The uncomfortable thing about a $115 million startup fraud: from the outside, it looked a lot like an ordinary founder-led company. One person owned the numbers, the customer history, the spreadsheet and most of the answers. This week, we look at why that setup hides plenty of things besides fraud, and four ways to fix it before the wrong number reaches your board.

Creators lock in holiday content calendars 90 days out, before brands figure out commissions. Waiting too long to launch an affiliate program means less runway to build demand and a missed shot at the best partnerships.

The 90-Day Holiday Sprint covers commissions, recruiting, and scaling a program at Day 30, 60, and 90.

Download the Free Guide

A founder owns the spreadsheet that calculates his company's annual recurring revenue. He does the math himself, every quarter and every year.

His board asks him to hire a chief financial officer to run day-to-day finance. He declines. They keep asking for more than two years, and he keeps declining.

He keeps a tight hold on operations, sales and record-keeping. Customer conversations run through him. Invoicing runs through him. He discourages employees from discussing the company's finances, with each other and with customers.

You know those habits. You've watched somebody else run a company that way, or you've done a version of it yourself on a Sunday afternoon with the laptop open, because the numbers are faster to do than to explain.

Those first three paragraphs come from the civil and criminal complaints against one founder. The company sold tools for testing mobile app compatibility and raised $115.7 million. It reached a valuation of about $1.1 billion on financial statements the founder had fabricated, and 29 investors bought stock at prices based on that number.

He reported first-quarter 2018 revenue of $6,043,369 and net income of $1,758,032. An auditor later found revenue of $1,300,381 and a loss of $274,250. He pled guilty and was sentenced to 18 months.

The fabrication is his alone. What he shares with you is an information structure, and from where his board was sitting, the two companies looked identical.

Two research teams published on venture fraud this year. Dyck, Fang, Hebert and Xu assembled 654 court cases against U.S. venture-backed startups between 2000 and 2023. Weiss and Radoynovska read the court files of 12 Silicon Valley ventures across 27 cases, companies that raised $1.8 billion between them and produced $687.6 million in losses and 73 years of prison sentences.

The first paper puts the rate this way: "Among US VC-backed firms founded since 2000 that raised at least $10 million, 1.75% have been involved in detected fraud." The figure counts cases somebody caught, so treat it as a floor. It's still 1.75%.

Almost every founder stays outside that number, and you will too. This article is about something else.

Weiss and Radoynovska found that the 12 founders protected their deceptions in three escalating ways. The first is passing blame to somebody else when questions arrive. The second is centralizing and controlling the flow of information. The third is entrenching organized secrecy inside the company.

The second one, centralizing and controlling the flow of information, is also a plain description of a founder who still does everything himself.

You keep the pricing model in your own head because explaining it takes longer than deciding it. The escalation from your largest account comes to you because you're the only one who remembers what you promised them in March. Finance runs on a spreadsheet you built, and the person who could take it over is a hire you keep deferring.

From outside the company, and from inside your own board meeting, the man hiding fabricated financials and the founder who has yet to hire a finance lead look the same. Both own the spreadsheet. Both declined the hire. Both answer the questions no one else can.

The behavior carries no signal, and that has nothing to do with your character.

Look at how these deceptions surfaced, in the seven cases where the court files describe it.

Start with the founder from the opening: an employee, reviewing the company's financial records, found customers who didn't exist. At a second company, an intern realized its confirmation forms were fakes: somebody had downloaded a form off the internet and edited it in Photoshop. At a third, a newly hired chief financial officer resigned after a few days and told the board the revenue figures were wrong, naming the specific transaction. At a fourth, the general counsel contacted an acquaintance at KPMG while scouting for a finance hire and learned that KPMG had never performed the audit the company said it had.

A first-time investor and their lawyer unwound a fifth with an ordinary reference check. Employees at a sixth leaked to regulators and the press. At a seventh, insurance companies probed, and the board opened its own investigation.

Prosecutors brought all seven cases. Not one of the discoveries was theirs. In each one, the first person to see the problem held routine access and was doing routine work, and in five of the seven that person collected a paycheck from the company they exposed.

Some founder is going to read that and conclude the lesson is to trust fewer people. And what does that buy him? The record answers it twice.

The company whose new finance hire resigned and went to the board filed for Chapter 11, survived, was acquired and still operates today. Four others in the study went bankrupt, entered liquidation or shut down. That's one case against four, and it settles nothing on its own, though it runs in the direction you'd expect.

The second answer is harder to get around. A founder who responds to this research by holding information tighter has adopted the second protection approach straight out of the paper. It will feel like protection while he does it.

Dyck, Fang, Hebert and Xu went looking for what predicts fraud. The answer, from their abstract: "Governance characteristics, rather than founder traits, are the strongest predictors of fraud." Those structural variables carry "5.1 times greater predictive power" than founder characteristics, and once you account for them, gender and age stop predicting fraud.

"Startups with founder-controlled boards are twice as likely to commit fraud as those with VC-controlled or shared-control boards." And a 10% higher investor ownership, on a converted basis, is associated with a 19.3% lower fraud rate relative to the mean.

The authors flag the limit themselves: the analysis "is not designed to establish causality." Founder-controlled boards predict more fraud, and prediction is as far as they take it. Founder control is also the common arrangement, 53% of the company-years in their panel, so the finding describes the default.

A finance lead who would have caught a fabricated number is the same finance lead who catches a pricing model that has been losing money for five months. The employee who found customers who didn't exist would also find the channel that stopped converting in April. The board that would have asked about the audit is the board that asks why the sales cycle got longer.

You dismantled that apparatus one deferral at a time, because the hire was slow and expensive and you were faster at the spreadsheet than at the handoff. Each of those calls was defensible on the day you made it. Together they leave nothing standing between a wrong number and your board deck, whatever put the wrong number there.

One of the 12 founders described the moment in his own sentencing hearing. "At that point, as opposed to being honest, I was not. At that point, as opposed to asking for help ... I did not. At that point, I made horrific decisions to try to buy time because I truly believed in what we were doing, truly believed." His revenue was falling, and the choice of what to say about it was his alone.

If you saw yourself in any of this, putting it back takes four moves, and all four serve one best practice: no important number, role or channel with a single person on it.

1. Make your claims verifiable. Weiss and Radoynovska close by asking how founders can be encouraged to craft "not just convincing stories but also verifiable ones." Take your last investor update, pick the three numbers carrying the most weight and name the person who could produce the underlying record inside an hour without asking you. If the answer is you all three times, you've written a convincing update that no one can verify. Hand one of the three records to somebody else before the next one goes out.

2. Fill the role somebody keeps asking for. The founder in the opening turned down more than two years of board requests for a finance lead. If you've been hearing a smaller version of that request, start there.

3. Install one question anyone can ask about any number without it landing as an accusation: what record backs this one? An intern ran a version of that check and turned out to be right.

4. Reroute one thing that runs only through you. The founder in the opening ran customer conversations and invoicing through himself, and he discouraged employees from discussing the company's finances. Each arrangement like that is one fewer person who could catch a wrong number. If anything of yours still works that way, pick one and step out of the middle.

The founder who owned the spreadsheet had a board that asked about his missing finance hire for more than two years. He had 29 investors who bought stock at prices built on his own arithmetic. He had prosecutors, eventually.

The one who caught him was an employee who could get at the records and read them.

You employ somebody who could do the same for you, on ordinary errors instead of fraud. If your numbers have one reader, add a second one this week.

Still paying GoDaddy or Namecheap prices? Porkbun sells most domains at cost for low, transparent registration and renewal pricing with no nonsense. Get free features like WHOIS privacy and SSL certificates, plus real human support 24/7, 365 days a year. Save $1 on your next domain name now.

Save On Your Domain Now

Thanks for reading. See you next Wednesday.
— Jason

P.S. This week is the problem fully grown. April's Two Sets of Books is how it starts: one number for the board, another for yourself