What to Do When a Testimonial Cites a Metric You Can No Longer Verify

The best testimonial you have is a specific one: "ProofShow cut our review cycle from three weeks to four days." Numbers convert because they're falsifiable — a reader trusts a claim more when it's precise enough to be wrong. But precision is a liability the moment you can no longer stand behind the number.

Here's the situation that catches teams off guard. A testimonial you collected eighteen months ago cites a hard metric — a 60% reduction, a 4x increase, $40,000 saved. It's on your homepage. Then something changes: the customer churns, the champion who gave the quote leaves, your product changes so the number would be different today, or you simply go looking for the source data and can't find it. Now you have a quote making a specific, measurable claim that you cannot substantiate if anyone asks.

This is not a rare edge case. Any testimonial with a number in it has a shelf life on the verifiability of that number, separate from whether the quote itself is still true. Here's how to handle it.

First, separate three different problems

People collapse these together and make the wrong call. They're distinct:

1. The number is unverifiable — you can't find the source data, but you have no reason to think it's wrong. The customer said it, you believed it at the time, the evidence is just gone.
2. The number is stale — it was true then, but the product or the customer's situation has changed enough that it wouldn't hold today.
3. The number is doubtful — you have a specific reason to think it was never accurate, or was exaggerated in the retelling.

These three call for three different actions. Unverifiable-but-plausible can often stay, with a tweak. Stale usually needs reframing as a point-in-time result. Doubtful should come down, full stop. The mistake is treating all three as "well, a customer said it, so it's fine" — or, at the other extreme, panic-deleting every numeric testimonial the moment you can't produce a spreadsheet.

The standard: could you defend it in a room with the customer present?

The practical test for whether a numeric testimonial can stay is not "do I have a screenshot of the dashboard." It's: if the customer who said this were standing next to you, would they repeat the number without flinching?

That reframes the question usefully. A testimonial is a representation by the customer, not a metric you published. The claim's credibility rests on their willingness to have said it and stand by it. If your champion said "you saved us about 40 grand a year" in a recorded call and would say it again today, the fact that you can't reconstruct the exact calculation two years later doesn't make the testimonial fraudulent — it makes it a customer's good-faith recollection, which is what testimonials always are.

The problem arises when you can no longer answer that question with a confident yes — because the person is gone, the relationship ended badly, or the number has drifted from anything currently true.

What to do in each case

Unverifiable but plausible, customer reachable: Ask them to reconfirm. One email: "We still feature your quote about the 60% reduction in onboarding time — is that still a number you're comfortable being attributed to you?" A yes gets you a fresh, dated confirmation and resets the clock. This is the cleanest outcome and it's worth the two minutes it costs.

Unverifiable, customer unreachable, no reason to doubt it: You can usually keep it, but strip the false precision. "Cut onboarding time by 60%" becomes "cut onboarding time by more than half" — still a strong, specific claim, but one that no longer invites a "prove the exact figure" challenge you can't win. You're not softening because the number was wrong; you're removing a precision you can no longer source, which is a different and honest reason.

Stale — was true, wouldn't hold today: Anchor it in time. "In their first year with us, [Customer] reduced review cycles from three weeks to four days." The past tense and the time anchor make it a documented historical result rather than an implied ongoing promise. This is the honest way to keep a genuinely impressive result that current conditions have moved past — and it's usually more credible than a vague present-tense claim anyway, because dated specifics read as real.

Doubtful — reason to think it was never accurate: Pull it. There is no version of "the number is probably wrong but it converts well" that's worth the risk. A single testimonial that a competitor, a journalist, or the customer themselves can publicly debunk does more damage than the testimonial ever did good — it retroactively taints every other proof point on your page, because now a reader has evidence you'll publish claims you can't back. The whole value of social proof is that it's trusted; one exposed exaggeration is expensive out of all proportion to its size.

Build the audit in before you need it

The reason this becomes a fire drill is that nobody owns the shelf life of a testimonial. It goes up, it converts, everyone forgets it exists until a legal review or a skeptical prospect surfaces it. A few lightweight habits prevent almost all of the pain:

• Record the source at collection time. When a testimonial with a number comes in, note where the number came from in the same place you store the quote — "self-reported by [name] on [date]" is enough. You're not building a compliance archive; you're leaving yourself a note about what kind of claim this is.
• Date every testimonial internally, even if you don't display the date publicly. A quote's age is the single best predictor of whether its numbers still hold.
• Re-confirm numeric testimonials on a cycle. Once a year, the quotes carrying hard metrics get a quick "still good?" email to the customer. Most say yes; the ones who've churned or gone quiet are exactly the ones you want to catch before a prospect does.
• Flag testimonials tied to features you're changing. When you deprecate, rename, or significantly rework something, the testimonials praising it need a look. The claim may still be true, but it may now describe something that no longer exists under that name.

The mindset shift

The instinct when you find an unverifiable metric is either to defend it ("a customer said it, we're covered") or to delete it in a panic. Both are shortcuts around the actual judgment.

The right frame is that a numeric testimonial is a dated, sourced representation by a specific person — and your job is to keep the display honest about all three of those properties. Keep the ones your customer would still stand behind. Time-anchor the ones that were true then. Soften the false precision on the ones you believe but can't source. And pull the ones you can't honestly defend in a room. Do that, and a stale number becomes a maintenance task instead of a credibility crisis.

The testimonials that survive this kind of scrutiny are the ones worth having. The ones that don't were quietly costing you trust the whole time — you just hadn't been asked to prove it yet.