Proof, Not Promises: Why the Next Era of African Venture Runs on Verified Traction

In the summer of 2021, JPMorgan Chase paid $175 million for a startup called Frank, whose founder told the bank it had 4.25 million customers. It had fewer than 300,000. When the bank’s due-diligence team asked to see the user list, Charlie Javice paid a mathematics professor to manufacture one: millions of synthetic names with plausible email addresses, invented on deadline for about $18,000.¹ The fraud surfaced only after the acquisition closed, when a marketing campaign to the “customer base” bounced. In September 2025 she was sentenced to 85 months in federal prison, with restitution set at $287.5 million.²

Two years after the Frank deal, the U.S. Securities and Exchange Commission described an even purer specimen of the genre. Tingo Group, a fintech with Nigerian operating subsidiaries, had reported $461.7 million of cash in its accounts. The SEC’s complaint put the actual combined balance at less than fifty dollars, and described “a brazen scheme using phony records and fictitious entities” to make the companies “appear highly profitable.”³

The direct answer to the question this piece exists to answer: self-reported traction is no longer a credible basis for investment decisions, and the fix is a two-layer one: cryptographic attestation that makes a founder’s documents tamper-proof, plus confirmation from the human counterparty who signed them. Neither layer works alone. Together they convert traction from a claim into an asset.

Why doesn’t due diligence catch fabricated traction?

The uncomfortable answer is that diligence is artisanal, and artisanal does not scale. The canonical study of venture practice, Gompers, Gornall, Kaplan, and Strebulaev’s survey of nearly nine hundred institutional VCs, found that investors spend an average of 118 hours on due diligence and call ten references for each deal they close.⁴ That cost structure was built for large checks in information-rich markets. Spread it over a $50,000 angel check in Kampala and the arithmetic stops working entirely; for most early-stage deals in emerging markets, the verification actually applied to a founder’s numbers is a fraction of those hours, or none.

And the failure rate at the top of the market shows what happens even when the hours are available. JPMorgan, a bank with effectively unlimited diligence resources, bought Frank. SoftBank led a $170 million round into the social app IRL on the strength of 12 million claimed users; a board investigation later concluded 95 percent of them were bots.⁵ HeadSpin’s founder reported $80 million of annual recurring revenue that was actually about $10 million, supported by invoices he fabricated himself; the board cut the company’s valuation from $1.1 billion to $300 million.⁶ These were not unsophisticated victims. They were sophisticated victims of an information architecture in which the only source for a startup’s numbers is the startup.

Africa’s ecosystem has lived its own version. Ghana’s Dash raised a record $32.8 million seed and claimed, within months, to have processed $1 billion in transaction volume; internal examinations later found inflated user numbers and roughly $25 million unaccounted for, and the company shut down.⁷ The response from the region’s investors was telling. “Enhanced both qualitative and quantitative due diligence will need to be done even at the earlier stages,” Norrsken22’s Lexi Novitske told TechCrunch⁷. It is a sentence that concedes the current system verifies too little, too late. PwC’s global economic crime survey suggests the base rate the system is up against: 41 percent of executives report fraud at their own companies within the past twenty-four months.⁸

The pattern across every case is structurally identical. The founder did not lie about something peripheral. Each fabricated the specific metric their category rewards: users for a consumer app, ARR for enterprise software, cash for a fintech. Self-reported traction fails precisely where it matters most, because that is where the incentive to inflate is strongest.

What happens to founders when nothing can be verified?

Here is where the story stops being about fraud and becomes about infrastructure, because the founders who pay the highest price for unverifiable claims are the honest ones.

Verification in rich markets rides on rails most participants never notice: near-universal legal identity, deep credit bureaus, searchable registries, audited-statement norms half a century old. Those rails are thin to absent across Sub-Saharan Africa and South Asia. The World Bank estimates that more than 800 million people still lack official identification, concentrated in exactly those regions.⁹ FounderWise operates in Sub-Saharan markets where private credit bureau coverage stands at 11 percent of adults, against 66.7 percent in high-income OECD economies.¹⁰ A founder in Frankfurt inherits a credibility apparatus at birth. A founder in Kampala must build one by hand, deal by deal.

The cost of that missing layer has a number. The IFC puts the financing gap for formal micro, small, and medium enterprises at $5.2 trillion globally, with roughly $331 billion of it in Sub-Saharan Africa. Its diagnosis is informational, not moral: financial institutions consider these businesses too costly to serve because they “generally lack the financial records, business plans, and assets required for traditional credit assessments.”¹¹

Economics has known the mechanism for four decades. Stiglitz and Weiss showed in 1981 that when lenders cannot distinguish good borrowers from bad, they do not simply charge everyone more. Beyond a point, raising the price selects for worse risks, so they ration instead.¹² The result is the cruelest feature of low-information markets, and every East African founder has felt it: unverifiable claims do not get worse terms. They get silence. The deal that dies without feedback, the investor who stops replying, the “we’ll circle back next quarter”: these are not judgments about the business. Often they are the rational response of capital that has no affordable way to check whether the business is what it says it is. The financing gap, seen this way, is not primarily a capital shortage. It is a verification shortage. (We mapped the six structural gaps behind this in The Six Gaps Holding Back Founders in Developing Countries.)

What is verified traction actually worth?

If verification is the missing input, installing it should show up in prices. It does, everywhere researchers have looked.

Start with the oldest verification product in commerce: the audit, which is nothing more than paid third-party confirmation of self-reported numbers. Michael Minnis’s landmark study of private U.S. firms found that lenders charge audited firms significantly less and “place more weight on audited financial information in setting the interest rate.”¹³ Follow-on work quantifies the discount at roughly 0.47 percentage points on the cost of debt.¹⁴ Verification is not compliance overhead. It is a discount rate.

Marketplaces tell the same story in cleaner laboratory conditions. In a controlled field experiment on eBay, Paul Resnick and colleagues sold identical goods from a high-reputation verified identity and from new accounts; buyers paid 8.1 percent more to the verified seller.¹⁵ On Airbnb, the Superhost badge, a platform-issued verification of past performance, carries a documented price premium.¹⁶ Same asset, same description; the verified version is worth more.

Then there is the country-scale natural experiment. India’s Aadhaar-based eKYC cut the cost of verifying a bank customer from roughly $23 to $0.15,¹⁷ infrastructure the Bank for International Settlements describes as a deliberate strategy of providing “digital financial infrastructure as a public good.”¹⁸ On top of those rails, India’s Account Aggregator framework, consented sharing of verified financial data, had moved roughly ₹42,300 crore (about $5 billion) in loans by March 2024.¹⁹ When verification gets cheap, credit markets reorganize around it. That is the precedent East Africa’s capital markets have not yet had, and the one platforms are now racing to build. (On why such platforms matter disproportionately here, see Why Capital Platforms Are Critical for Developing Economies.)

What can cryptography prove, and what can it not?

Cryptography can prove that a document existed, that it has not been altered, and which key sealed it. It cannot prove that the document is true. That limit defines this entire category.

The technology for portable, tamper-proof claims went mainstream while nobody in venture was watching. The signature scheme FounderWise uses to seal documents, Ed25519, is not exotic cryptography; it is the workhorse standardized by the internet’s engineering body in RFC 8032,²⁰ the default key type securing SSH connections to most of the world’s servers,²¹ and the machinery behind the signed software provenance that npm now publishes for packages downloaded hundreds of millions of times.²² In May 2025 the W3C made Verifiable Credentials 2.0 a full web standard (a common envelope for digitally signed claims²³), and the European Union has legally obligated every member state to issue digital identity wallets to its citizens by the end of 2026.²⁴ Identity, software, media, money: every high-stakes domain of digital life has converged on the same architecture. Sign the claim. Log the signature. Let anyone verify.

Founder traction documents (the LOIs, purchase orders, invoices, and contracts that anchor every emerging-market pitch) are among the last high-stakes claims still circulating as unsigned PDF attachments. Applying the standard architecture to them is overdue. But here the argument must slow down and say something most explainers in this category will not.

Cryptography has a hard limit, and the standards bodies state it themselves. The W3C’s specification says, in normative text: “Verifiability of a credential does not imply the truth of claims encoded therein.”²³ The Ethereum Attestation Service, infrastructure that has processed over nine million signed attestations, concedes in its own documentation that “the value of an attestation comes from the relative reputation of the entity making the claim.”²⁵ Blockchain engineers call the general form the oracle problem: a system can guarantee the integrity of what is inside it, never the veracity of what came in from the world.²⁶

Call this the Truth Gap. A cryptographic hash of a purchase order proves three things perfectly: this exact document existed at this moment; it has not been altered since; this key attested to it. It proves nothing about whether the purchase order is real. A forged LPO hashes as immutably as a genuine one. Garbage in, cryptographically immortalized garbage out. Any platform that stops at the hash has built a faster pipe for unverified claims. That is why the interesting question was never how do we sign founder documents but who can truthfully confirm them.

How do you close the Truth Gap?

You close it with the oldest fix in finance, made portable: a human confirmation wrapped in a tamper-evident process. Every mature verification institution humanity has built is, at its core, exactly that. The notary is a witness whose seal makes the witnessing portable. The audit is a person checking claims against source records; as Minnis showed, lenders literally pay for that human layer.¹³ The back-channel reference check (calling “the people you find entirely on your own,” as one guide to venture practice puts it²⁷) is how investors already verify founders, ten phone calls and 118 hours at a time.⁴ The gold standard has always been the counterparty’s word. The problem was only ever cost and portability: a diligence phone call expires the moment it ends, and no one else can reuse it.

This is the gap the Proof Vault and its TRUST Line close, and it is worth being precise about the design. A founder uploads the document (the LPO from a distributor, the supply contract, the signed MOU) and it is sealed with an Ed25519 attestation hash. That is layer one: from that moment, the document cannot be quietly altered, backdated, or swapped. Then the system contacts the human who actually signed it (the customer who issued the purchase order, the partner who countersigned the agreement) through a signed, expiring link, without exposing their contact details to anyone. The signer confirms, in writing: yes, this is real, I stand behind it. That is layer two. The confirmation binds cryptographically to the exact hashed artifact, so what an investor later sees is not a founder’s claim but a counterparty’s confirmation of a tamper-proof document: a back-channel reference call that no longer expires, executed at near-zero marginal cost instead of billable hours. In effect, it is an API for counterparty confirmation in markets where the revenue systems have no API. (Fintech underwriters like Pipe and Capchase already verify revenue by reading it from live billing systems²⁸, but a Kampala distributor’s purchase order flows through no such rail. For those documents, the only oracle is the human who signed.)

FounderWise ranks confirmed proofs by the credibility hierarchy investors actually price: from pre-product signals through soft commitments up to hard commercial evidence. Founders learn, by using the vault, which proofs move capital. FounderWise’s $39 Investor-Readiness System teaches founders where each piece of evidence sits on the credibility hierarchy and what converts it to the next rung. Verification stops being a compliance cost and becomes a fundraising curriculum.

Honesty requires naming what this does not solve. A colluding counterparty can confirm a fake document; signer confirmation raises the cost of fraud from “edit a PDF” to “recruit a co-conspirator willing to put their name to it in writing”. That is a large improvement, not an infinite one. Confirmation also depends on the identity layer beneath it, which is exactly why the maturing of African ID rails (Nigeria’s 130 million digital ID enrollments, Ghana’s single-card KYC mandate, Kenya’s Maisha Namba²⁹) matters to this design: each strengthens the answer to who is this signer, on which is this claim true ultimately rests. No verification system abolishes fraud. The realistic goal is the one audit economics demonstrated: shift the cost of lying high enough, and the price of honesty low enough, that verified claims become the market default.

What should founders, investors, and development partners do now?

Each seat has one move: founders build the verified evidence file before the raise, investors start diligence from counterparty-confirmed claims, and development partners fund the record rails that turn identity investment into credit flow.

You are carrying the old model if your data room is a folder of PDFs only you have touched. You are ahead of it if a third party can verify your three strongest commercial claims without calling you.

Investors stopped believing founder-reported numbers somewhere between Frank and Tingo; the market simply hasn’t finished replacing belief with infrastructure. The founders who win the next decade in East Africa will not be the ones who pitch best. They will be the ones whose claims check out without a phone call, because in a capital market starved of verification, proof is the scarcest asset a founder can own, and the only one that gets cheaper to hold the earlier you start. The decision in front of you fits in a week: pick your three strongest commercial claims, seal them, and ask the people who signed them to confirm them in writing.

FounderWise’s Proof Vault seals your traction documents and routes them to the people who signed them for confirmation: founderwise.io.