ElevenLabs

ElevenLabs’ carousel ad: “Transcrire de l'audio réel de mauvaise qualité”

Running 69 days·Facebook and Instagram·Winner

How it opens

“Transformez n'importe quel enregistrement en texte structuré. Le système de diarisation identifie automatiquement qui parle.”

Angle: Feature

Why it works

It opens on a concrete, relatable difficulty instead of a generic product claim, then backs it with named capabilities that show how the difficulty is solved. The carousel format suits a feature set that can be split across cards. It has also run for 69 days and is still active on 5 ads across six placements, which suggests the advertiser keeps finding it worthwhile, though the data alone doesn't show why.

Who it speaks to: Developers, product teams and content or media teams who need to transcribe large volumes of audio or video, including interviews, podcasts and recorded meetings.

How it’s built

1. Hook: The headline names the hard case of low quality real-world audio.
2. Promise: The post text says any recording becomes structured text.
3. Feature: Diarisation automatically identifies who is speaking.
4. Proof claim: The description calls it the most accurate Speech to Text for bulk use, with no figures given.
5. Capabilities: It lists emotion and sound effect detection and keyword guidance.
6. CTA: The Learn more button sends people to elevenlabs.io.

What to borrow

• Open with the hardest, most specific version of your customer's problem, since it signals you can handle the easy cases too.
• Pair one outcome line with a short list of named features so the claim feels concrete.
• Use a carousel to give each capability its own card instead of cramming them into one block of text.
• If you claim to be the most accurate or best, back it with a figure or example, because this ad leaves that to the landing page.

Its words

Transcrire de l'audio réel de mauvaise qualité

Transformez n'importe quel enregistrement en texte structuré. Le système de diarisation identifie automatiquement qui parle.

La transcription Speech to Text la plus précise pour les usages en masse. Détectez les émotions et effets sonores, et orientez la transcription avec des mots-clés.

Button: Learn more

Where and how long

• First ran 19 Aug 2026; still running.
• Run as 5 ads, 5 of them running now.
• Links to elevenlabs.io.
• Style: Demo, Product shot, Carousel.