When an investor splits your round into milestone tranches, they are buying an option on it. The first cheque funds today. The second funds only if you hit an agreed trigger, which hands them the right, without the obligation, to continue.1 So accept tranching only when you have named that trigger yourself, and charged for the option instead of writing it for free.
The decision has two parts. Accept the tranche or refuse it. And if you accept, choose the single observable measure that releases the money. Get the second part right and the first stops mattering. Get it wrong and a strong headline number becomes a slow, conditional maybe.
This sits at the sharp edge of the idea behind everything we publish on capital. African venture prices the cost of checking you, not the risk you carry. A tranche is that cost turned into a schedule: the investor pays to verify you in stages, and keeps the option to stop paying if a stage disappoints. Your work is to move that option onto ground you control.
Why these three lenses
A tranche sits where three separate mechanics cross, and no single model reads all three.
The first is that the deferred cheque is a financial option, and options carry a price that grows with your uncertainty and with how long the investor gets to wait. The second is that the trigger only works when it is observable, because a claim you cannot prove moves no money. The third is that the terms of the tranche quietly quote a probability: how likely the investor thinks you are to hit the milestone by the deadline.
Run all three and you get a decision, not a description. One tells you what you are giving away. One tells you what to write into the contract. One tells you how the price was set and how to move it. I fold the behavioural layer, founder optimism about clearing the trigger, into the levers below rather than ship it as its own card, because alone it changes no term. The governance layer, the reporting covenants that ride between closings,3 lives inside the second model, because that is where it does its work.
The framework: three machines behind one deferred cheque
1. The option lens: the second tranche is a right you are granting
Staged financing is old and well understood. Venture capitalists fund in stages precisely so they can gather information and keep the option to discontinue a company that is not working.6 Recent work models the whole relationship as a portfolio of compound options, in which investors hold the right to continue financing or to abandon a startup at each stage.7
An option has value, and the value is not small. It rises with two things you influence more than you think. It rises with the variance of your outcome: the wilder the range of where you might land, the more the right to wait is worth. And it rises with tenor: the longer the window before the trigger resolves, the more the investor learns for free before committing.
So a tranche with an eighteen-month milestone is a costlier option than the same tranche with a six-month one, and you are the party granting it. The canonical practitioner text is blunt about who ends up holding what: when the trigger sits under the investor’s control, the second tranche behaves like a put option they hold; when it sits under yours, it behaves like a call option you hold.2
An option you write for free is still an option. Price it, or the investor keeps the difference.
Assumes the investor can walk or wait at low cost, and your milestone is genuinely uncertain.
Fits because a milestone tranche is a compound option: the second cheque funds only on the trigger, so the investor holds a right rather than a duty.
Breaks when the contract forces the investor to fund on the trigger with no discretion, which turns the option back into an obligation.
Counteracts the founder belief that committed capital is committed.
May reinforce investor caution if you concede the option at no charge.
2. The proof lens: only observable triggers move money
Here is why you cannot talk your way to a smaller tranche. Your forecast is cheap to state and hard to verify. In the language of the Crawford and Sobel cheap-talk model, a costless and unprovable claim carries no separating information: the investor cannot tell your honest projection from the projection of a founder who will miss. The money ignores your words and attaches to something a stranger can check.
The evidence is direct. In a detailed study of real venture contracts, future financing is frequently made contingent on observable measures of financial and non-financial performance.5 Observable is the operative word. Not promised, not projected, not pitched. Measured.
Development finance runs on the same logic. A development finance institution pays out funds in stages, on the condition that certain agreed steps are completed, written into the legal agreement.4 The milestone is simply the form your claim has to take before this money will recognise it.
Which hands you the lever. If the trigger has to be observable, then the choice of what to observe is yours to shape. Name the milestone yourself, and you turn their put into your call. A trigger you largely control, a signed anchor contract, a licence granted, a collections figure cleared, is a call you hold: you decide when to draw. A trigger set on a market outcome you cannot move is a put they hold: they decide whether to fund.
Assumes your forecast is costless to state and expensive to verify.
Fits because an unprovable claim cannot separate you from a founder who will miss, so capital attaches only to observable, contractible measures.
Breaks when the milestone is audited cheaply and objectively, closing the gap between claim and proof.
Counteracts the belief that a sharper pitch will shrink the tranche.
May reinforce over-instrumentation if you accept a measure you cannot move.
3. The odds lens: the tranche terms quote a probability
A price reveals an expectation. That is the core result of the prediction-markets literature: a market price reveals the market’s expectation of the underlying quantity, and simple markets aggregate scattered information into forecasts that are usually accurate.8
Read your tranche the same way. The size of the second cheque, the discount, the way the price often steps up for the later tranche,2 all of it encodes one number the investor will not say aloud: the odds they place on you hitting the milestone by the release date. The tranche is a bet, and the terms are the odds.
The catch is the thinness of the market. Prediction-market prices are reliable when many independent participants trade. Your tranche is priced by one risk-averse investor. A one-trader market returns a wide, defensive quote, tilted to protect the only person setting it. So treat the low implied probability inside your tranche as the quote of a single cautious counterparty with nobody bidding against them. It measures the thinness of the market as much as the state of your company.
That points at two moves. Widen the market, because a competing term sheet forces the implied probability to be set by competition instead of caution. And make the trigger cleanly resolvable, because a market needs an unambiguous settlement rule. One number, one date, one source of truth. A fuzzy milestone hands the investor a second option: the option to dispute the payout at the exact moment you need the cash.
Assumes the tranche terms encode the investor’s probability estimate on your milestone.
Fits because a price reveals an expectation, and a staged deal prices the odds you clear the trigger by the deadline.
Breaks when the market is one cautious investor, so the quote is wide and defensive rather than a fair aggregate.
Counteracts reading the tranche discount as an objective score of your business.
May reinforce a low implied probability if you bring no competing bid.
GEER: the moves, cheapest and most reversible first
Name the trigger before the investor drafts it. Bring one observable, self-executing measure you largely control to the table in your own markup. The party who writes the milestone sets the strike. Let it be you.
Make it cleanly resolvable. Reduce the trigger to a single number, a single date, and a single agreed source. Ambiguity is the investor’s option, not yours. Remove it.
Cap the tenor. A shorter window is a cheaper option to grant. If you can clear the trigger in six months, refuse a twelve-month clock that hands the investor six free months of waiting.
Price the option or convert it. If the trigger stays under their control, it is a put you wrote them: charge for it in the first-tranche pre-money, or attach upside to yourself for clearing it early. Better, convert it into a call you hold by making the draw your decision at a pre-agreed price.
Bring a second bidder. Even one competing sheet re-prices the implied probability. This is the highest-leverage move on the list, and the least used.
RADAR: what to settle before you sign
Do now. Define your one trigger and write it into your own term-sheet markup before the investor’s draft lands. Reversible, costs a morning, dominates every scenario. By T+3 days you should hold a single sentence naming the measure, the number, the date, and who verifies it.
Hedge. Buy cheap insurance against a near-miss. Negotiate a grace period, and a clause letting the investor proceed at the original price if you land close.1 Add a partial-release fallback so a small miss releases some of the tranche rather than none. Table these by T+14, before terms harden.
Defer and trigger. The irreversible move is signing a tranche whose trigger you do not control. Pre-commit now to the rule you will hold at T+28, when the papers are final: sign only if the trigger is a call you hold, or it carries repriced upside to you for clearing it. If neither holds, raise a smaller round with no conditions attached. Fix that line before a live deadline can move it.
CHAIN: what usually happens after you sign one
Match this to the right precedent. Choose the reference class with care: staged financings where the release trigger sat with the investor, a much narrower set than rounds that closed in general. Inside that class the recurring outcome is written into the mechanics. Because the investor holds a right and not an obligation,2, 6 a slipped or contested milestone frequently means the second cheque never arrives, and the company is left short of capital it had already planned around.
Set the base rate there: assume a meaningful share of investor-controlled second tranches do not release on the original terms, because the option was designed to let them lapse.
Now adjust for where you stand. Two present-state modifiers push against you in this market. Development finance, a large share of African capital, is staged on conditions by construction,4 so tranching is the default here, not the exception. And the investor pool is thin, which means fewer competing bids and a wider, more defensive probability on your milestone. Both widen the discount.
Subtract the counterfactual before you credit the structure. An un-tranched round of the same size would have handed you the full amount and no option at all. The tranche added no capital. It added a condition and parked the optionality on the investor’s side. What you surrendered to win the headline number is the thing to price.
One matrix-break flag. The National Venture Capital Association standardised tranched-financing language in its October 2025 model documents.3, 1 As the template hardens into the default, the negotiation shifts from whether to tranche toward which milestone goes in the annex. When it does, the founder who has already named a clean, controllable trigger negotiates from a draft. Everyone else negotiates from the investor’s.
What these three lenses cannot price
The ensemble prices the option, tells you to make the trigger observable, and reads the odds behind the terms. It stays blind to the relationship. It cannot tell you whether this investor exercises discretion in good faith or punitively, whether a missed milestone opens a supportive conversation or a quiet exit. It cannot see the correlation between your trigger and a shock you do not control: a currency move, a regulator, a delayed government payment that stalls the very metric your money depends on. And it cannot price trust, which is what actually decides how a near-miss goes.
So do not pretend the arithmetic settles it. Do this instead. This week, find the one trigger you genuinely control and write it into your markup, with a number, a date, and a named source of verification. If you search honestly and cannot find a single controllable, cleanly resolvable measure, that is your answer: the round is not tranche-ready, and the move is to raise a smaller amount with no conditions rather than write an investor a free option on a milestone whose outcome luck will decide more than you will.
Tranching is not a discount on your company. It is a price for the uncertainty in your forecast. Name the milestone, price the option, and the price becomes one you set.
Sources and notes
- New York Venture Hub, “Milestone Money: NVCA Standardizes Tranched Financings” (2025). Defines a tranched financing as installments where later tranches close only if the company hits agreed milestones, and describes founder-friendly softeners such as a grace period and a “proceed at original price if close enough” option. Verified: 200, article body contains the quoted definitions.
- Venture Deals, “Tranched Financings”. States that the trigger releasing the second tranche can be under the investor’s control (“similar to a put option”) or the company’s control (“similar to a call option”), and that the price can increase for the later tranche. Verified: 200, body contains the put and call language.
- Foley & Lardner, “Breaking Down the October 2, 2025 NVCA Updates to the Model Legal Documents” (2025). Notes that milestone-based or “tranched” financings are now formally addressed in the model Stock Purchase Agreement, and that staged funding tends to bring tighter operational covenants between closings. Verified: 200, body contains the SPA and covenant language.
- International Finance Corporation, “IFC Project Cycle”. Disbursement stage: “Funds are often paid out in stages or on condition of certain steps being completed as agreed in the legal agreement.” Verified: 200, body contains the quoted sentence.
- Kaplan, S. and Strömberg, P., “Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts,” NBER Working Paper 7660 (2000). Full text: PDF. Documents that future financings are frequently contingent on observable measures of financial and non-financial performance. Verified: 200, PDF text contains the “observable measures” finding.
- Gompers, P., “Optimal Investment, Monitoring, and the Staging of Venture Capital,” Journal of Finance 50 (1995), 1461-1489. Finds that venture capitalists stage investments to gather information and maintain the option to discontinue funding projects with little probability of going public, on a sample of 794 venture-capital-backed firms. Verified: 200, abstract body contains the staging and discontinuation-option language.
- Hillenbrand, S. and Stafford, E., “Venture Capital as Portfolios of Compound Options,” Harvard Business School (2025). Models venture financing as compound options in which investors hold the option to continue financing or to abandon a startup at each stage. Verified: 200, PDF text contains the compound-option and continue-or-abandon language.
- Wolfers, J. and Zitzewitz, E., “Prediction Markets,” Journal of Economic Perspectives 18(2) (2004), 107-126. Shows that a market price reveals the market’s expectation of the underlying quantity, and that simple markets aggregate dispersed information into forecasts that are typically accurate. Verified: 200, PDF text contains the price-reveals-expectation result.
Model note: the cheap-talk framing follows Crawford, V. and Sobel, J., “Strategic Information Transmission,” Econometrica 50 (1982). The only openly hosted copy is an image scan with no text layer, so it is named here for provenance and not cited for any figure; the substantive empirical claim rests on Kaplan and Strömberg (note 5).