Receipts, Not Vibes: The Substantiation Standard

Ask an agency for the export behind their favorite case-study number and watch the meeting reschedule itself. Marketing doesn't lie, exactly — it rounds up until the truth stops being load-bearing.

This post defends one position: an objective performance claim must be substantiated, or it must become qualitative. If you publish a number, you can produce the receipt behind it — the export, the date range, the definition of what was counted — the day someone asks. If you can't, you don't get the number; you get an honest adjective instead. And because nobody's editorial discipline survives every deadline, the only enforcement that separates an honest marketing agency from a merely confident one is automated: a machine that reads every claim, every time, and stops the ones that fail.

That standard applies to our own copy first. Every document we commit — including the post you are reading — passes through an automated scanner that hard-fails any performance claim it cannot substantiate, before the standard is ever applied to anyone else's marketing. What follows is the framework the scanner enforces. We call it the Substantiation Standard, and it has three rules.

1. The Industry Has a Rounding Problem

Nobody sets out to fabricate a case study. The drift is quieter than that. A result that was mostly organic gets rounded, by the third retelling, into an absolute that no invoice supports. A strong quarter becomes a clean multiple because clean multiples fit on slides. A number gets inherited from last year's deck, and nobody re-pulls the data because the slide already looks right. Every edit shaves off a qualifier, and the qualifiers were the part that made the sentence true.

The damage is collective. A prospect reading two proposals cannot tell disciplined measurement from confident fiction, because both arrive in the same font — so every rounded-up claim taxes every substantiated one. This is not an indictment of any particular firm. It is what entropy does to copy when nothing is checking.

2. The First Copy We Scan Is Our Own

The easiest place to preach a substantiation standard is at someone else's expense. The only credible place to enforce it is at home. So the standard points inward first: every claim in our own marketing must produce its receipts on demand, and any absolute that outruns its receipt gets demoted to the qualified, qualitative version — mechanically, on every commit, not as a matter of editorial willpower.

That ordering matters more than it looks. An agency that grades only its clients' copy has an incentive problem: the grader's own slides stay ungraded. An agency that runs its own marketing through the same gate — before the gate ever touches a client's — has put its credibility where its checklist is. When we say the standard below is enforced, we mean a machine reads this post before it publishes, the same way it reads everything else we ship.

Here is the standard it enforces, one rule at a time.

3. Rule One: If It's a Number, You Can Produce the Receipt on Demand

An objective claim is anything with a unit on it — a multiple, a percentage, a rank, a dollar figure, a timeframe. Under Rule One, every objective claim maps to a named artifact that existed before the claim ran and can be produced the day someone asks for it. Not "the data is around here somewhere." A specific export, a specific date range, a specific definition of what was counted.

Worked example. Say a proposal wants to claim that a client's booked calls doubled in ninety days. Before that sentence ships, three artifacts have to exist: the call-tracking export covering those ninety days, the baseline period the doubling is measured against, and a written definition of "booked call" — including how the paid channels running alongside it get credited. Miss any one of the three and the sentence does not ship. Notice how far the rule reaches: it covers what the claim implies, not just what it states. "Doubled in ninety days" implies attribution, and the receipt has to reach as far as the implication does.

4. Rule Two: If You Can't Produce the Receipt, the Claim Becomes Qualitative

A missing receipt does not kill the story. It demotes the claim. That is the release valve that makes the standard livable: you keep the truth, you lose the arithmetic you could not back.

This is the rule our own copy lives under. When a result is real but the receipt cannot reach as far as the absolute implies — a growth story that was largely organic but not purely so — the sentence keeps the story and loses the absolute: it describes how the growth happened instead of asserting a purity the invoices don't support. Qualitative claims are legitimate: a plain description of how a result came about is still checkable; it just no longer borrows the authority of arithmetic it cannot repay. The trade runs the same direction every time — the copy loses precision it never actually had, and keeps the credibility it needs.

5. Rule Three: Enforcement Is a Machine, Not Willpower

Every agency claims editorial standards.

Standards enforced by willpower die at the first deadline, because the person enforcing them is also the person shipping late.

Rule Three closes the loop: a machine checks every claim, every time, and the machine does not have a launch to protect.

Here is ours. We built an automated forbidden-claims scanner into our workflow. It runs every time work is committed, and it hard-fails the moment a document makes a performance claim it cannot substantiate — a violation does not log a warning, it halts the process. The catalog of banned patterns lives in a data file outside the scanner's code, so the net extends without anyone touching the machinery. The scanner is one enforcement layer inside an operations system that audits its own health the same way. Enforcement runs in two tiers: Tier A hard-blocks the absolutes — fabricated numbers, all-organic absolutes, number-one claims, promised rankings, and promises that nothing can go wrong. Tier B flags gray-zone language and routes it to a human for review.

The worked example for Rule Three is the document you are reading. This post will pass through the same gate before it publishes — which is why the Tier A patterns above are described rather than quoted.

6. The FTC Wrote This Standard First: Marketing Claim Substantiation

None of this is a novel ethic; it is an old rule with new enforcement. The FTC's substantiation principle requires an advertiser to hold a reasonable basis — "adequate substantiation" — for objective claims before those claims run, and it covers implied claims as well as express ones (16 CFR Part 255, Section 255.2(a)). The Commission revised its Endorsement Guides on July 26, 2023, tightening the standards around testimonials and endorsements.

Set that standard next to the industry's habits and the gap is plain: substantiation is supposed to be a precondition, and rounding up treats it as paperwork to reconstruct later if anyone asks. We did not invent a standard. We automated one that already existed — and we point it at our own copy before we point it at anything else.

7. What We Will Hold Ourselves — and Our Clients' Copy — To

The commitments are simple to state because a machine holds us to them. We will not publish a number we cannot back with an artifact on demand. Where results proof does not exist yet, we will keep leading with structural proof — our pages already do this today, leading with commitments like "the 30-day re-audit clause is written into the contract," because a contract term you can read outranks a metric you would have to take on faith. Every claim we draft for a client will pass through the same scanner before it ships, under the same standing rules: no fabricated numbers, no all-organic absolutes, no number-one claims, no promised rankings, no promises that nothing can go wrong.

And here is the test to carry into any agency conversation, ours included: pick the number you like best in the pitch, and ask for the receipt. A shop that runs on substantiation will hand you an export and a definition. A shop that runs on rounding will handle the question — and that handling tells you everything. When we put a number in front of you, you will be able to ask, and we will produce it. That is the entire standard, and it fits on a sticky note: receipts, not vibes.