You are paying yourself a third of what you paid your last hire. You describe this to investors as commitment, and to yourself as temporary.
It has been fourteen months. The company’s largest single-point exposure is the personal solvency of the person who cannot be replaced, and nobody has ever put a number on it.
Why these three models
The decision is what to pay yourself, and when to change it. The features that fire are an outcome that can end the company through a route that never appears in the financial model, a number that will be evaluated against a comparison nobody has stated out loud, and a question about what is normal that founders answer from folklore.
Three lenses. Ruin sizing produces a random answer about the exposure you are carrying personally. Reference dependence produces an equilibrium answer about how the number is read. Base rates produce a cycle answer about what is actually normal, which is the input that turns this from a values conversation into an arithmetic one.
1. Your personal runway is a company exposure
Start with the failure mode, because it is the one that never gets modelled.
A founder who cannot pay their rent takes a consulting engagement, or moves cities, or accepts a job. None of those decisions are made in a board meeting and all of them end the company as reliably as running out of corporate cash. The exposure sits on the balance sheet of a person, so it is invisible to every instrument the company uses to watch itself.
Where an outcome can remove you from the game, the relevant quantity stops being the average and becomes the size of the worst case. That principle is routinely applied to customer concentration and to a single large bet, and almost never applied to the founder’s own solvency, which is the most concentrated exposure in the business.1
The measurement is one line and most founders have never written it. How many months can you personally continue at the current salary before you would have to change something. Not how long you intend to; how long you can. If that number is smaller than the company’s runway, the binding constraint on the business is not the bank balance, and every plan built on the bank balance is describing a longer horizon than you actually have.
Your personal runway is the shorter of the two numbers and it is the one nobody reports.
There is a second-order version worth naming. A founder under personal financial pressure makes different decisions: shorter horizons, more willingness to take a bad term for faster cash, more reluctance to fire someone whose exit would raise their own workload. Those are not character failures, they are what pressure produces, and the company pays for them without ever attributing the cost.
Put numbers on it, because the abstract version never moves anyone. Rent in a decent Kampala neighbourhood. School fees arriving in three termly lumps rather than twelve even instalments. A medical event with no cover. A vehicle that has to keep running. Write the annual total, divide by twelve, then write what you actually take. The gap is not a lifestyle question. It is the monthly rate at which you are drawing down a reserve, and when the reserve reaches zero the company loses its founder to a consulting contract. Most founders who run this find the fees month is the binding one, not the average month, which is why a salary set against the average still produces a crisis in January.
If you cannot raise the number this quarter, the cheaper partial version is to move the personal fixed costs instead. A smaller lease. Fees spread by arrangement with the school rather than by lump. One vehicle instead of two. That buys the same months of personal runway at no cost to company cash. Do it explicitly, with the months written down, so it counts as a decision rather than as quiet erosion.
Resist the third option, which is to accrue the shortfall as unpaid salary owed to you. It reads as free runway and it is not. Accrued founder compensation sits on the balance sheet as a liability that a new investor will ask you to waive or convert, and the ask arrives at the least convenient point in a term-sheet negotiation. By then the figure is large enough that waiving it is a real loss and insisting on it is a real signal.
2. The number is read, and against something you did not choose
The second lens explains why this is not simply a matter of taking what you need.
Any figure is evaluated relative to a comparison point rather than in absolute terms, and the comparison is supplied by the reader.2 Your salary will be read against at least three different references, and they point in different directions.
An investor reads it against what founders at your stage typically take. Too high and it reads as extraction. Conspicuously low and it reads, to an experienced investor, as either a personal-runway risk or a signal that you will need to renegotiate soon, which is a conversation they would rather have now.
Your team reads it against their own pay. A founder taking dramatically less is sometimes read as commitment and sometimes as a standard being set for everyone, and which of those depends on whether the founder has other assets, which the team will speculate about either way.
And you read it against your own previous income, which is the reference that does the most damage. A large drop from a prior salary is experienced as an ongoing loss rather than as a chosen position, and that framing is expensive over a period of years in ways that arrive as resentment rather than as a line item.
There is an uncomfortable implication in how the low number reads. A signal carries information in proportion to what it would cost someone else to imitate, and a conspicuously low salary is cheap for a founder with savings or a second household income and expensive for one without.4 Underpaying yourself signals your circumstances far more reliably than it signals your commitment.
The practical response is not to optimise for one reader. It is to state the number and its basis once, in writing, to the board: this is the figure, here is what it is benchmarked against, here is the trigger at which it changes. A stated basis replaces three speculations with one fact, and the trigger removes the annual conversation.
The distinction that conversation usually blurs is between the salary and the total household draw. A founder taking a modest figure while the company covers the vehicle, the fuel, the phone and the travel is not taking a modest figure, and everyone who processes those expense claims knows it. That gap is where the goodwill from a visible sacrifice gets spent, quietly and at full price. State the whole draw as one number when you state the basis. A slightly larger salary with nothing attached is read better than a small one trailing company-paid costs, and it is the only version anyone can benchmark.
When a board member pushes back on a raise, do not argue whether it is deserved. Restate it as the exposure it removes and give them two numbers: months of personal runway before, months after. An investor who then declines to fund the difference has told you the risk is acceptable to them, which is useful and should be minuted.
3. What is actually normal
The third lens converts the first two into a number, and it is the one founders skip because the answer feels like something they should already know.
Under a weak specific signal, the sensible estimate sits close to the underlying frequency and moves away from it only in proportion to how diagnostic your particular information is.3 Applied here: start from what founders at your stage, in your market, with your funding position actually take, and adjust for your specifics rather than reasoning from first principles about virtue.
That base rate is obtainable and rarely obtained. Three founders one stage ahead of you will answer the question directly if asked directly, and the answer is more useful than any published survey because it is drawn from your own market. In East African markets in particular, the imported benchmarks are close to useless, because the cost base, the currency exposure and the funding norms are all different, and a number lifted from a US survey is not conservative or generous, it is simply unrelated.
The output should be a band rather than a figure: the level below which you are creating a personal-ruin exposure, and the level above which an investor would reasonably object. Most founders discover their current number sits below the bottom of their own band, which is the finding, and it is a finding about risk rather than about deserving.
Ask the question in a form that returns a number. “What do you pay yourself” invites a range and a caveat. “What hit your account last month, and does the company cover anything else” gets an answer you can use. Ask three, not one, and choose three whose funding position resembles yours rather than three you admire, because admiration and comparability are usually inversely related here.
One correction the band needs in this region: set it in the currency your obligations are in. A founder whose rent, fees and loan repayments are in shillings, drawing a salary denominated in dollars from a foreign account, has taken a currency position on top of everything else. A ten percent move in the rate moves the floor of the band with it, and the floor is the number that was protecting you. Choose the currency first, then the figure, and rebase the floor when the rate moves rather than at the annual review.
What the three say together
- Write your personal runway in months at the current number. Compare it to the company’s. The shorter one is the real constraint.
- Ask three founders one stage ahead what they take. That is your base rate and it takes three messages.
- Set a band, not a figure: the floor below which you are an exposure, the ceiling above which it reads as extraction.
- State the number, its basis and the trigger that changes it, once, in writing, to the board.
Where they disagree
The exposure argument and the signal argument point in opposite directions, and the conflict is the whole difficulty.
The exposure argument says raise it: an underpaid founder is a single point of failure and the company should pay to remove that. The signal argument says that in a market where investors watch for founders extracting cash, and where your own team can see the number, raising it carries a cost that is real even when the raise is justified.
The resolution is that the two are not weighed against each other, they are sequenced. Establish the base rate first, then position within it. Inside the normal band, the signalling cost of moving is small and the exposure reduction is large, so move. Outside the band, the signalling cost dominates and you should change something else instead: the runway, the raise, the burn. Founders usually run this backwards, treating the signal as the binding constraint without ever establishing what normal is, which means they are managing an impression against an unknown standard.
What none of them contain
None of the three handles the founder with other assets or a partner’s income. That changes the exposure entirely and it changes nothing about how the number is read, which means two founders in the same company can face genuinely different versions of this decision and the fair answer is not the same number.
None of them prices the co-founder comparison. All three treat this as one person’s decision, and in practice the hardest version is two founders with different obligations, different outside options and the same title, where any answer other than identical requires a conversation most teams avoid until it is overdue.
And one property the ensemble will not produce: the number changes what you are willing to hear. A founder who has taken a large personal sacrifice becomes less able to consider shutting the company down, because the sacrifice is only redeemed by continuing. None of these lenses sees that, and it may be the most expensive thing on this page.
The one action that survives the ignorance: today, write down how many months you can personally continue at your current salary. If that number is smaller than the company’s runway, you have found the binding constraint on the business, and it is not in any document the company produces.
Who has to move
The founder, and the reason it does not happen is that raising your own salary is the one proposal nobody else in the company can make on your behalf. The instinct is to defer it to the next round, which converts a solvable problem into a dependency on an event you do not control. The cheapest first test is three messages to founders one stage ahead asking what they pay themselves. Almost all of them answer, and the number is usually higher than the one you have been calling responsible.
Sources and notes
- Michael J. Mauboussin, The Success Equation: Untangling Skill and Luck in Business, Sports, and Investing, Harvard Business Review Press, together with the standard treatment of absorbing barriers. The principle used in section 1 is that where an outcome removes you from the game, the expected value across outcomes ceases to be the decision-relevant quantity and the size of the worst case governs. Applied here to personal rather than corporate solvency, which is an extension of the standard framing rather than a result in the source.
- Sanjit S. Dhami, The Foundations of Behavioral Economic Analysis, Oxford University Press. Reference dependence, under which outcomes are evaluated as movements from a comparison point rather than in absolute terms, and the asymmetry between losses and gains relative to that point, are developed in the part on behavioural decision theory. Used in section 2 for the direction only. No coefficient is quoted, because the two-to-one figure that circulates in business writing is contested and context-dependent.
- Sanjit S. Dhami, The Foundations of Behavioral Economic Analysis, Oxford University Press. Base rate neglect and the tendency to substitute case-specific reasoning for the underlying frequency sit in the part on bounded rationality. Used in section 3 for the direction of the error. The recommendation to source the base rate locally rather than from published surveys is a judgement about data quality in East African markets, not a claim from this source.
- Jeffrey Carpenter and Andrea Robbett, Game Theory and Behavior, MIT Press. Signalling, and the result that an action carries information in proportion to what it would cost a different type to imitate, underpins the treatment in section 2 of how a conspicuously low salary is read. Note the implication the section draws: because a low salary is cheap to adopt for a founder with outside assets and expensive for one without, it is a poor signal of commitment and a reasonable signal of personal circumstances, which is the opposite of how it is usually intended.
A note on a number this article does not give. There is no correct founder salary and no multiple of median that transfers. The band depends on your market, your obligations and your funding position, and the only reliable source is three founders one stage ahead of you. The method is the transferable part.
Joshua Agonya Pi’Rwot, Founder.