The Founder Dependency Audit
Not how busy you are. How concentrated you are. Those are different problems and only one of them gets better if you work less. This one measures how much of the business runs through a single person, what it would cost to change that, and how long you could vanish before anybody outside noticed.
Excel (.xlsx), eight sheets, and a 12-page PDF. No email, no signup, no drip sequence.
This is not a complaint about how hard you work
The register is going to read like an indictment for about twenty minutes. Line after line with your name in the owner column, most of them marked as things nobody else can do. The instinct at that point is to feel either defensive or slightly heroic, and both are wrong, because neither one is a finding.
Here is the finding. Every one of those items ended up with you because at some point that was the fastest option available. It usually was. What nobody did was revisit the decision once the business grew past the point where it stayed true — because there is no meeting where that gets revisited and no number that makes it visible. This is that number.
You can live with founder dependency indefinitely, and plenty of people do, profitably, for decades. It becomes somebody else’s problem the moment you want to sell, raise, borrow, hire a real leadership layer, or take eight weeks off. An acquirer calls it key-person risk. An insurer prices it. A strong senior hire notices it in the second interview and takes the other offer.
Three kinds of dependency, three different fixes
Most handovers fail because all three get treated as the same problem — usually as a documentation problem, which only one of them actually is. Writing a document does not move a decision dependency. It moves a knowledge dependency, and it makes everybody feel like something happened.
Things wait for your judgment. Approvals, scope calls, pricing exceptions, who gets hired.
What it takes to move: A written limit, not a document. “You decide up to X. Above X you ask me.” Then you honour it, including the time they decide something you would not have.
Only you know how it actually works — the pricing logic, why that client is handled differently, where the bodies are.
What it takes to move: The one kind documentation genuinely fixes. Also the one that never happens, because it is never urgent until the week it is catastrophic.
A customer, supplier or partner deals with you and quietly considers that part of the arrangement.
What it takes to move: Months, and a deliberate handover with the other party in the room. It cannot be documented and it cannot be done quietly. Slowest to move and worth the most.
Think about the last handover that did not take. If it was a decision dependency and you answered with documentation, the person read it, understood the process perfectly, and came and asked you anyway — because what they lacked was not information, it was permission and some evidence that using it was safe.
The five numbers
The workbook computes nine signals off your register. These are the five that decide whether you have a problem, and the last one is the only one anybody outside your company will ever ask you about.
| Signal | Flags at | What it means |
|---|---|---|
| Share of load that is you | Above 30% | Share of all recurring hours that run through one person. Past a third you are not the founder of this business, you are its operating system — whatever the org chart says. |
| Items only you can do | Anything above zero, examined | The list that decides whether you can take a holiday. Not whether you will. Whether you can. |
| Money on a single owner | Anything above zero | Revenue or compliance depending on one person being reachable. This is the number an acquirer, an insurer or a bank finds in about ten minutes. |
| Days before it shows | Under 14 days | How long until the first thing only you can do comes round again. Not when it becomes a crisis — when it starts. The crisis is usually a week behind. |
| Hours to hand it all over | Context, not a verdict | The whole bill in one number. Almost always smaller than people expect, and almost always larger than the time anyone has set aside for it. |
Four ways a handover quietly fails
Almost invisible from the inside, because from where you sit it looks like you delegated and they did not step up. What actually happened is that you gave someone the work and kept the right to be disappointed with how they did it, without ever stating the standard. Three of these four are yours, which is annoying and also good news, because it means they are fixable.
| Pattern | What happens | The fix |
|---|---|---|
| The undefined edge | You hand over approvals but never name the limit. Every unusual case comes back to you, and unusual cases are most of them. | Write the number. Then live with what happens below the line, including the one you would have called differently. |
| The invisible standard | They do it, you redo it. Sometimes silently, which is worse — they think it went fine and you have quietly concluded they cannot do it. | Say what good looks like before, not after. If you cannot articulate it in advance, the standard lives in your judgment and this is a knowledge problem wearing a task costume. |
| The rescue | It starts going wrong and you step back in. The problem gets solved and the handover ends, permanently, without either of you noticing. | Decide in advance what failure you will absorb. If the answer is none, you are not delegating. You are auditioning them. |
| The half-handover | They own the work; you own the relationship it touches. The client keeps calling you, you keep relaying, and the job now takes two people. | Move the relationship or do not move the work. A half-handover costs more than no handover, and it is the most common state in a growing company. |
The handover that works
Nothing original — it is how anyone has ever been taught anything practical. Worth writing down anyway, because under time pressure the two middle steps are always the ones that get skipped, and they are the two doing the work.
You do it, they watch
Narrate the judgment out loud, not the mechanics. They could have read the mechanics. The reasoning only exists in your head and this is the one step where it gets out.
You do it together
They drive, you are in the room. The point is that they hit the first awkward case while you are still there, and the awkward cases are the entire job.
They do it, you review after
They decide and tell you. You do not reverse it unless it is genuinely damaging. Reversing here teaches them the step was theatre, and you are back to step one permanently.
They do it. Full stop.
Write the limit down somewhere other than your memory. An undocumented handover reverts the first time that person is on holiday and somebody asks you instead.
About six weeks of low-grade attention per item. Which is why the workbook asks you to estimate hours-to-hand-over and works out the payback: some items are not worth it, and you want to know which ones on purpose rather than by never getting round to any of them.
What you should never hand over
Handing over everything is not the goal and it is not even a good outcome. A founder who has delegated all judgment has not built a resilient company — they have built one with an expensive figurehead. Usually worth keeping: what the company will not do, the pricing philosophy (not the prices), who joins the leadership layer, the first serious conversation when something goes badly wrong with a client, and anything where your judgment genuinely is the product.
“I like doing this one” is a legitimate reason to keep something. It is just not a business reason, so write it in the register as what it is and stop pretending nobody else could. Then apply the only test that matters: if you were unreachable for two weeks, would this need to happen anyway? If yes, it needs a deputy even if you go on doing it yourself ninety percent of the time.
What costs you is the third state — where something is neither handed over nor deliberately kept, and you simply carry it while intending to sort it out later.
What is in the workbook
- Start Here
- The legend, the colour convention, and two rules for filling in the register without lying to yourself.
- Register
- One row per recurring decision, approval, task, relationship or piece of knowledge. Nine columns are yours; five more compute hours a month, whether it is you, the leverage of handing it over, and a plain-language flag naming what is wrong with that row.
- Exposure
- Nine signals read straight off the register, each with an OK or a CHECK. Do not type here.
- Two Weeks Out
- You are unreachable for a fortnight. What stalls in the first three days, inside the first week, by the end of it, and what is waiting when you get back.
- Handover Plan
- Ten rows. What, to whom, by when, and what “handed over” actually means — with the payback in months worked out for you.
- Decision Rights
- Who decides what, by function, and what they need before they can. The one page here that outlives the audit.
- Thresholds
- The numbers Exposure judges against, each with the reasoning beside it.
- Snapshot
- The one you send back. Seven numbers fill themselves in.
It ships with six example rows belonging to an invented founder called Alex Rivera, so Exposure has something to chew on and you can see what a flag looks like before your own numbers start doing it to you. Delete them and put your own name in the box before you start.
When you have finished it, send it back
The last sheet is Snapshot. Seven numbers fill themselves in from your register — including days before your absence shows, and the total hours it would take to hand over everything only you do. Three boxes ask what no spreadsheet can work out: who you are, what you have already tried to hand over and what happened, and what you would do with the hours if you got them back.
Fill those in and email the file to connect@cnnctd.work. We read it before we talk, so the first conversation starts from your register rather than an hour spent building one out loud.
That third question is the one we care most about. Founder dependency is only a problem if the hours would go somewhere better. If the honest answer is that you would fill them with the same kind of work, we will say so — and that is a much cheaper thing to find out on a spreadsheet than on an invoice.
This is the operating half of Discipline Stacking — capability built by layering a few practices that hold each other up, rather than adopting a framework wholesale and hoping the culture shows up with it.
The other two
Three tools, three failures that feel identical from the inside — the vague sense that everything takes longer than it should. Worth knowing which one you actually have, because the fixes have nothing in common.
The Operating Cadence
Whether the record keeps up with reality, or the board is a description of last Tuesday.
Workbook and field guide →
Measures ForesightThe RAID Log
Whether anything that goes wrong was ever anticipated, or every problem arrives as a surprise and gets handled well.
Workbook and field guide →
All three are free, ungated, and listed at the toolkit index.
Most founders already know what this file will say.
What they do not have is the number, in writing, in a form they can hand to a co-founder or a board without it sounding like a complaint. That is what this is for. What you do about it afterwards is a separate and much harder question, and if you want somebody in the room for that part, this is where to start.