Key person risk is the portion of your revenue that depends on one person showing up — almost always you. The test is blunt: if you were out of action for two weeks, what would stall? Every function that answers yes is a concentration you are carrying silently, and most founders have never added up what those functions are actually worth. This audit does exactly that, in four steps, with a spreadsheet and honest inputs.
Key takeaways
- Key person risk is measured per function as annual revenue at risk multiplied by the probability you are genuinely unavailable in the next 12 months.
- The audit is four steps and one table: list what stalls, price it, estimate your absence odds, then multiply and sum.
- Exposure is usually concentrated, which is good news — two or three functions tend to carry most of the total, so the fix is small and specific.
- The recovery is 30 days against your top three lines, one action each: document it, delegate it, or automate it.
Why the gut-feel number is wrong
Ask a founder what happens if they take two weeks off and you usually get a list of tasks. Answering emails, sending invoices, running the weekly call. That list is not wrong, it is just cheap. The expensive exposures are rarely tasks. They are decisions nobody else is authorized to make, approvals that sit waiting, and relationships where you are the only name the client or supplier knows. A task you can hand over in an afternoon. A relationship you have owned for three years is a different kind of asset entirely.
The audit forces you to see both, in the same column, with the same arithmetic.
The key person risk audit, step by step
Step 1 — Inventory everything that stalls
Open a spreadsheet and make four columns: function, annual revenue at risk, probability of absence, expected annual loss. Then work through three sweeps. Tasks: what recurring work stops producing output. Decisions: what waits for your signature, approval, or judgment. Relationships: which clients, suppliers, contractors, or partners will notice within days that you are gone and start behaving differently.
Be boring about it. "Invoicing" and "the Henderson account" are separate lines because they fail differently.
Step 2 — Price the annual revenue at risk
For each line, ask what a two-week stall actually costs across a year. A stalled invoice is a cash-timing problem. A stalled proposal pipeline is a revenue problem that compounds quietly for months. A stalled client relationship is a renewal problem. Use real numbers from your own books where you can, and mark the ones you are estimating.
Step 3 — Estimate the probability of your absence
Not "do I take holidays." The question is whether you are out of action for two full weeks in the next twelve months — illness, surgery, burnout, a family emergency, a stomach bug that flattens you at the worst moment. Convert your answer to a rough decimal: a low chance is around 0.1, a moderate chance around 0.3, a high chance around 0.6. These bands are judgment, not precision, and that is fine. The point is to stop treating your own availability as a fixed constant.
Step 4 — Multiply and sum
Annual revenue at risk times probability gives you an expected annual loss per function. Add the column. That total is your key person risk exposure.
| Step | What you record | What it produces |
|---|---|---|
| 1. Inventory | Every task, decision, and relationship that stalls | The exposure list |
| 2. Price | Annual revenue at risk per line | A figure per function |
| 3. Probability | Odds you are out for two weeks, as a decimal | A weight between 0 and 1 |
| 4. Multiply and sum | Revenue at risk × probability, then total the column | Expected annual loss |
Worked example: a solo consultant
The numbers below are hypothetical. The arithmetic is the point.
Say a solo consultant carries four lines. Delivery for their largest client: if they vanish for two weeks, a milestone slips and payment is withheld, so they put the milestone at 20,000 at risk with a 0.3 probability of absence — an expected 6,000. Proposal writing, which only they do: a two-week gap quiets the pipeline for a quarter, so they put 30,000 of future work at risk at 0.3 — 9,000. Invoicing: 5,000 exposed to delay at 0.3 — 1,500. One subcontractor relationship only they manage: 10,000 at risk at 0.2 — 2,000.
Total expected annual loss: 18,500. Notice where it sits. The two biggest lines are not tasks. One is a relationship the client values, the other is a judgment-heavy activity the consultant has never treated as a system. Those are the two worth fixing first, and neither requires hiring anyone.
What to do with the number
An expected annual loss is a real input, not a feeling. Plug it into the free RADAR calculator at https://tools.exiuss.com/radar — a five-step diagnostic that turns six inputs into your Chaos Tax, the annual cost of manual work, urgent problems, missed revenue, and wasted tools, then breaks it down by category. Your audit total gives you a defensible figure for the missed-revenue and urgent-problem inputs instead of a shrug, and the breakdown tells you how much of your chaos tax is pure single-point-of-failure risk.
The free tier shows you the split. The full diagnosis is a one-time $7 with no subscription, and adds your chaos archetype, a sovereignty maturity level from Survivor to Sovereign, and a prioritized 30-day recovery plan tagged Systemize, Delegate, Automate, Eliminate. The SOP and RACI tools draft from your own inputs. A free 14-day EXIUSS trial is available if you would rather build than read a report.
The 30-day plan to cut your top three exposures
Sort your spreadsheet by expected annual loss and take the top three. One per week.
Week 1 — Systemize the biggest line. Write the procedure exactly as you do it, including the decisions you make without noticing. The test is whether someone competent could follow it cold. If it involves a judgment call, define the threshold that triggers an escalation to you.
Week 2 — Delegate the second. Hand the documented function to one named person with explicit authority, not just instructions. Authority is the part founders forget, and it is the reason delegation quietly reverts.
Week 3 — Automate the third. Automate only what is now documented. Automating an undocumented process just makes the mess faster. Invoicing, reminders, and status updates are usually the cheapest wins here.
Week 4 — Eliminate and re-score. Some lines will not survive the question "why does this exist at all." Kill them. Then re-run the audit with the same probability inputs and watch the total move.
Quick checklist
- Build the four-column register: function, revenue at risk, probability, expected loss
- Sweep tasks, decisions, and relationships separately
- Price each line against real revenue, not a feeling
- Assign a probability band to each and multiply
- Sort by expected loss and pick the top three
- Schedule one document, delegate, or automate action per week for 30 days
You are the single point of failure until you build the architecture that says otherwise. The audit gives you the figure. The next 30 days decide whether it grows or shrinks. Run yours at https://tools.exiuss.com/radar, and be honest — the diagnosis only works if the inputs are real.