Send a dated note to your investors every month and you are building a record a future investor cannot manufacture. It costs an hour. It compounds into the one thing diligence is actually paying for: a reason to trust your numbers without re-checking every one. The founders who start the month they need the money have already skipped the part that mattered.
Most founders treat the update as a chore for the months that went well. That is backwards. Almost all of the value sits in the months you did not want to send it.
This is not the memo you write to open a round. That one travels to a room you are not in and is judged once. The update is the opposite instrument. It never opens a round, it is judged over two years, and its whole power comes from having existed before you needed it.
Start this month. Fix four numbers. Send them whether the month was good or not.
Why these three models
This piece runs the Wire Model: score the decision, route it to a small ensemble of formal models, force dated actions out of them. Three features carried the routing.
- Strategic disclosure under asymmetric information (0.85). You know which months flatter you. The reader knows you know. Whether your silence gets priced is the whole game.
- Nonlinear accumulation (0.75). Credibility does not add up in a straight line. It sits inert, then it tips.
- Belief formation under noise (0.7). An investor holds a moving estimate of you and updates it on fragments that arrive out of order.
That routes to three models: verifiable disclosure and unraveling (whether your silence reads as bad news), threshold (why the record tips instead of climbs), and prediction markets (how a stranger prices you in the gaps between updates). They span equilibrium, complex and random, so their errors point in different directions.
Two more models earned a seat and did not get a card. Behavioral loss aversion explains why founders skip the bad month, but it changes how you read every lever below without adding one of its own, so it rides inside the cards instead. Repeated-game reputation is the connective logic across all three, not a fourth lens: the compounding you are buying is a reputation built through repeated, observed play, where a small doubt about your type is resolved by a succession of kept promises.6 I folded it in rather than spend 350 words on a card the argument does not need.
The framework: a record is worth more than the month it records
1. Disclosure: your silence has a price, and the record sets it
Start with the theory of what silence means. Milgrom’s unraveling result says a seller who can prove any true claim ends up revealing everything, because the buyer treats each non-disclosure as the worst case, and the best of the silent types keeps peeling off to separate itself, until only the genuinely bad stays quiet.2 In the clean model, no news is bad news.
Reality is softer, and the gap is your opening. When Jin, Luca and Martin ran the disclosure game with real people, senders disclosed favorable draws about 95% of the time and quietly withheld the unfavorable ones, and they got away with it, because receivers were insufficiently skeptical about what was not shown.3 Withholding paid.
In a one-off raise, that is exactly what happens to you, in reverse. You can hide a weak quarter and the investor cannot fully punish the silence, because there is no series to hold it against. A monthly cadence removes the hiding place on purpose. Once you have sent eighteen dated notes, a missing month is loud. You have walked yourself into the equilibrium where your good news is believed precisely because your bad news was already on the record.
So the separating move is the bad-month send. A founder who reports only wins is indistinguishable from one who is burying a loss. A founder who wrote “we missed, here is why, here is the fix” in March and hit the fixed number by July has done the one thing a liar will not copy, because copying it means surfacing a real loss on the record. The update is not a report on the month. It is the asset the month produced.
For an African founder the raw material already sits in your systems. Mobile-money settlement logs, agent-network daily actives, signed LPOs, a WhatsApp-order export: dated, hard to fake, and yours. The update is where they turn into a series a stranger can trust.
Verifiable disclosure and unraveling
Assumes: someone eventually notices the pattern of what you send and what you skip.
Fits because: strategic disclosure scored 0.85; a cadence turns each silence into information a reader can price.
Breaks when: nobody is keeping count. A first-time reader with no history cannot read your silence, so the effect only switches on once the series exists.
Counteracts: the reflex to report wins and go quiet on losses.
May reinforce: cosmetic honesty, framing every loss as a lesson while fixing none of them.
2. Threshold: credibility tips, it does not climb
Granovetter’s threshold model says a person acts once the count of others who have already acted crosses their own trigger, and that a population of different triggers produces wildly different outcomes from nearly identical averages.4 Point it at belief rather than behavior. Each investor carries a private threshold: the number of consistent months before they lean in, write a check, or make the introduction. You cannot see whose threshold you are about to cross.
That is why the record cannot be built at the moment of need. You do not climb into an investor’s confidence a percentage point at a time. You sit below the line as one more unverified founder until a count of kept promises tips you over it, and then, suddenly, you are the founder who has hit the number they said they would for six straight quarters. The tip is discontinuous. The months that produce it are banked long before it happens.
Which reprices the boring month. The earliest, cheapest, most uneventful update is the highest-return one, because it sits furthest below every reader’s threshold and there is no way to buy it back later. Start the count now. The founder starting the same count next year is a full year of proof behind you for the rest of the company’s life.
Visible’s platform data puts a number on the tip: founders who send consistent updates are twice as likely to raise follow-on funding.1 Read that as a threshold effect, not a linear one. The updates did not each add a slice of probability. They accumulated until enough readers crossed their trigger at once.
Threshold (Granovetter)
Assumes: each reader has a private trigger, a count of consistent months before they act.
Fits because: credibility accumulates and then tips; nonlinear accumulation scored 0.75.
Breaks when: the underlying number is trending down. A steady record of decline crosses no one’s threshold, and should not; the model rewards consistency, never direction.
Counteracts: the belief that you can climb into confidence gradually, on demand, when the round opens.
May reinforce: cadence theater, mistaking the act of sending for the substance being sent.
3. Prediction markets: a stranger prices you between the updates
Picture every investor watching you as holding a position in a thin market on one question: will this founder be a good bet at the next round. Wolfers and Zitzewitz show that a market price aggregates dispersed information into a probability, and that the estimate is only as reliable as the market is liquid.5 Your updates are the trades. Each dated note moves your price and, more usefully, narrows the spread: the uncertainty a reader carries about you between one note and the next.
Low variance is the asset. An investor can move fast and pay up on a founder whose next number they can predict inside a tight band. They discount a founder whose outcome is a wide guess. The width is what they are pricing, and it is what your silence widens.
Now picture the founder who goes quiet for eleven months and arrives with a polished deck the week the round opens. That is one large trade placed into an illiquid market, at the exact moment their interest is most obvious. The price barely moves and the timing is its own tell: why now, because you are raising. A stranger cannot rebuild a record you assembled the week you needed it. That is the same reason a top-down market size lifted from a consultancy PDF gets discounted toward zero.
Feed the market steadily instead. Small, regular, dated trades keep you liquid and keep your variance low, and variance is the thing that earns you a discount you can compute.
Prediction markets
Assumes: readers hold a moving estimate of you and update it on the fragments you send.
Fits because: belief formation under noise scored 0.7; dated updates aggregate into a price and shrink its spread.
Breaks when: the market is one reader deep. A single decision-maker backing you in the room is not aggregating anything, so the thin-market logic does not apply.
Counteracts: saving your proof for one big reveal.
May reinforce: over-managing the narrative instead of the company underneath it.
The levers, cheapest and most reversible first
Every lever here costs less than the discount it removes. Run them in order.
- Send this month, flat or not. Four numbers, three sentences of context, one specific ask. An hour, fully reversible, and it starts the count every later lever leans on.
- Fix the four numbers in writing now, and never swap them. Metrics chosen after you have seen which ones flatter you carry no information. Pick four you will report for twenty-four months before you know how they will land. The full case for pre-committing a metric set is its own decision; here it is enough that the four never change.
- Send in the bad month, on time. This is the separating move and the only expensive lever, because it means writing down a real loss. It is reversible in cost and irreversible in value: the bad-month note is the line an investor rereads when deciding whether to trust the good ones.
- Publish the cadence. Tell your list the update lands monthly, on a fixed day. This is the least reversible lever, because it prices your future silence. Once you have promised the cadence, a skipped month is a signal you cannot take back. Do it only when you intend to keep it.
What to do before, during and after the next round
Do now (T+0 to T+3). Send the update this week and fix the four metrics. This dominates every scenario: if you raise soon it is your warm list, if you do not it is your record, and it costs the same either way.
Hedge (T+7 to T+14). Put a recurring block on the calendar and build a one-screen template, so no single bad month can break the chain. Add the investors who passed but asked to stay in touch. A pass-then-watch investor is the cheapest lead you will ever have, and the update is the only thing keeping that option alive.
Defer and trigger (T+28 and beyond). Do not open the raise on the strength of the record until you have banked roughly six dated months. Pre-commit the trigger now: the month your headline number crosses the level you named in your first update is the month you send the “we are raising” note. Naming the trigger in advance is what stops you opening late, when the record still reads as assembled.
What usually happens next
Match the class on structure, not on surface. The wrong reference class is “startups that email their investors.” The right one is founders whose numbers a stranger can verify from a dated series, set against founders who assemble the same proof at the moment of need. Same documents, opposite credibility, because one is timestamped across two years and the other is timestamped last Tuesday.
The base rate for the disciplined class is favorable: twice the follow-on likelihood in Visible’s data.1 Now modify it for today. African funding fell to 2.2 billion dollars in 2024, down 25% on the year, as developed-market investors pulled back, and debt funding alone dropped about 40%.8 Longer, thinner, more skeptical diligence raises the price of checking, which is exactly the price a verified record removes. The instrument is worth more in this market than it was in 2021.
Subtract the counterfactual before you credit the update with the round. The record does not raise the money; the business does. Strip out the founders who would have raised regardless, and what the update actually moves is the marginal deal in a slow market: the investor on the fence, the partner who has to convince a committee, the local co-investor being asked to vouch. That is where twice-as-likely lives.7
The matrix-break flag: a model can now write your update for you. When the cost that made the signal separate, your own time and your honesty in a bad month, collapses into a prompt, investors will re-price the whole instrument, and the separating layer moves to the artifacts a model cannot fake on your behalf. Settlement logs, signed contracts, cohort exports pulled straight off the rail. Keep those close, because that is where this signal is heading.
The reading these three models cannot give you
None of this sees whether your business is any good. A flawless record of a dying company documents the death in high resolution. The update is a credibility instrument, not a survival one, and a founder can win the cadence while losing the market. It also cannot see the reader who will never move: the investor who does not touch your sector, whom no run of kept promises will convert.
So keep the claim narrow. A record cannot save a business the numbers already condemn, and it cannot invent a quality you do not have. Its job is smaller and surer. It makes a real business legible to a stranger before you know which stranger, and in a market that reprices the whole continent on one founder’s fraud, that legibility is the cheapest edge on the table.
You cannot predict the reader whose threshold you are about to cross. You can only make sure the record is already there when they look. Open a blank note. Put this month’s four numbers in it. Send it before you know who needs to see them.
Sources and notes
- Visible, “How to Write the Perfect Investor Update,” visible.vc. States that companies which regularly communicate with their investors are “twice as likely to raise follow-up funding.” visible.vc/blog/how-to-write-the-perfect-investor-update
- Paul Milgrom, “Good News and Bad News: Representation Theorems and Applications,” Bell Journal of Economics 12 (1981): 380-391. The origin of the unraveling result. Cited qualitatively: the Stanford author copy is a scanned image with no machine-readable text layer, so it is not cited for a figure. milgrom.people.stanford.edu
- Ginger Zhe Jin, Michael Luca and Daniel Martin, “Is No News (Perceived As) Bad News? An Experimental Investigation of Information Disclosure,” HBS Working Paper 15-078 (2015, rev. 2017). Senders disclosed favorable draws roughly 95% of the time and withheld unfavorable ones; receivers were “insufficiently skeptical about non-disclosed information,” so “senders increase expected returns by strategically withholding unfavorable information.” hbs.edu
- Mark Granovetter, “Threshold Models of Collective Behavior,” American Journal of Sociology 83 (1978): 1420-1443. A threshold is “the proportion of the other members of the group who must make one choice before a given individual does so,” and heterogeneous thresholds yield very different aggregate outcomes. cse.cuhk.edu.hk
- Justin Wolfers and Eric Zitzewitz, “Prediction Markets,” NBER Working Paper 10504 (2004), Journal of Economic Perspectives 18(2): 107-126. Simple markets “aggregate disperse information into” forecasts and reveal “the market’s expectations about probabilities,” with reliability rising in liquidity. nber.org/papers/w10504
- David M. Kreps and Robert Wilson, “Reputation and Imperfect Information,” Journal of Economic Theory 27 (1982): 253-279. A small amount of incomplete information about a player’s payoffs, in a game played against a succession of observers who see earlier moves, “is sufficient to give rise to the reputation effect that one intuitively expects.” ideas.repec.org
- Visible, “Investor Reporting for Startups,” visible.vc. Early-stage founders “send updates monthly,” shifting to quarterly as they grow, and “consistent communication often unlocks introductions, strategic advice, and a more supportive network.” visible.vc/blog/investor-reporting
- “African startups raised $2.2B in 2024, 25% less than 2023,” Connecting Africa, reporting Africa: The Big Deal data. Funding fell to 2.2 billion dollars, down 25% on 2023, as developed-market investors pulled back; debt funding dropped about 40% year on year. connectingafrica.com