The decision is which documents to sign this month, before an investor asks to see them. Sign the five that turn your cap table into a record a stranger can check: the founders’ IP assignments, the team’s invention assignments, founder reverse-vesting, a board-authorized option pool, and a share register that matches what you have filed. Each is cheap today and costly to explain later, because at diligence an undocumented item reads as the worst thing it could be.
A cap table is a claim about what a buyer of your shares actually receives, read line by line. The lines you never wrote down are the ones a diligence lawyer prices against you.
Silence on a document is not the absence of information. To the investor’s associate, it is information, and it points one way.
This piece is not the folder structure. Every law firm publishes one. It is about why a missing signature converts good news into a discount, and which missing signature to sign first.
Why these three models
Three mechanisms decide what an undocumented cap-table item costs you, and they fail in different places, which is why they run together rather than alone.
The first is an information model: the investor cannot cheaply tell a clean cap table from a messy one, so an undocumented line gets pooled with the worst case and priced accordingly. The second is a communication model: it explains why saying the line is fine adds nothing, and why only the signed instrument moves the price. The third is a structural model: it explains why the five gaps are not equal, so you can rank them and fix the one that contaminates the most first. One says the gap is expensive. One says talk will not close it. One says which gap to close first. Equilibrium and complex, two outcome types, so a weakness in the information story does not sink the structural one.
The model-thinking discipline would append a behavioral layer and a governance layer. We folded both in. The behavioral layer is the founder’s honest belief that everyone remembers what was agreed, and it explains the skipping without changing the price, so it sits in the blind spot. The governance layer is the board resolution that authorizes the option pool, and it is not a separate lens: it is one of the five documents, so it lives inside the structural read rather than in its own card.
The framework: three reads on one undocumented line
1. The pool: why the missing document is priced as the worst case
George Akerlof’s 1970 result is the spine. When a buyer cannot verify quality before purchase, the uninformed price drives the high-quality sellers out, average quality falls, the buyer revises down again, and the market can collapse toward the bottom.1 The condition that produces the collapse is a deficiency of public quality assurance: reputation, regulation, or a checkable guarantee.9
Your cap table is a used-car lot to the investor. They cannot see, from the outside, whether your IP is actually assigned to the company, whether the pool was ever authorized, whether the 2% you promised a developer over WhatsApp is on the register. So they price the average of the pool they think you belong to, and that pool contains every founder who did not own their IP and every unissued option that surfaced mid-deal. The price of the pool shows up off the term sheet: deeper diligence, an indemnity holdback, a lower cap, or the weeks the deal sits open.4
Certification is the cure Akerlof named. A signed IP assignment, a board resolution, a filed return: each is a public quality assurance that separates you from the pool. It is worth signing precisely because it is checkable by someone who does not trust you yet.
Adverse selection and certification.
- Assumes the investor cannot verify each cap-table claim cheaply and prices the gap as the pool’s average.
- Fits because an undocumented line is hidden quality, and hidden quality gets discounted to the worst case.
- Breaks when the investor already trusts you and treats the gaps as formalities, so no pooling penalty applies.
- Counteracts the belief that a gap is neutral until raised.
- May reinforce box-ticking, if founders chase documents instead of building the business under them.
2. The talk: why saying it is fine changes nothing
“We agreed she gets two percent.” “The IP is ours, obviously.” “The pool is basically ten percent.” These are cheap talk in the exact technical sense: costless to say, non-binding, and unverifiable by a third party such as a court.2 Crawford and Sobel’s result is that a costless message carries information only when the sender’s interests are aligned with the receiver’s. Yours are not. You want the round to close, so you have every incentive to say the cap table is clean whether it is or not, which means the statement carries no information at all and gets discounted to zero.
Worse than zero. The verbal promise you disclose is now a claim the associate has to chase, and the promise you forgot is the one that surfaces in a reference call and stalls the wire. A signed instrument is the opposite of cheap talk on every axis: it costs something to execute, it binds you, and a third party can verify it.
The signature is the part a liar will not pay for.
This is why repeating the assurance does nothing and executing the document does everything. The founder who says “trust me, it is clean” and the founder who is hiding an unissued pool make the identical sound. Only one of them files the return.
Cheap talk.
- Assumes the founder’s interest is to have the cap table appear clean regardless of whether it is.
- Fits because a costless, non-binding claim from a misaligned sender transmits nothing and is discounted to zero.
- Breaks when the claim is independently checkable at low cost, at which point it stops being cheap talk and starts carrying weight.
- Counteracts the instinct to reassure investors verbally instead of signing.
- May reinforce silence, if a founder decides that saying nothing beats saying something unprovable.
3. The web: which gap to close first
The five gaps are not equal, and network centrality says why. A centrality measure ranks the nodes of a graph by their position, to identify the ones whose removal or failure matters most: the most influential person in a network, the key node in an infrastructure map.3 A cap table is that kind of graph. Every claim on it depends on other claims being sound.
Unassigned founder IP is the most central node. If the company does not clearly own its core technology, every downstream claim is contaminated: the investor’s preferred shares, the option pool, an acquirer’s rights three years out. One unsigned assignment expands diligence across the whole structure, because the associate now has to check whether you own the thing you are selling access to.5 A single advisor’s undocumented quarter-percent is a leaf: it should be fixed, but its ambiguity infects nothing else.
So centrality gives you the order. Sign the founders’ and team’s IP assignments first, because they sit under everything. Put the founders on vesting next, because an unvested founder who leaves takes a live claim with them. Authorize the pool third. Reconcile the advisor and angel promises last. Fix the node that everything hangs on, then work outward to the leaves.
Network centrality.
- Assumes cap-table claims depend on one another, so an ambiguity’s damage scales with how many claims sit downstream of it.
- Fits because unassigned IP and unvested founders are load-bearing nodes, while a small undocumented grant is peripheral.
- Breaks when the gaps are genuinely independent, so fixing order does not matter and any sequence works.
- Counteracts the habit of fixing the easy document first because it is easy.
- May reinforce neglect of the leaves, which still surface in a reference call.
GEER: the moves, starting with the one that costs nothing
Order them by cost and by how easily you can reverse them.
Write down every promise you have made. One dated list: name, amount, instrument, whether it is signed. The verbal grant to the developer, the handshake with the angel, the advisor’s quarter-percent. This costs an hour, it is fully reversible, and it converts a set of things you carry in your head into a set of things you can act on.
Sign the IP assignments. Founders assign inventions to the company. Every employee and every contractor who touched the product signs the same. Investors check whether the startup owns the intellectual property behind its products, and the assignment is the answer.6 Cheap template, and the one document that sits under all the others.
Put the founders on reverse-vesting. A schedule with a cliff, so a founder who walks in year one does not keep a slab of the company. This is a document, not a valuation, so it is cheap to draft and easy to agree before anyone has a reason to fight about it.
Authorize the pool with a board resolution. An option pool that lives only on a pitch-deck slide is not issued. The board resolution that reserves it is the act that makes it real, and it is a governance step, not a legal bill.
File the record where a stranger can read it. A reconciled share register, allotments and returns filed with the registrar, is the most external and least fakeable proof you own. In Nigeria, for instance, annual returns must be filed at the Companies’ Registry each year, which turns your ownership into a record the investor can pull rather than take on your word.7
RADAR: what to sign long before the round opens
Do now, T+0 to T+3. Write the promises list and sign the founders’ and team’s IP assignments. This dominates every scenario. It is nearly free, it is reversible in effect if a relationship changes, and it removes the most central ambiguity on the table. If you do one thing from this piece, do this.
Hedge, by T+14. Put the founders on reverse-vesting and pass the board resolution that authorizes the pool. This is cheap insurance against two specific tail events: a co-founder split that strands equity, and a mid-diligence finding that your pool was never issued. Both are survivable if the documents predate them and expensive if they do not.
Defer and trigger, T+28 and beyond. The lawyer-reconciled register, the filed returns, and a defensible valuation for option strikes cost real money and are not free to unwind, so do not buy them on a calendar. Pre-commit the trigger instead: the first serious investor conversation, or the first option grant you actually issue to a hire. When the trigger fires, buy the reconciliation. Not before, and not after the term sheet, when it becomes a condition to closing rather than a fact on file.
CHAIN: what usually happens to an undocumented claim at diligence
Match on structure, not on the word “startup,” and the reference class is anyone asked to prove clean title to an asset they never documented as they acquired it. The property seller with an unregistered transfer. The borrower pledging collateral under a defective lien. The applicant whose claimed credential cannot be verified by the issuer. Across that class the deal rarely dies outright. It slows, it reprices, or a warranty gets wrapped around the gap so the buyer can proceed and still recover if the gap turns out to matter.
The base rate sits against the founder with gaps, because cap-table, IP and vesting problems are among the most routine findings in venture diligence, and the clean-up is a known, costly exercise rather than a novel one.4 Two present-state modifiers push it further. African diligence runs a legal review covering company registration, corporate compliance and IP protection, so the gaps get looked at directly rather than waved through.8 And the standard document set is public: the model financing documents investors work from are industry-embraced and specific, so a missing piece is conspicuous against a known template rather than hidden in bespoke paperwork.10
Now subtract the counterfactual. The round that closed on a messy cap table closed because the investor priced the fix in, as a holdback, a condition, or a lower number, and absorbed it. Do not read that outcome as proof the documents did not matter. The founder paid for them anyway, in worse terms, and got the paperwork as homework after signing.
Matrix-break flag. If a registry becomes a real-time, publicly verifiable source of truth for ownership, or if cap tables settle on a shared ledger a stranger can query, the silence this whole piece prices simply disappears. Hygiene stops being a signal and becomes the default, and the founder who kept a clean record gains nothing over the one who did not. Watch for it. Until it arrives, the gap is unverifiable from the outside, and the unverifiable gap is priced as the worst case.
Where a clean cap table tells the investor nothing
The ensemble reads one thing: whether the record exists and is signed. It is blind to whether the deals the record memorializes were good ones. A perfectly executed fifty-fifty split with no re-split trigger is a clean record of a dispute waiting to happen, and the signatures prove that you agreed while saying nothing about whether you agreed well. How you decide the splits is a different question, and a separate one. This ensemble does not touch it.
It is also blind to trust, which can make the whole exercise moot, as the counterfactual showed. And it cannot see the real reason founders skip these five documents, which is a conviction, honestly held, that everyone remembers what was agreed and the paperwork can wait. That is the behavioral layer, and the cheap-talk model already priced it: memory is costless, non-binding and unverifiable, which is exactly why it will not close your round.
Here is the action that survives all of that. You cannot document the past during diligence. You can only sign the present, so sign it now. Pick the most central undocumented claim on your cap table, the IP assignment under everything else, and execute it this week. Sign it this month, while it costs a signature. Wait for the term sheet, and the same signature costs a discount.
Sources and notes
- George A. Akerlof, “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism,” Quarterly Journal of Economics 84, no. 3 (1970): 488 to 500, as summarized here: when buyers cannot verify quality, the uninformed price “drives the high-quality cars from the market,” average quality falls, and adverse selection can lead to market collapse. en.wikipedia.org
- “Cheap talk,” on Crawford and Sobel’s framework: cheap talk is communication that is “costless to transmit and receive,” “non-binding,” and “unverifiable (i.e. cannot be verified by a third party like a court),” and it transmits information only when sender and receiver interests align. en.wikipedia.org
- “Centrality,” on ranking nodes by network position: centrality indicators “assign numbers or rankings to nodes within a graph corresponding to their network position,” to identify “the most influential person(s) in a social network” or “key infrastructure nodes.” en.wikipedia.org
- Cooley GO, “What Is a Cap Table and Why Do You Need One?”: “A poorly managed cap table can lead to bad decisions and diligence issues in transactions, as well as costly and time-consuming ‘clean-up’ exercises.” cooleygo.com
- Cooley GO, “What Is a Cap Table and Why Do You Need One?”: the more out-of-platform information a founder or lawyer must track, “the more susceptible you are to errors stemming from reliance on an incomplete cap table,” including complex vesting schedules with milestone and time-based components that must be tracked. cooleygo.com
- Kruze Consulting, “Startup Due Diligence”: among the questions VCs ask, “Does the startup own the intellectual property (IP) linked to its products and services?” kruzeconsulting.com
- ICLG, “Corporate Governance Laws and Regulations: Nigeria”: “there is a requirement to file annual returns at the Companies’ Registry not later than 30 June each year,” at the Corporate Affairs Commission, the registry in charge of incorporation. iclg.com
- Founders Factory Africa, “A Guide for African Tech Startup Founders”: a raise ends in “a legal review to ensure everything’s up to code,” covering “company registration, tax compliance, corporate compliance, employment laws, and protecting your intellectual property.” foundersfactory.africa
- “The Market for ‘Lemons'”: a lemon market forms given a “deficiency of effective public quality assurances (by reputation or regulation and/or of effective guarantees/warranties),” which is why a checkable assurance separates a seller from the pool. en.wikipedia.org
- National Venture Capital Association, “Model Legal Documents”: “The NVCA Model Legal Documents serve as the industry-embraced model documents to be used in venture capital financings,” and the set includes the Certificate of Incorporation, Stock Purchase Agreement, Investors’ Rights Agreement, Voting Agreement, and Right of First Refusal and Co-Sale Agreement. nvca.org