ElevenLabs: Building an AI Sales Machine & Why We Set a 20x Sales Quota

TL;DR

ElevenLabs CRO Carles Raina explains how AI agents replace traditional sales tools, why a 20x quota filters for elite talent, and how the company scaled to $350M ARR.

Key Points

• 1.AI SDR outbound tools don't work. Response rates on outbound emails have dropped below 0.01% because AI tools treat every lead identically — ElevenLabs instead built internal AI agents for inbound handling, RFP scanning, and CS draft emails that are perceived as human and have already closed deals.
• 2.ElevenLabs targets 50% productivity improvement, not headcount reduction. Carles prefers a smaller, highly compensated team where AI agents close deals and humans still earn full commission on those AI-closed contracts.
• 3.The 20x quota is intentional and aggressive. Two reps hit their full quarterly quota by February; Carles posted a 'Mount Olympus of sales' Slack message, arguing high quotas attract top talent and keep performance culture alive — with a commitment to lower or compensate fairly if the bar is genuinely too high.
• 4.Commission structure starts at 5% and accelerates in tiers. Accelerators kick in at 1.1x, 1.2x, 1.3x, 1.5x and beyond quota; spiffs target specific products per quarter, but commissions are never paid on pilots — only on signed annual or multi-year contracts.
• 5.Customer success must be a revenue-generation function, not a satisfaction role. Carles disagrees with the 'CS is BS, charge for professional services' view (attributed to Snowflake's Chris Degnan), arguing that in an AI world where competitors can be spun up in two days, deep CS relationships and community trust are essential for retention and expansion.
• 6.Opening multiple markets in parallel beats the traditional sequential VC playbook. Carles calls the 'open one market, go deep, then move to the next' approach outdated — with 100 competitors potentially emerging within a month, parallel market bets and portfolio-style experimentation (channels, products, partners, affiliates) is the correct strategy.
• 7.Verticalizing sales teams too early backfired in India. Segmenting too early with too small a team depressed revenues for a full quarter; ElevenLabs reverted to horizontal selling, used name-account lists with portfolio-construction logic, and saw deals close within the first month of the reset.
• 8.Media and entertainment was a failed ICP; AI agents became the breakout bet. Large entertainment studios weren't ready for the technology — ElevenLabs pivoted to media creation platforms, then made an early bet on agentic systems that is now 'fantastically huge' in revenue contribution.
• 9.Brand dramatically shortens enterprise sales cycles — cursor, OpenAI, and Anthropic are the current blue-chip procurement brands. Carles frames the ideal as 'no one gets fired for buying IBM' and acknowledges ElevenLabs' brand was weak in India a year ago versus the UK/US, directly impacting close rates.
• 10.ElevenLabs powers its own competitors' voice infrastructure and is transparent about it. When launching its own agents product, Carles personally called the largest platform customers to warn them of upcoming competition — all founders accepted it, creating a healthier co-opetition ecosystem similar to Nvidia powering everyone while competing with everyone.
• 11.Doubling the sales team from 130 to ~250 this year risks culture dilution. The biggest pitfall is unclear expectation-setting at hire; ElevenLabs runs a public per-rep leaderboard posted weekly via Slack bot showing quota attainment year-to-date, and Carles personally checks calendars — an empty calendar is an immediate red flag.