▶️ Listen to this episode
[ Apple Podcasts ] [ YouTube ]
Two friends annoyed by flat, lifeless movie dubbing built a company now worth $11 billion. I’ve run marketing teams for over fifteen years, and the part of this story that grabs me isn’t the voice technology. It’s how methodically ElevenLabs engineered virality instead of waiting for it to happen on its own.
Most founders treat marketing as a function you bolt on after the product works. ElevenLabs treated it as part of the product from day one.
From Annoyance to Ambition
Mati Staniszewski and Piotr Dąbkowski didn’t set out to build a marketing machine. They wanted machines to sound human.
That’s a familiar founder origin story. What’s less common is what happened once the product started working.
Both founders came out of Palantir and Google, places where shipping fast and reading usage data closely is just how you operate. That background shows up in how the company reacted to its first real signal.
Early users weren’t using the beta text-to-speech tool the way it was designed. They were pasting in entire manuscripts and building full audiobooks, chunk by chunk.
I’ve seen this pattern in my own campaigns. When customers misuse your product in a consistent, repeatable way, that’s not noise. That’s your next product line announcing itself.
ElevenLabs listened. It stopped selling a voice API and started selling an audio workflow — dubbing, reading, voice design, conversational agents.
Workflow Ownership Is a Marketing Decision Too
Choosing to own a workflow instead of a single feature is itself a distribution strategy. Most operators miss this.
A point solution gets compared on price. A workflow gets compared on outcomes, and outcomes are much harder for a competitor to undercut.
Foundation models commoditize fast. Everyone knows this by now. What ElevenLabs understood early is that the application layer wrapped around the model is where the durable business actually lives.
There’s a marketing upside to this too. You can’t build a viral moment around a raw API endpoint. You can build one around a creator publishing an audiobook nobody can tell was AI-generated.
How ElevenLabs Engineered Virality, Step by Step
A great product doesn’t guarantee a growth curve this steep on its own. I’ve managed budgets large enough to know paid acquisition alone never produces this.
ElevenLabs engineered virality through three deliberate levers, not luck.
Shareable Demos, Not Press Releases
When the team built voices that could laugh or carry real emotion, they didn’t write a whitepaper. They sent samples to creators and let the reaction do the talking.
The hackathon project Gibberlink is the clearest example. Two AI agents recognizing each other as machines and switching to a sound-based protocol is strange enough to share without anyone asking you to.
Most companies optimize for what they want to announce. ElevenLabs optimized for what people would actually want to send a friend.
SEO Built for Intent, Not Volume
Fighting for “AI voice generator” alone is expensive and crowded. ElevenLabs went long-tail instead — specific accents, sound effects, languages, and direct “[competitor] alternative” pages.
That’s unglamorous work. It’s also the kind of work that compounds quietly while rivals keep bidding on the obvious keyword.
Turning Creators Into a Sales Force
Instead of buying ad placements, ElevenLabs seeded the product with YouTubers and writers who had real production problems to solve.
A creator who genuinely saves time, attaches an affiliate link, and posts a tutorial is doing three jobs at once: building trust, demonstrating utility, and generating a backlink. No media buyer delivers all three together.
The Trade-off Nobody Likes to Admit
Creators kept asking for a simple speed and pacing slider. The research team resisted, insisting the model should infer pacing on its own.
Growth stalled over what turned out to be a small UX gap.
Leadership made the right call: give research a deadline, and if it can’t solve the problem inside that window, let the product team ship the manual workaround.
I’ve had this exact argument inside my own teams. Engineers want the elegant solution. Customers are waiting right now. The lesson holds in any industry — ship the workaround today, replace it with the elegant version later.
Defusing the Adversary Instead of Fighting It
Generative voice technology gives professional voice actors every reason to see ElevenLabs as a threat. That’s a brand risk most companies underestimate until it turns into a regulatory one.
ElevenLabs built a Voice Marketplace instead, letting actors license their own cloned voices and earn royalties whenever someone uses them.
You can’t out-message a community that believes you’re taking its livelihood. You can only change the economics underneath the relationship.
Running Two Clocks at Once
Revenue splits roughly evenly between self-serve creators and enterprise clients like Deutsche Telekom and Revolut. Those two audiences run on entirely different timelines.
A creator judges a new voice model within hours. An enterprise buyer takes six to twelve months of security review before signing anything.
ElevenLabs organized into small, autonomous pods to match each clock — consumer teams moving like startups, enterprise teams operating like embedded consultants.
Most companies pick one motion and force-fit the other audience into it. That’s usually where growth caps out.
What This Means for the Rest of Us
Every marketing leader watching this should take one thing seriously: virality isn’t an accident you wait for.
ElevenLabs didn’t get lucky three times in a row. It built a company where engineered virality was the default outcome of how the product, the SEO architecture, and the partner network were structured — not a campaign bolted on at the end of a quarter.
That’s the real gap between a company that hopes for a viral moment and one that’s built to produce them on schedule.