Reading a startup's true financial state in 48 hours.

When you have 48 hours to understand the real financial state of a startup, whether as an investor deciding to invest or as an incoming manager deciding whether the situation is recoverable, the financial statements are the worst place to start. The statements are the story the company tells about itself, prepared to present the business in the best defensible light, and the gap between that story and the truth is exactly what 48 hours of focused reading is meant to find.

The truth is in a different set of numbers, and the discipline is knowing which ones to ask for and what their answers mean. Most of what matters can be established in two days, because the questions that reveal the real state are specific, and the answers either come quickly or the delay itself is the answer.

Start with the bank statement, not the P&L

The first document to ask for is the bank statement, the last three months, raw. Not the cash flow statement, which is prepared, but the actual bank transactions. The bank statement is the one document that cannot be dressed up, because it records what actually happened to the cash.

The bank statement answers the questions that matter most, fastest. How much cash is actually there today. What the real monthly burn is, read from the net movement rather than from a prepared figure. Whether revenue is actually arriving, visible as real inbound payments rather than as invoiced-but-uncollected receivables. Whether the company is paying its bills on time or stretching payables to manage cash. The pattern of the cash over twelve weeks tells you more about the real state than the annual accounts do.

The specific thing to look for is the trend. A company with low cash but improving net movement is in a different state from a company with more cash but accelerating burn. The slope of the cash line over the last three months is the single most informative number in the first hour, and it comes straight off a document that cannot be edited.

Read the revenue quality, not the revenue number

The revenue figure on the P&L is the most-managed number in the company, so the work is to decompose it into the parts that reveal quality.

Recurring versus one-time.

What share of the revenue recurs and what share is one-time. A company reporting strong revenue that is mostly one-time project work is a different business from one with the same revenue recurring monthly. The recurring share is the real revenue base; the one-time share is a sales achievement that has to be repeated next period to stand still.

Concentration.

What share of revenue comes from the top one, three, and five customers. A company where the top customer is 40% of revenue has a different risk profile from one where the top customer is 8%. Concentration is the number most likely to be absent from the prepared materials and most likely to change the valuation once you have it.

Collected versus invoiced.

What share of recognised revenue has actually been collected. Revenue recognised but sitting in receivables for 90-plus days is revenue the company may never see, and a growing receivables balance against flat collections is a sign that the revenue quality is worse than the P&L shows.

Cohort retention.

Whether the customers acquired twelve months ago are still paying. A company growing revenue while losing old customers as fast as it adds new ones is running to stand still, and the growth is masking a retention problem that will surface the moment acquisition slows.

Find the liabilities the statements understate

The balance sheet shows the liabilities the company has recorded. The 48-hour read is about finding the ones it has not, because the gap between recorded and real liabilities is where the unpleasant surprises live.

The questions to ask directly: are there deferred payments, supplier arrears, or payment plans not visible as debt on the balance sheet. Are there tax liabilities, VAT, payroll taxes, corporate tax, that are accruing and not yet paid. Are there contractual commitments, minimum spends with providers, take-or-pay clauses, lease obligations, that do not appear as liabilities until they trigger. Are there contingent liabilities, customer disputes, regulatory matters, pending claims, that could become real liabilities.

In a fintech specifically, there are categories that matter more: safeguarding and client-money positions, regulatory capital requirements, provider liabilities that sit off the standard balance sheet, and chargeback or fraud exposures that may not be fully provisioned. The 48-hour read in a regulated business asks these explicitly, because they are the liabilities most likely to be material and least likely to be volunteered.

The speed of the answer is itself informative. A company that can produce a clean liability schedule in a day has its house in order. A company that takes a week to tell you what it owes, or produces it in fragments, is telling you something about both its controls and its candour.

Reconstruct the real burn and runway

The runway figure the company gives you is built on the burn the company reports and the cash the company shows. Both are usually optimistic, so the 48-hour read reconstructs the runway from the ground up.

The real burn is the net cash movement from the bank statement, averaged over three months and adjusted for any one-time items that flatter the recent figure. The real cash is the bank balance minus the near-term obligations that have to be paid regardless: payroll, taxes due, supplier arrears that cannot be stretched further. The real runway is the real cash divided by the real burn, and it is almost always shorter than the figure the company presents.

The adjustment that matters most is the one-time items hiding in the recent burn. A company that took a large prepayment last month shows artificially low burn that month, and a runway built on that month overstates the real position. The three-month average smooths this, and the question “what one-time items affected cash in the last three months” surfaces the rest.

Read the unit economics for the floor

The last piece in the 48-hour read is the unit economics, because they determine whether the business has a floor or whether more revenue makes things worse.

The question is whether the company makes money on a customer over that customer’s life, after the real cost of acquiring and serving them. A company with positive unit economics has a floor: it can cut growth spend and approach profitability. A company with negative unit economics has no floor, because every new customer loses money and growth makes the burn worse, which means the apparent growth is actually the mechanism of the decline.

The 48-hour version does not need a perfect model, but at least the rough answer: does a typical customer, over a reasonable life, generate more margin than they cost to acquire and serve. If the answer is clearly yes, the business has a recoverable core. If the answer is clearly no, no amount of growth fixes it and the situation requires a different intervention. If the answer is unclear in 48 hours, that uncertainty is itself a finding, because a company that cannot show you its unit economics quickly may not know them.

What the 48 hours tells you

At the end of the read, the picture resolves into one of a few states. A company with improving cash, quality recurring revenue, clean liabilities, honest runway, and positive unit economics is in good shape and the statements probably told the truth. A company with deteriorating cash, low-quality or concentrated revenue, hidden liabilities, overstated runway, and negative unit economics is in trouble that the statements were arranged to obscure. Most companies are somewhere between, and the value of the 48-hour read is locating them precisely enough to make the decision.

The discipline is to work from the documents that cannot be dressed up toward the ones that can, and to treat the speed and quality of the answers as data in their own right. The numbers matter, but how readily the company can produce them, and how well they reconcile with the documents that cannot be edited, often matters more. A company in honest difficulty answers quickly and reconciles cleanly. A company hiding its state answers slowly and reconciles poorly, and that pattern is visible well inside 48 hours.

The financial statements are where the company tells you what it wants you to believe. The bank statement, the revenue decomposition, the liability schedule, the reconstructed runway, and the unit economics are where the company tells you the truth, whether it means to or not. Two days spent on the second set is worth more than two weeks spent on the first.