Video, Voice or Text: Which Testimonial Should You Ask For

Rank them by the only thing that actually matters — how hard each one is to fake:

1. Video. A real person, visibly meaning it. Hardest to fabricate, most persuasive.
2. Voice over a photo of the work. Nearly as strong, and far easier to obtain. A real voice is very hard to fake.
3. Text. A long way behind. Anyone can type a sentence and put a first name under it.

Text is the weakest format, and it is the one almost every small business uses — because it is the one you can produce without asking anyone for anything. That convenience is precisely why it persuades nobody: a viewer knows you could have written it yourself in four seconds, because you could have.

The conversion numbers you will see are invented

You will find pages claiming video testimonials convert 80% better than text. Or 25%. Or that they triple sales.

None of these has a source. Chase any of them and you land on a marketing blog citing another marketing blog citing nothing. I am not going to add a number to that pile, and neither should you when writing your own material — a number without a source is a decoration, not a fact, and nobody who checks will cite it.

What can be said without inventing anything is the mechanism, and the mechanism is enough to decide with: proof works to the degree that it could not have been manufactured by the person who benefits from it. Everything below follows from that one sentence.

Why text is so weak

A written testimonial — “Amazing service, highly recommend! — Maria” — has three fatal properties.

It is trivially fabricable. You could have typed it. The viewer knows you could have typed it. There is nothing in the object itself that proves a human being other than you was involved, and no amount of sincerity fixes that.

It has no voice. The hesitation, the emphasis, the small laugh, the way a person’s tone lifts when they mean something — all of the evidence is stripped out, and what remains is a sentence, which is the format advertising has always used.

It gets tidied. This one is fatal and nearly universal. Nobody transcribes a customer word for word. You fix the grammar. You cut the “honestly”. You make it into a better sentence. And you have now written her testimonial for her, in your voice, with her name on it — which is a small fabrication, and reads as one.

Text is not useless. It is simply the format that gives up almost everything that made the proof valuable, in exchange for being easy.

A screenshot of a real review beats a typed line

One version of text carries more weight than the rest. A screenshot of a real Google review is harder to dismiss than the same words typed onto your page. The stars are there, the platform’s frame, often a reviewer photo and date. None of it is proof — a faker could stage it — but it is more than a bare sentence with a first name, because it points at a place the viewer can go and check.

So if text is all you have, take the screenshot, not the retype. Turning a written review into a post keeps the evidence that retyping throws away.

Why voice-over-a-photo is the underrated one

It is treated as the fallback when video fails. It should be the default.

A voice cannot be typed. The stumble, the pause, the real intonation of a person who is pleased and slightly self-conscious — those are all present, and they carry nearly the full weight of a video. Meanwhile the photo shows the thing the viewer actually wants to see: the colour, the plate, the repaired car.

The endorsement and the evidence arrive together, in the same three seconds. And it is enormously easier to obtain, because nobody has to be on camera — which removes the objection behind most refusals. “I look terrible today” is not a complaint about your work; it is a complaint about being filmed, and this format simply deletes it.

A voice testimonial that exists beats a video testimonial that does not. The craft of it is worth getting right.

Why video still wins when you can get it

A face makes the endorsement unfalsifiable in a way nothing else does. The viewer sees an actual person, in an actual place, visibly pleased — and the doubts they were holding (is this made up, is this a friend of the owner) become much harder to sustain.

Video is also the only format the discovery channels want. Reels, TikTok and YouTube Shorts are built to push vertical video to people who have never heard of you. A photo with a voice over it can be posted there and will do fine, but it is not what those feeds were designed to promote.

So video is worth wanting. It is just not worth pressing for — because a customer talked into being filmed produces something flat and hurried, and a bad video is far worse than a good voice note.

What it looks like at the counter

Picture a barber on a Saturday. A regular gets out of the chair, looks in the mirror, and says the thing every barber wants to hear — that’s the best it’s looked in a year. Said like that, that sentence is the testimonial. The job now is to not lose it.

Reach for a comment card and the moment cools: he writes “Great cut, thanks”, flatter than what he said aloud, because writing sands the life off speech. Ask him to be filmed and he stiffens, checks his hair, delivers a hurried line.

But hold up the phone, catch the last ten seconds of what he was already saying, over a photo of the fresh cut — and the real sentence survives, in his own voice, the small laugh still in it. Same customer, same three seconds, three different pieces of proof. The format you reach for decides which one you keep.

The format matters less than the words being real

Here is the trap. Having read a ranking, the temptation is to go and get video at any cost, and to improve it.

Do not. A polished video is worse than an unpolished voice note, and it is worse than nothing.

The moment you feed her the line, do a second take, or trim the “ehm” out, you have converted your best format into your weakest: something that looks produced, which reads as advertising, which is exactly what text already was. You have gone all the way round the circle and arrived back at a sentence you wrote yourself.

Whatever the format, the words go out verbatim — not tidied, not shortened, not improved, not in the subtitles. A testimonial that reads better than the customer speaks is a fake one, in video, voice or text alike.

But what if she only wants to write something

Then let her, and take it — but ask one small thing first: “Could I just get that on a quick voice note?” If she says no, fine; a written line from a real, willing customer is worth having as a supporting quote. The refusal is the filter doing its job — a testimonial pushed out of someone who did not want to give it is the flat, fake-sounding kind, whatever format it lands in.

What you do not do is write it for her. Three plain words go up as three plain words. The caption is yours; her words are hers. Fixing her grammar or fattening her sentence turns a modest true thing into a polished false one.

Ask for the format she will say yes to

Default to voice with a photo of the work. Upgrade to video when the customer’s own enthusiasm makes it obvious — when she asks if she should look at the phone.

Use text only for what it is good for: a supporting quote under something real, or a written note from a client who is never going to be recorded. Never as your primary proof.

And whichever you get, do not improve it. How to make the choice at the counter, in one glance, is the practical version of this.