The Blog

A crowded category is not evidence of demand. Positions cluster because clustering is individually rational, and the returns do not divide evenly.

Three published medians for the same population in the same year, and they differ by a factor of eight. Why that keeps happening, and the four numbers to track instead.

In a contest where everyone is valuing the same underlying thing, the winner is whoever estimated highest. Winning is evidence your estimate was wrong.

Nine months of n8n security advisories, a benchmark collapse nobody mentions in the demo, and the three bills that arrive after the workflow is already running.

Against a stream of one-shot counterparties, the most you can ever earn is what you would get by committing publicly first. Cleverness inside the relationship cannot beat it.

Someone in your company is making a decision and never sees its consequence. Connecting those two is usually a meeting invitation.

You invented the weights. Nothing has ever checked whether your own stated priorities contradict each other, and they usually do.

The consequence you did not plan for is not a property of the world. It is a report on the boundary you drew, and the boundary was yours.

You will never see inside the algorithm, the buyer or the market. You do not need to. Vary one input and read the output.

A gain at step one that terminates below where you started is not a gain. And when the tree comes out symmetric, runway decides.

The first order is not a small version of the deal. It is the instrument you use to find out who you are dealing with.

Correct work does not carry its own recognition. Proof is a property of the work. Credit is the output of an institution, and it arrives late.

Reputation is not a balance you built and hold. It is a stock with an outflow, and the outflow runs whether or not you are watching.

If a different person in the same chair would behave the same way, you do not have a people problem. Replacing them resets nothing.

An open system reaches the same end from different starting points by different routes. Their sequence encodes conditions you do not have.

Present bias over money is modest. Over work it is large. That single split tells you where to put every deadline you set.

Your record of absorbing shocks is evidence about small suppliers. It says nothing about the one you cannot replace.

Count the distinct situations arriving. Count the distinct responses you have. If the second is smaller, hiring changes neither number.

Adjusted for publication bias, the average nudge effect is 0.04 with a confidence interval that includes zero. Change the payoff instead.

The failure in your operations is not a people problem. It has been tested against experience, incentives and competition, and it survived all three.

Whether a counterparty keeps their word is decided by four numbers, and none of them is character.

Take the shiny hire, channel, or customer only if at least one still-acceptable future remains after the grab. One counterexample fails the rule.

Your investor, your buyer and your competitor are reasoning one move ahead. Your clever second-order signal is a cost you are paying alone.

Decide this week which records may be overwritten and which must only be appended. Investors reconstruct from a journal.

Expected value is computed across parallel worlds. Your company lives in one, and it has an absorbing state at zero.

A sponsor-bank arrangement lets you launch without a licence. It also hands the sponsor the right to change the price the day their regulator moves.

A flip is a taxable disposal in one country and a tax-free rollover in another. The founder who learns the difference after signing pays it out of their own pocket.

Connecting all sites cheaply is a tree. Visiting every city and returning is a tour. Name the class, then ship an approximation.

Grant and challenge-fund capital is priced in founder-weeks of reporting, procurement and audit. Cost it before you apply, not after you win.

The founder owns the number on the homepage, the pitch, and the monthly update. The contractor who made it pretty owns nothing, unless the brief put the claim in their…

You will concede something to close the round. Concede the price and you lose a number once. Concede pro rata casually and you write your best investor a call option…

Name the one-sentence point before you pick a graphic for this week's investor update. If there is no point, do not graph.

Your investor is graded on cash returned, not on paper marks, and that one fact explains the pressure you are getting and what to offer instead of what is being…

Write two lists before you spend the next introduction: the loud rolodex, and the person who sits between a group you can reach and a group you cannot.

Undocumented promises, an unissued option pool, unsigned IP and unvested founders are read as bad news the day a term sheet lands. Execute the five documents this month, while each…

Schemas are the API of the relationship. Additive first. A break needs both shapes live, a dated close, and one writer who owns the cut.

A top-down market size lifted from a consultancy PDF is priced near zero by an investor who cannot check it. Build the number a stranger can rebuild, and the gap…

A milestone tranche converts your unverifiable forecast into an option the investor can buy. Name the trigger yourself, and price it, before they draft it for you.

The likeliest cash your cap table ever produces is a partial sale inside a later round, and it is a term you negotiate on the way in while you still…

If every candidate still needs the same three reference calls, remaining spend cannot buy a faster loop. Name the unit the tool deletes, or do not pay.

An objection is not a rejection.

A revenue-share facility forgives a bad month and even a bad year. What it cannot carry is an unpredictable one, so the instrument fits a smooth collection cycle and quietly…

The customer money you hold sits on a partner's books as their asset and your liability, and the morning their problem becomes yours, no assurance unfreezes it.

The cap you celebrate today is a strike price on a round you have not raised. Set it above what you can clear, and the good news reprices into a…

Two readers of one stream will disagree while writes continue. Bound the lag, or serialize the join.

Which affiliate and content loops may run unattended, and which must stop for a human. Variety, reputation, and ruin, applied to a thirteen recipe automation playbook.

The advice you have been given is to build a network.

A second tap or a webhook retry can debit twice. Deduplicate on a stable id. Exactly-once does not travel across HTTP or a bank.

Treat the audit as a purchase of verification, then run the return on it: the discount an investor drops when your numbers stop needing to be trusted is a number…

The realistic African buyer set is small, strategic and slow. It buys licences, distribution and books, not growth. That changes what you build three years before you sell.

A director's guarantee reattaches the downside your incorporation removed, so cap it, carve it to acts you control, and put an institution's balance sheet where your house is being asked…

You have solved the same problem for twenty clients. Each time you scoped it fresh, quoted it fresh, and negotiated it fresh.

A term sheet is not a price. It is a set of instructions for the bad states, and the instructions do the real deciding.

Reincorporating into Delaware or Mauritius buys your investor a legal system they already know how to check. The only live question is when to pay for it.

A data room assembled during a raise dates itself in a suspicious cluster. One kept an hour a month dates itself honestly, and the date is the part a liar…

You have found the number. The work you are proposing is worth about fifty thousand a year to the buyer.

An option pays only when there is a sale to exercise into. In a market of two dozen exits a year, most option grants are a currency the holder can…

You built the thing. It works. The investor said no.

Debt is 41% of the capital African startups raise, and it is growing faster than equity. It is the larger of the two pools most founders can actually reach, and…

When a founder in your category is caught lying, the market raises the checking bill for everyone who looks like them, and mails part of it to you.

A company can be profitable in local currency and still have that profit trapped, because repatriability is a separate property of the cash, with its own queue, its own price,…

A 0m fund and a 00m fund need different exits from the same company. Read the fund and you know what your investor needs before the first meeting.

Founders price an agent network like software that scales for nearly free. It is a labour force with recruitment, training, churn and management costs, and that changes both the unit…

In African B2B and B2G, the biggest contract you sign is often the biggest loan you make. Price the term, or the term prices you.

A dated note you send every month, especially the bad ones, builds a track record a future investor cannot manufacture, and that record is what decides the next round.

In African markets a regulator rewrites your unit economics faster than any competitor can. Reading primary sources on a schedule is the cheapest edge you are not using.

The valuation decides how the story reads. The preference stack decides how the money moves. In a market of small exits, only one of those is your outcome.

Where you incorporate is read by a foreign fund and a local regulator in opposite directions. Pick for the counterparty you cannot afford to lose.

When no top-down TAM survives diligence, the credible market size is the one you build from counts a stranger can check.

The fastest capital is not the cheapest. It is the one whose checking machine already fits the records you hold.

Acquisition cost is a price set in an auction you do not run, and the bidder who moved it never appears on your dashboard.

A fixed price in a moving currency is an option you wrote for your customer. It has a market price, and you charged nothing for it.

A rival's round is a price change in your market, and the only decision that pays is which fronts you concede on purpose.

A raise relieves every constraint that carries a price. Find out whether yours does before you sign.

Every overload panic on record was answered the same two ways. Knowing which two tells you exactly where to spend the next hour.

Miller measured two limits, and neither one is the rule you are quoting. Build for four chunks, then spend everything else on making each chunk bigger.

Three measurable inputs govern how well you decide across a working day, and they imply a different calendar from the one most founders run.

A corner does not raise your intelligence. It cuts the price of dropping the assumption that was capping your options.

It is the only slide that puts an expiry date on the opportunity, and most founders fill it with trends.

An intro is a loan against someone else's reputation, and the terms are set long before you ask for it.

Volume multiplies whatever sign your contribution margin already carries, and somebody else is paying for the sign you have now.

The number you cannot explain is a customer doing a job you never built for, and the story you invent to explain it is how you lose them.

The technique works because certainty changes what the mind is allowed to retrieve, and what the room is allowed to say.

A deck is a visual aid for a narrator, and the room where your round gets decided has no narrator.

A no with a condition attached is a specification someone else wrote for you. A no without one is an exit, and building against it wastes a quarter.

Two numbers set the dial: what it costs to correct the call once it is wrong, and how much error the extra time actually removes.

Every other number on your dashboard is a stock somebody else gets to re-count. Retention is a rate inside a closed cohort, which is why it holds its shape.

The evidence of your business already exists. It sits in WhatsApp threads, till books and mobile money statements, in a format no investor can price.

An investor prices the class before they read the company, and the checkable part before the true part. Run that order on yourself first.

Dated calls, access nobody else can buy, and contact recent enough to still be true. That is the whole of it.

The cheapest fix for a default-dead company takes about three months to run, so a quarterly review finds the problem after the window to solve it has closed.

Your inputs are a portfolio. Price every one of them by the decision it can change, then cancel the rest.

Every problem has a class, and the class picks the method. Get the class wrong and effort stops helping.

A complicated problem has an answer an expert can hand you. A complex one only releases its answer to a probe you can afford to lose.

When Gutenberg's press flooded Europe with more books than anyone could read, the scholars who survived the deluge did so not by consuming more—but by deciding better.

The customer interview is product discovery's most trusted tool — and its most dangerous, because founders routinely design conversations that produce the answers they already believe.

In 1971, a Nobel laureate wrote one sentence that predicted the defining constraint of modern business — and most founders are still misreading it.

The science of locking in your best judgment before fatigue, pressure, and distraction corrupt it — and why the most effective founders treat pre-commitment as an operating system, not a…

Most founders drown in data and call it strategy — the DIKW pyramid reveals exactly where the confusion happens and how to climb toward decisions that actually compound.

Five well-documented biases — anchoring, loss aversion, overconfidence, confirmation bias, and the sunk-cost fallacy — distort how founders structure equity, negotiate terms, and decide when to walk away; recognising them…

Bush's memex wasn't a technology prediction — it was a decision-making manifesto, and eighty years later most founders are still ignoring the core argument.

Founders don't lack courage—they lack a disciplined vocabulary for the specific kind of uncertainty that precedes all markets, all data, and all proof.

The bias that keeps founders pouring resources into failing ventures is not a logic error — it is an identity crisis dressed up as strategy.

Herbert Simon proved in 1971 that information abundance creates a scarcity of attention — and every founder who has ever lost a morning to a feed instead of a decision…

Clay Shirky was right that the feed was never the problem — but the deeper crisis for founders is not filtering inputs, it's structuring the decisions those inputs are supposed…

Most dashboards are built to display data — but the cognitive science of how operators actually make decisions demands something far more deliberate.

The information-overload crisis that founders blame on social feeds was solved once before—in wood and paper—and the design logic still works.

Investors have quietly stopped believing founder-reported numbers — and the replacement for belief is not more diligence, it is infrastructure. Here is the case for verified traction, and the honest…

Insights are not accidents — they arrive through three distinct cognitive pathways, and founders who learn to work each one deliberately will out-decide any competitor who is merely consuming information.

The Data–Information–Knowledge–Wisdom hierarchy has a contested, 90-year genealogy — and understanding it reveals why most founders are optimising the wrong layer.

The science behind decision fatigue is far messier than the productivity industry admits — and that ambiguity is precisely what founders should act on.

In a world where attention is the scarcest resource, the founders who win are not those who consume the most information—they are those who have built a deliberate system for…

The willpower-as-fuel theory that launched a thousand morning routines has largely collapsed under scientific scrutiny — here is what founders should do instead.

Owning every outcome — not just the wins — is the single trait that separates founders who scale from those who stall.

The move from employee to founder is not a career change—it is a cognitive rewiring that restructures every decision a founder makes, from resource allocation to risk tolerance to the…

Decision velocity is not a personality trait—it is a compounding operating system that separates founders who build durable advantages from those who wait for certainty that never arrives.

Most founders over-deliberate cheap choices and under-deliberate expensive ones — Bezos's one-way/two-way door framework fixes both failure modes simultaneously.

Operators who underask on raises, pricing, and scope don't just leave money on the table—they set a ceiling on everything that follows.

Independent-mindedness and a willingness to hold unpopular positions are not personality flaws in founders — they are measurable predictors of venture success.

Early manual effort is not a workaround — it is the mechanism by which durable competitive moats are built, and knowing when to stop is the discipline that separates founders…

When the payoff structure is nonlinear, the correct default is not caution — it is disciplined aggression, bounded by survivability.

Inaction feels safe but carries a measurable, compounding price—and for capable operators, it is the most common path to irrelevance.

The founder who cycles through reality faster than rivals compounds an advantage that no amount of strategic planning can replicate.

Cereal box theory of fundraising: how Airbnb's Obama O's story reveals the principle that scrappy, unscalable action beats waiting for the right funding channel

Grit is overrated: Credé et al.'s 88-study meta-analysis shows grit weakly predicts performance (r≈.18) and duplicates conscientiousness. Here's what actually d

Luck surface area explained: Naval Ravikant's four kinds of luck and a tactical checklist for founders to engineer more serendipity this quarter.

Start before you're ready: why most credential gates are self-imposed and how decisive action, not preparation, builds lasting founder competence.

Permissionless action is the operating system of every founder who ships. Learn why waiting for permission is the default failure mode of talented people.

Agency vs. luck in startups: research-backed framework for founders to separate what they controlled from what fortune handed them — and build better systems.

Melanie Perkins faced 100+ investor rejections before building Canva into a $40B company. Reverse-engineer her playbook on reframing rejection, affordable loss,

Agency gap: why locus of control and self-efficacy — not talent or credentials — compound into wildly different career and founder outcomes over time.

Nobody is coming to save your startup. Grounded in Rotter's locus of control and self-authorship theory, here's why founder agency is your only durable edge.

Self-efficacy vs confidence: how Bandura's four sources — mastery, vicarious experience, persuasion, physiological state — help founders build execution belief

"You can just do things" — tracing the founder-agency ethos from Steve Jobs 1994 to the 2024 meme, separating load-bearing truth from tech-elite entitlement.

High agency defined: use George Mack's jail-cell test to measure clear thinking, bias to action, and productive disagreeability in yourself and your team.

Founder agency starts with one insight: the world is built, not given. John Collison's "museum of passion projects" reframed as a call to decisive action.

Survivorship bias distorts the 'you can just do things' founder narrative. Here's what the data on mobility, failure rates, and capital access actually says — a

Ask forgiveness not permission: the practical framework for high-agency founders — when to act without sign-off and when caution is non-negotiable.

Capital platforms in developing economies: how software-mediated trust closes information, trust, and distribution asymmetries to unlock the $5.7T SME finance g

AI-verified diligence compresses document review, fraud detection, and risk scoring from weeks to hours—here is what it reliably catches and where human judgmen

From Mesopotamian clay tablets to W3C Verifiable Credentials, trace how humanity has scaled trust across strangers — and what it means for cross-border startup

Physical verification remains a high-signal trust tool in developing economies. Learn why hybrid physical-cryptographic systems beat digital-only KYC in thin-re

How transaction infrastructure—SAFE notes, e-signature, escrow, and KYC verification—unlocks cross-border startup investing. Evidence-led analysis for founders

How alternative credit data—mobile money, payments, supplier records, traction proofs—underwrites frontier startups where credit bureaus are thin. Global resear

Six structural gaps keep frontier founders under-capitalised. Discover which is the binding constraint and how verification infrastructure attacks several at on

How trust develops in developing economies, why capital flows through personal networks in low-trust markets, and how verification infrastructure substitutes fo

KYC-verified founders: why a 3-step identity check (government ID, proof of address, facial liveness) plus human review is the optimal trust layer for cross-bor