Definition
Most startup revenue you read online was typed in by the company. Some of it was read from the money. The two look identical on a page, and the difference is the whole reason one of them is worth anything.
Self-reported means the company supplied the figure. A form, a profile field, a slide, a landing page. Nothing connects it to the account the money arrived in, so its accuracy is a property of the person who typed it rather than of the system that holds it.
Verified means the figure was read from the payment processor itself. The company chose to connect it and chose to publish, but did not choose the number.
That is the entire distinction, and it is not a spectrum. Either something other than the company produced the figure, or it did not.
The failure mode people imagine is invention. The common one is much duller: a number that was true once and was never revised. A founder types their MRR into a profile in March, has a good March, and the field still says March in November. Nobody lied; nothing updated.
Close behind it are honest ambiguities. Is that monthly or annual, divided by twelve. Gross or net of refunds. Does it include the one-off setup fees, the consulting on the side, the deal that has been signed but not paid. Every one of those is a defensible choice, and every one moves the figure, and none of them is visible to a reader looking at a single bold number on a profile.
Verification does not make a founder more honest. It removes the room in which all of this happens by accident.
This is the part that does the real damage. Product directories and startup listings often carry both kinds of figure in one layout, in the same typeface, in the same position on the card. The only difference is a small mark — and the absence of a mark is a signal almost nobody consciously registers.
The result is that verified figures lend their credibility to unverified ones sitting beside them. A reader skims ten listings and comes away with an impression of the whole page rather than a per-listing judgement. Verification that is not obvious does not function as verification.
Whether you are reading someone else’s number or publishing your own, the same three apply, and most published figures fail at least one.
Where did it come from? Name the system. “Read from Stripe” is an answer. “Verified” on its own is a word.
What rule turned payments into this number? Raw charges are not MRR. Somebody decided how annual plans, trials, refunds and one-off payments are treated, and a figure whose counting rule is not published cannot be compared to any other figure.
When was it last read? A verified figure from eighteen months ago is a historical fact being presented as a current one. Freshness is part of the claim.
Worth stating plainly, because the term gets stretched until it stops meaning anything. Reading a payment processor tells you what that processor processed. It is not an audit. It says nothing about profitability, runway, contracts, or whether the business is any good. Revenue collected outside that processor is not in it.
A narrow claim that is exactly true is worth more than a broad one that needs defending — and a service that describes a processor reading as an audit has told you something about its standards.
The conversation stops being about you. A self-reported number puts your character in the room: the reader is deciding whether to trust a person they have never met. A verified one moves the question to the method, which they can examine without any feelings about you at all.
That is a better position to argue from even when your number is modest, because “small and checkable” beats “larger, and you would have to take my word for it” with anyone who is actually evaluating rather than browsing.
Which processors can be read directly · why a screenshot is not verification · what our own mark does and does not claim
Connect your processor read-only and the figures come from your billing instead of from you. Nothing is public until you say so.