Six ways to back up a revenue number.

1. Nothing — the number on your landing page

“Trusted by 400 businesses. $40k MRR.” Typed by you, connected to nothing. It is still the most common form of revenue claim on the internet.

Proves: nothing, but it does set an expectation, and a figure that later turns out to be stale is worse than no figure. Costs: nothing. Use when: the reader is not really evaluating you — browsing, early curiosity. It is not an argument, it is framing.

2. A screenshot of your dashboard

The default upgrade, and a smaller upgrade than it feels like. An image cannot show whose account it is, whether a filter was applied, whether the figure is recurring revenue or total volume, or whether a number was edited before the capture.

Proves: that the dashboard renders. Costs: a minute, and some credibility with readers who have learned to discount images. Use when: honestly, rarely — and never as the strongest thing in a deal. The longer argument.

3. A live dashboard view

Screen-sharing your processor on a call, or adding someone as a read-only user. Genuinely more convincing, because the reader watches it load rather than taking an image on faith.

Proves: a great deal, to the people in the room. Costs: more than founders realise — default views carry customer names, individual payments, disputes and refunds. And it does not survive forwarding, so it convinces the attendee rather than the person who signs off. Use when: one trusted counterparty, one conversation. How to do it without exposing customer data.

4. Bank statements

Standard in lending and in acquisition diligence, and poorly matched to recurring revenue. Deposits arrive batched and net of processor fees, so reconstructing what recurs from what landed is interpretation rather than measurement.

Proves: money genuinely arrived. Costs: full disclosure of your banking activity to whoever receives them. Use when: a lender or buyer asks, under an agreement. Never for a public claim.

5. An accountant’s letter or prepared accounts

A professional attests to figures for a period. Real weight in a negotiation, and close to useless for the everyday version of this problem: it is a point-in-time document that starts ageing the day it is signed, and it is slow and billable enough that nobody produces one for a prospect who is merely curious.

Proves: a qualified third party reviewed a defined period. Costs: fees and lead time. Use when: diligence, lending, a material contract.

6. A direct read of the payment processor

A read-only connection to Stripe, Paddle or a similar system computes the figure from the billing data itself, continuously, without the company typing anything. It is the only method on this list that is both current by default and cheap enough to leave running.

Proves: exactly what that processor processed, counted under a stated rule, as of a stated date — no more. Costs: a scoped key, which can be revoked, and the discipline of not being able to edit the result afterwards. Use when: anyone outside the company is evaluating you and is not going to sit through a call.

Two things determine whether a given implementation is worth anything: whether the counting rule is published and whether the reading date is shown. A figure with neither is a badge, not a verification.

They stack — they do not compete

The common mistake is treating this as one choice. A processor-read record does not replace accountant-prepared numbers in a sale, and audited accounts do not give you something to put beside your pricing. Most businesses end up with two: a continuous public one for everyone who is merely deciding whether to believe you, and a formal one produced on demand for the rare process that requires it.

The public one is the one almost nobody has, and it is the one that gets asked for most often.

A note on tools that do not exist

This category attracts comparison articles faster than it attracts working products. Several tools recommended in “best verified revenue tools” round-ups turn out, on inspection, to be landing pages collecting emails or repositories with a handful of commits and no deployment.

Two checks, a minute each, before you trust any such list. Open the site and try to reach a signup — a “launching soon” form is not a product. If it is open source, look at the commit history and the date of the most recent one. Both checks are worth doing on us as well as on anyone we compete with. How Thonest compares to the products that do exist.

Questions founders ask

How is startup revenue verified?

In practice, by one of six things: nothing at all, a screenshot, a live dashboard view, a letter or statement from an accountant, a direct read of the payment processor, or a full audit. They differ enormously in what they prove and what they cost. Reading the payment processor is the only one that is both continuous and cheap, which is why it has become the default for software and subscription businesses.

Which method should I use?

Match it to who is asking. A curious reader needs a checkable public figure. A customer running a vendor review needs continuity over years. An acquirer in diligence will want accountant-prepared numbers regardless of what you publish, and a regulated or late-stage process needs an audit. These stack rather than compete — a processor-verified record does not replace an audit, and an audit does not give you something to put on your pricing page.

Is a bank statement good proof of revenue?

It proves money arrived, which is more than a screenshot does, and it is badly suited to recurring revenue. A statement shows deposits, often netted of processor fees and batched across customers, so reconstructing MRR from it is guesswork. It also exposes your full banking activity to whoever you send it to. Useful in a lending or diligence context; wrong for a public claim.

Does an accountant's letter count as verification?

It carries real weight in a diligence conversation and almost none in a public one, for two reasons: it is a point-in-time document that ages immediately, and the reader has to trust both you and a firm they have never heard of. It is also slow and billable, so nobody produces one for a prospect who is merely curious.

Why do so many 'verified revenue' tools not seem to exist?

This category attracts listicles faster than it attracts products, and many of the tools recommended in them are placeholders or abandoned side projects. Two checks take a minute: open the site and see whether you can actually sign up, and if it is open-source, look at the commit history and the date of the last one. Do both before trusting a comparison article that recommended it.

What can't a processor read tell you?

Anything outside that processor — invoiced contracts, revenue on another platform, cash collected elsewhere. It also says nothing about profitability, costs, runway or contract terms. It is a narrow claim: this is what this processor processed, counted this way, read on this date. The narrowness is what makes it reliable, and any service describing it as an audit is overselling it.