A customer sends you a testimonial you'd frame and hang on the wall: "ProofShow cut our onboarding time by 60% and paid for itself in the first month." It's specific, it's enthusiastic, and it's exactly the kind of concrete result that persuades. So you publish it. Then, months later, someone asks the uncomfortable question: how do you know it was 60%? You don't. The customer said it in good faith, but nobody measured it, and now that number is sitting on your homepage with your name attached. This is one of the most common — and most quietly risky — situations in testimonial marketing, and the instinct to just run the impressive quote is the wrong one. Here's what's actually going on and how to handle it.
The number stops being their claim and becomes yours
The core thing founders miss: when you publish a customer's specific, measurable claim, you adopt it. In the eyes of a regulator and a skeptical prospect, a testimonial that says "cut our costs 40%" is no longer just one customer's impression — it's a performance claim your company is now making, using a customer's mouth to make it.
In the United States, the FTC's position is direct: an advertiser must have substantiation for the objective claims a testimonial conveys, including any implied claim that the endorser's experience is what others can expect. A vague sentiment — "the team is a pleasure to work with" — carries no measurable claim and needs no proof. But the moment a testimonial contains a number (a percentage, a dollar figure, a timeframe, a multiple), you need to be able to back it up as if you had written it yourself. "A customer said it, not us" is not a defense; you chose to publish it.
This is the same principle behind the "results not typical" problem. If one customer genuinely saw a 60% improvement but most see 10%, running the 60% figure implies a typical result you can't support. Disclaimers like "results may vary" help but do not cure a fundamentally unsupported implied claim — the FTC has been explicit that a strong claim plus a small-print disclaimer can still mislead.
First, figure out which kind of claim you're holding
Not every impressive-sounding line is a substantiation problem. Sort the claim into one of three buckets before you decide anything:
- Subjective / experiential — "It saved us so much time," "the best tool we've used." No objective metric, nothing to prove. Publish freely; this is the safe majority of testimonials.
- Specific but verifiable — "cut our onboarding from 10 days to 4." There's a number, but it's a fact you can check against your own data, the customer's records, or a documented before/after. Verifiable if you do the work.
- Specific and unverifiable — "paid for itself in a month," "doubled our conversion." A hard number that neither you nor the customer actually measured, or measured in a way you can't reconstruct. This is the one that bites.
Most panic comes from treating a bucket-one sentiment as if it were bucket three, or from publishing a bucket-three claim as if it were bucket two. Do the sort first.
The four ways to handle an unverifiable number
Once you've confirmed you're holding a specific claim you can't substantiate, you have four honest options, roughly in order of preference:
1. Substantiate it — go back and get the proof. Often the fastest fix is to ask the customer where the number came from. Sometimes it's real and documented — they have the dashboard export, the before/after report, the invoice math — and you just needed to ask. If you can attach that record, the claim moves from bucket three to bucket two and you can run it as-is. This is the ideal outcome, and it's why a light verification step belongs in your intake; see how to prove your testimonials are real without embarrassing your customers.
2. Qualify it — publish the number with its context. If the result is real but clearly not typical, present it as a single case rather than an implied norm: attribute it precisely ("a 12-person agency that switched from spreadsheets"), and let the specificity signal that this is one story, not a promise. Concrete, narrow attribution is more believable and more defensible than a bare percentage floating on its own.
3. Soften it — keep the testimonial, drop the number. You rarely need the metric to keep the quote's value. "ProofShow cut our onboarding time dramatically and paid for itself fast" carries almost the same emotional punch as the 60% version without asserting a figure you can't defend. Edit for the claim, not for the voice — and always get the customer's sign-off on the edited wording, because you may not alter the substance of what they said. If the edit starts to sound like copy you wrote, you've gone too far; that's the exact failure mode described in why your testimonials sound fake and the edits that fix it.
4. Shelve it — don't publish. If the claim is central to the testimonial, you can't verify it, and the customer won't or can't stand behind the specifics, the right move is to not run it. A quote you can't defend is a liability regardless of how good it looks, and one unsupported number can cast doubt on every other testimonial on the page.
The habit that prevents this
The reason unverifiable claims pile up is that verification happens after collection, when the enthusiasm has cooled and the customer has moved on. Flip it: at the moment you collect, if a testimonial contains a number, ask one follow-up — "That's great — where did the 60% come from, so we can show it accurately?" Either you get the proof and a stronger testimonial, or you learn it was a figure of speech and you soften it on the spot. Asking at collection time costs one sentence. Asking after publication costs a takedown, an awkward email, and sometimes a credibility hit you can't see happening. Treat every number in a testimonial as a claim you'll have to defend, and handle it before it's live, not after someone asks.