How Juicebox Hit $10M ARR in 2 Years and Raised $30M from Sequoia

David Paffenholz killed a viral app, stayed lean with just two co-founders for over a year, and fixed brutal churn by manually reviewing 100 searches a day. The result: 10x growth in a single year and a Sequoia-led Series A.

Key Facts

• ARR at Series A: $10M ARR announced with Series A (David Paffenholz)
• Series A size: $30M led by Sequoia Capital (David Paffenholz)
• 2025 growth rate: Approximately 10x year-over-year (David Paffenholz)
• Top acquisition channel: Over 50% of signups from word of mouth; 500+ free signups per day (David Paffenholz)
• Team size at $10M ARR: 24 people (David Paffenholz)

Why David Killed a Viral App with 50,000 Users

Bad retention made the music app a dead end. Despite viral TikTok traction and 50,000 users, day-30 retention was catastrophic. David had seen elite retention benchmarks at Snapchat and knew the music app couldn't compare — so he shut it down before YC even started.

Most founders would double down on 50,000 users and TikTok virality. David Paffenholz walked away. His time on Snap's international growth team had shown him what genuinely retentive products look like — and his music app wasn't one of them.

The top of funnel looked healthy. New users were downloading the app because it was catchy and had viral momentum. But the underlying habit simply wasn't there. Users wouldn't return after seven days, let alone thirty. The only number growing was new installs — and that masked a product with no real staying power.

Killing the app before YC was an uncomfortable decision. David admits he and his co-founder Ishan were 'really worked up about it' when telling their YC partner. In hindsight, that intellectual honesty became a defining trait of how they built Juicebox.

"Our day thirty user retention was catastrophic. People would not come back, not even after seven days, let alone thirty days. The only thing that was growing for us was the top of the funnel." — David Paffenholz

"I had seen what those metrics looked like at Snap, and they were phenomenal there. And then I saw what they looked like for us." — David Paffenholz

The 90-Second LinkedIn Video That Launched Juicebox

In May 2023, a simple Loom demo posted to LinkedIn generated 1–2 million views and converted zero users into 100 paid customers overnight. The video required no budget, no production team — just a fast-paced product walkthrough filmed in a WeWork.

Juicebox launched PeopleGPT with a ninety-second screen recording on LinkedIn. The name was deliberately chosen to piggyback on ChatGPT familiarity at a time when most people had heard of it but few understood LLMs. The product was live and payable from day one.

The video spread because it spoke directly to what recruiters were already searching for: a way to bring AI into their workflow. David describes it as 'message-market fit' — proof that the audience existed, even before the product fully delivered.

The aftermath was a reality check. Within weeks, those 100 paid users were churning, requesting refunds, and leaving unhappy. David frames that moment honestly: 'We had just come off this high... but then we felt like we couldn't deliver on it.' The launch validated demand. Now the real work began.

"We went from zero users, zero customers, no one had ever paid us anything before, to having like a hundred paid users on pretty cheap, fifty-dollar-a-month subscriptions." — David Paffenholz

"It was literally like a Loom recording... filmed in a WeWork with Loom." — David Paffenholz

• Ninety-second Loom recording, no production budget, filmed in a WeWork.
• 1–2 million LinkedIn views; approximately 7,000 likes.
• Zero to 100 paid users overnight at $50/month.
• Churn hit within weeks — message-market fit ≠ product-market fit.

Fixing Brutal Churn: The Manual Playbook

For six months post-launch, Juicebox sat at roughly $200K ARR with 20–30% monthly churn. The fix was entirely manual: David and Ishan rebuilt a Slack bot that pinged them every time a user ran a search, then manually recreated up to 100 searches per day to find and fix product gaps.

There was no clever growth hack. The path from churny $200K ARR to product-market fit was methodical, unglamorous work. The team tracked every search users ran, recreated each one manually, and logged every failure mode. Because search has an enormous long tail of edge cases, the only way to improve was to go through them one by one.

By January 2024, the product crossed a threshold. Net revenue retention climbed from roughly 70–80% month-on-month to above 100%. The change wasn't driven by more signups — it was driven by existing customers actually staying and expanding. Recruiters started using Juicebox as a primary sourcing tool, not just a curiosity.

The insight David draws from this period is simple: staying lean made fast iteration possible. With just two co-founders, every decision was instant, every fix was immediate, and nothing was wasted on coordination overhead.

"We had an internal Slack bot that pinged us every time a user ran a search. Then we would manually go and recreate that search... at some point, that became a hundred searches a day." — David Paffenholz

"Net revenue retention month on month, I think in the beginning, I'd ballpark it at seventy to eighty percent month on month, which is pretty brutal." — David Paffenholz

Why Juicebox Beats Traditional Recruiting Search

Traditional recruiting search is keyword-dependent: a candidate must have the exact word 'fintech' on their profile to appear in a fintech search. Juicebox uses LLMs to review profiles semantically, surfacing candidates who match the intent of a search even without the exact keywords — delivering results competitors literally cannot find.

The core product insight is that keyword search fails at the edges, and the edges are where competitive advantage lives. A standard LinkedIn Recruiter search for a fintech software engineer requires either the word 'fintech' or a manually compiled list of fintech company names. Both approaches miss candidates who worked in adjacent roles — a payroll engineer at a non-branded company, for example.

Juicebox runs an LLM against every profile at scale, making judgment calls a human recruiter would make but that no traditional search system can replicate. As David explains: 'Because we can just brute force an LLM to do this at the scale of hundreds of thousands of profiles, we can suddenly use human level judgment against profiles at a scale that was just not possible previously.'

In a zero-sum hiring market, that edge compounds fast. If your recruiting team finds candidates a competitor's team never sees, you win those hires by default.

"Recruiting is so competitive. It's a zero sum game. Only one company is going to hire the right person. And so if you have an edge or a way of finding additional profiles that a different recruiting team does not, that means you're going to be winning those candidates against them because they don't even know they exist." — David Paffenholz

"Our goal was always to do the work for the user rather than provide a platform for the user to do the work." — David Paffenholz

Scaling to $10M ARR: PLG, Lean Team, and Founder-Led Sales

Juicebox reached $10M ARR with 24 people, no outbound team until 2025, and over 50% of signups driven by word of mouth. The go-to-market engine was PLG self-serve plus founder-led demos for larger accounts — David personally ran every sales call until roughly $1.5M ARR.

The growth model was intentionally simple. Free signups, fast time-to-value, and a one-week trial process for larger teams. Self-serve drove the majority of revenue. David ran every demo himself until the team hired its first account executive in January 2025. By mid-2025, there were six AEs replicating the same process.

A landmark signal came in mid-2024 when a major AI lab — competing intensely for engineering talent — signed on and then expanded. That enterprise validation, in one of the most demanding hiring markets, confirmed the product had crossed the threshold.

The team stayed deliberately small. At $700K ARR, Juicebox had three people. At $10M ARR, it had twenty-four. David raised $5M in seed funding in fall 2024 and spent almost none of it before closing the $30M Series A — burning only around $750K in between rounds.

"Almost all of our growth is purely inbound. Word of mouth is by far the biggest channel. Over half of our signups come from word of mouth. We got over 500 free user signups a day." — David Paffenholz

"Keeping the team super lean and then being honest with yourself... one of our core values is intellectual honesty." — David Paffenholz

• 500+ free signups per day at peak PLG growth.
• First AE hired January 2025; six AEs by mid-2025.
• Seed round: $5M raised; ~$750K spent before Series A.
• 24 employees at $10M ARR — very high revenue per employee.
• Major AI lab signed and expanded in mid-2024 as PMF signal.

Traditional Recruiting Search vs. Juicebox AI Search