A skincare client sent me their new Meta ads with a note: "We used the customer research this time." They had. The hook came straight from a Reddit thread, "my face looked like a pizza for two years". It sat inside this sentence: "Say goodbye to the days when your face looked like a pizza, with our clinically formulated, dermatologist-approved routine."

Every word of the quote was real. The ad still read like a brand wearing a customer costume.

Short answer: Customer language in ads fails when you copy the words and keep the brand's register. Vocabulary is which nouns your buyer uses. Register is how formal, hedged or blunt they are when they say it. A verbatim quote inside a polished brand sentence reads worse than either alone, because the sentence around it gives the register away.

The takeaways

Register is a separate layer from vocabulary. A 400-word Amazon review, a two-line Reddit reply and an App Store rant can name the same problem in three registers, and each fits a different placement.
The tell is a line nobody would say out loud. Read the hook as a sentence spoken to a friend in that exact moment. If it breaks there, the register is wrong.
Brand register can win. High-price, high-consideration and regulated categories often need it, so match the sentence shape to the evidence and keep the brand vocabulary.

Why does a real customer quote still read fake in an ad?

A real customer quote reads fake when the sentence around it speaks a different register. "My face looked like a pizza" is self-deprecating and a bit crude. "Clinically formulated, dermatologist-approved" is a label on a box. Put them in one line and the reader hears two speakers. The second one is obviously the company.

Most voice-of-customer guides stop at the harvest, which works (I have written about running the research itself). The trouble starts when a copywriter trained on a style guide rebuilds the quote into a "proper" sentence. The nouns survive. Everything that made them sound human gets edited out on the way.

What separates vocabulary, register and emotion in ad copy?

Ad copy borrows three layers from a customer quote, and teams usually collapse them into one thing called "brand voice". Vocabulary is the words for the problem: "breakouts", "pizza face", "cystic". Register is the delivery: casual or formal, hedged or blunt, technical or plain, sworn or clean. Emotion is what the person feels while saying it, whether that is embarrassment, resignation or anger at a product that failed.

You can get one layer right and the other two wrong. The pizza ad nailed vocabulary, kept the emotion, and then wrote it in the register of a pharmacy shelf. Readers catch a wrong register before they have parsed the words, the way you hear a call-centre script.

How do you read register off the source?

Read register off the place the quote came from, then match it to where the ad will run. A 400-word Amazon review is considered and structured, often written after the problem was solved. A two-line Reddit reply is quick, peer-to-peer and unguarded. An App Store rant is angry and short. Same product, three registers.

Pick the one that fits the placement and the buyer's awareness stage. A Reels hook for someone who has not yet named the problem wants the Reddit register. A retargeting ad for someone comparing two products can carry the review register. Let the evidence pick the register and let the style guide set the limits around it. Matching message to stage has its own post on awareness stages for ad copy.

What is the tell that the register is wrong?

The tell is a line no human would say out loud in that situation. Imagine the buyer saying the hook to a friend over coffee, at the moment they feel the problem. "Say goodbye to the days when your face looked like a pizza" fails at once. Nobody talks like that. "Two years of pizza face and three routines that did nothing" passes, with the same quote.

I run this test on every hook before it goes to a client. It also catches "elevate", "transform" and "unlock", which mark a line as written by marketing. When a hook fails, rewrite the frame around the phrase and leave the phrase alone. The phrase is the part that came from a buyer.

When is the brand register the right answer?

The brand register is the right answer when sounding casual makes the buyer trust you less. High-price purchases, long consideration cycles and regulated categories (finance, health claims, software sold to a procurement team) sit here. Someone spending €4,000 on a hearing aid does not want the seller to sound like a mate on Reddit.

The customer's words still name the real problem here. But the register worth copying may be the most careful one in your corpus: the long review, the forum post from someone who did their homework. Match the buyer closest to buying. In a careful category, that buyer is careful too.

What if the brand guidelines forbid the register the evidence points at?

When brand guidelines forbid the register the evidence points at, change the sentence shape and keep the approved vocabulary. Register lives mostly in structure. Short sentences, first person, a concrete number and an admission read human even with clean words. Long sentences stacked with adjectives read corporate even with slang in them.

A brand that bans "pizza face" can still write: "Two years. Three routines. Nothing worked." No forbidden word, and the register is right. Most guidelines police vocabulary and say nothing about rhythm. I have got more copy past brand teams by shortening sentences than by arguing over words.

Keep the register in the research

A quote doc that lists "pizza face" without noting it came from a resigned two-line reply has already lost half the information. Adscalr's audience research searches five named sources (Reddit, Amazon reviews, App Store, Google Play and custom forums seeded with your and competitors' URLs), tags each quote with two to four phrase markers plus its emotion and language register, and maps it to one of Eugene Schwartz's five awareness stages. A copywriting critic then scores every concept against six direct-response principles, pass/fail, quoting the line that failed. It will not write your brand voice for you. The audience intelligence pillar shows how quotes are sourced and tagged.

Brand voice vs customer language in ads