How ElevenLabs took outbound from 5% to 35% of pipeline (without sending more email)

How ElevenLabs took outbound from 5% to 35% of pipeline (without sending more email)

[ Read the LinkedIn breakdown → ]
[ Get the 12-page guide → ]

Ask most outbound teams how to build more pipeline and you get the same answer: send more. More emails, more domains, more inboxes, more reps hammering the same list.

Jonathan Chemouny did the opposite.

He took over outbound at ElevenLabs in early 2025 when it was 5% of pipeline. Two quarters later it was 35% and climbing, and the team closed that quarter at 210% of quota. Four SDRs, thirteen AEs, roughly $50k ARR a deal, closing in under two months.

The engine behind it is deeply unglamorous. No growth hack, no magic tool. A strict definition, a tight target list, and a sequence built to warm people up rather than pin them down. Here's the part worth stealing.

The number only works because the definition is strict

Every team has a meetings target. Jonathan's counts differently, and that one design choice is what makes the number mean anything.

The target is 20 accepted SQLs per rep, per month. Accepted is the operative word. An SQL only counts when the SDR has qualified it and the AE has accepted it. If the AE marks it closed-lost on arrival, it never existed.

That's two signatures on every opportunity, and it quietly kills the thing that wrecks most SDR targets: booking bodies to hit a number. When the AE is the quality gate, volume with no substance has nowhere to go.

Three more rules hold the line:

• SDRs are comped past the handover, on revenue from the outbound deals they helped unlock, not just the booking. So they don't walk away at the pass.
• The ICP is published, so reps can't quietly widen the list when they're behind. Out-of-ICP accounts can't be counted.
• The rules of engagement live in the data, not in anyone's memory. Anything touched in the last 30 days is filtered out automatically, so two reps never work the same account.

His one-liner for it: a meeting is not the unit of work, an opportunity an AE will own is the unit of work. Fix the definition before you touch anything else. It's the cheapest change on this list and it moves rep behaviour faster than any of the others.

A universal product is a targeting problem

ElevenLabs sells voice AI, which sits under almost any use case. That makes "who do we sell to" genuinely hard, so instead of selling to everyone they picked four lanes: AI agents, media creation tools, media and entertainment, and gaming. Reps prospect inside those lanes, not around them.

Inside each lane, three titles are worth hunting: Head of AI, Innovation Lead, and CX or support leadership. The Head of AI one is the tell. Most teams assume the role doesn't exist yet, so they never search for it. Search it anyway. A brand-new title is the clearest budget signal you'll find in an emerging category.

And the list is built on behaviour, not firmographics. Jonathan's framing here is the sharpest line in the whole thing: a perfect-fit account with no movement is a cold call, but a decent-fit account that just hired a Head of AI has a reason to reply this week. Fit tells you who could buy. Behaviour tells you who might reply now.

The sequence warms the lead, it doesn't chase it

The sequence runs six steps across email, LinkedIn and WhatsApp, built on one idea: each channel makes the next one land warmer.

A few moves that cut against the standard playbook:

• The first email has no CTA. Nothing to accept or decline means nothing to reject. Its only job is to be worth reading.
• The LinkedIn connect goes out with no note. Acceptance rates are higher without one, and around 70% connect.
• The whole thing branches on behaviour, not the calendar. Engage on LinkedIn and a new thread opens. Go quiet and the next touch becomes an audio message or WhatsApp. The branch is built once, then it runs itself.

"We're not trying to sell in the first email," as Jonathan puts it. "We're trying to build a relationship." Unfashionable, and it works.

The metric swap, and the test he was sure he'd win

Two habits most teams have backwards.

First: stop reporting open rate. Tracking opens hurts deliverability, and the number is misleading anyway. Report bounce rate instead. It's the signal inboxes read as "this sender is hitting dead mailboxes," and a high one gets your domain blocked. It's the one number Jonathan is relentless about.

Second: always ship a B version. The team's rule is blunt. If you're not testing, you're not learning, you're just sending what feels right. And the story that earns the rule: he was certain a personalised subject line ("ElevenLabs, let's connect [First Name]") would beat the plain one ("ElevenLabs, let's chat"). It lost. That is the entire argument for testing in one line. You are usually sure which version will win. You are usually wrong.

Five things you can run this week

None of this needs budget. Most of it needs you to change what you measure.

1. Rewrite your SQL definition so it only counts when the AE accepts it, then comp something past the handover.
2. Search the job title you assume doesn't exist yet, and build the list on activity, not fit alone.
3. Strip the CTA out of your first email, and send your next connection request with no note.
4. Stop reporting open rate. Report bounce.
5. Never send one version of anything again.

The full 12-page guide has the rest of the engine: the complete six-step sequence with the intent behind each touch, the five buying signals and how each one becomes a first line, the deliverability benchmarks his team holds themselves to, and the field-first way they actually develop SDRs (it involves planes, not more inboxes).

Grab it below. The LinkedIn breakdown is there too if you'd rather start there.

[ Get the 12-page guide → ]
[ Read the LinkedIn post → ]

Source: Jonathan Chemouny's outbound playbook interview, ElevenLabs, 2025.