A customer sends you a glowing testimonial. You are about to paste it onto the homepage when you notice the problem: they credit a feature you have never built, or they claim "ProofShow saved us $40,000 a year" when you have no idea where that number came from, or they say the integration was "instant" when it actually took them a day. The praise is sincere. The facts are wrong.
This is more common than it sounds, and more consequential. A testimonial is not just marketing copy — in most jurisdictions it is an advertising claim, and a false claim does not become safe just because a customer said it instead of you. You are responsible for what appears on your page. Here is how to handle it without throwing away a good testimonial or embarrassing the person who gave it to you.
First, sort the error into one of three types
Not all factual errors carry the same risk. Before you do anything, identify which kind you are dealing with.
- A harmless detail. They said "I signed up in March" but it was April, or they called your plan "the Pro tier" when it is now "the Growth tier." Nothing rides on it. These you can fix silently or leave alone.
- A claim you cannot substantiate. "It doubled our conversion rate." "We saved $40,000." These are results claims, and they are the ones that get companies into trouble. The number may be completely true for that customer — but if you cannot back it up and a regulator or competitor asks, "true for one customer" is not the same as "substantiated."
- A claim about your product that is simply false. They praise an AI feature you do not offer, or describe an integration that does not exist. Publishing this misleads every reader, even though the customer meant well.
The middle and last categories are where the work is. Treat them differently from the harmless one — and never let a results claim onto the page without a paper trail. We go deeper on the verification side in our guide on how to verify testimonial authenticity.
The rule: you may shorten, you may not rewrite
There is a bright line in correcting testimonials, and it is worth stating plainly: you can remove a false or unverifiable claim, but you cannot put words in the customer's mouth.
Trimming is fine. If a customer's quote is 90% accurate and 10% wrong, you can publish the accurate part and drop the inaccurate sentence — as long as the edit does not change the meaning of what remains. That is ordinary editing, the same as cutting a quote for length.
Inventing is not fine. You cannot change "it doubled our conversions" to "it improved our conversions" and present it as their words, even if the softer version is more defensible. You cannot add a feature name they did not mention. The published quote has to be something the customer actually said or would unhesitatingly endorse. The moment you author the claim yourself, it stops being a testimonial and becomes a fabricated endorsement — which is exactly the thing testimonials are supposed to protect you from. We make the broader argument in why fabricated-sounding testimonials destroy trust.
How to fix each type
Harmless detail: correct it for accuracy, no drama. If it changes nothing material, you do not even need to go back to the customer. Just publish the right plan name.
Unverifiable results claim: go back to the customer and ask for the basis. "We'd love to feature the $40,000 figure — can you share roughly how you calculated it, so we can stand behind it?" One of three things happens: they explain it and now you have substantiation; they realize it was a rough guess and you soften it with their agreement; or they cannot back it at all and you drop the number and keep the rest. All three outcomes are good. The number you publish should be one you can defend.
False product claim: you must remove it. Go back warmly: "This made our day — one small thing, we actually don't have the AI summarizer you mentioned; I think you might be describing the smart filters. Mind if we adjust that line?" Customers are almost always happy to help. They wrote the testimonial to support you, not to create a problem, and they would rather be accurate than be quoted saying something untrue.
Always go back to the customer for anything that matters
The single most important habit here: for any non-trivial correction, get the customer's sign-off on the final wording. This protects you three ways at once. It keeps the published quote truthful. It keeps it genuinely theirs, so it remains a real endorsement and not your invention. And it protects the relationship — a customer who later sees their words altered without permission feels misrepresented, even if your edit was reasonable.
A simple confirmation does the job: "Here's how we'd like to feature your words on the site — does this look right to you?" A one-line yes is all you need, and it is worth keeping on file.
When to walk away from the testimonial
Occasionally a testimonial is wrong in a way you cannot fix. The customer insists on a claim you cannot substantiate and will not soften. The error is structural — the entire quote is built around a benefit your product does not deliver. The person turns out not to be a real customer at all.
In those cases, do not publish it. A testimonial you cannot stand behind is worse than no testimonial: it invites exactly the scrutiny social proof is meant to avoid, and one debunked claim casts doubt on every other quote on the page. You have plenty of other customers. Use the ones whose praise is both genuine and true.
Make corrections part of the workflow, not a fire drill
Factual errors stop being scary when checking for them is a step in your process rather than a surprise at publish time. ProofShow keeps the original submission, the customer's approval of the final wording, and the date — so every published quote has a record showing it is accurate, attributable, and signed off. When you can prove a claim is true and the customer endorsed the exact words, a factual error is a five-minute fix instead of a liability.