that change behaviour, not just dashboards.)*

6. Communicate the risk in the language of money

• Keyperson dependence hurts valuation and resilience; buyers discount businesses where the founder is the product. Use that to justify the shortterm cost of crosstraining, tooling, and succession. 

A pragmatic 90day plan

• Weeks 1–2: Map the choke points.
  List the decisions/flows that currently need a specific person to move. Tag by revenue impact and risk.
• Weeks 3–6: Lift the knowledge.
  Cowrite the runbooks with the star(s). Pair on live work once a week. Move “why” notes into your actual systems.
• Weeks 7–10: Raise the bus factor.
  Rotate ownership on two critical flows. Add a backup approver. Run one “day without me” pilot.
• Weeks 11–13: Test the bench.
  Have the star step back from one decision domain. Measure speed, quality, and stress. Fix gaps. Repeat with the next domain.

The uncomfortable (but necessary) question

Apple soared when Jobs returned. It’s thrived under Cook’s operational excellence—but the world still wonders about the next “outofnowhere” product. NVIDIA’s rise is extraordinary—but how much of that trajectory depends on Jensen Huang’s ongoing presence?

If your answer for your own business makes you nervous, good. That’s your cue to start designing resilience that outlives individuals.