Jun 21, 2023 · 1h 9m · news
Gina Gotthilf: Lessons Scaling Duolingo from 3-200M Users; How to Master PR and Comms | E1028 · 20VC with Harry Stebbings
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this episode of 20VC, growth expert Gina Gotthilf sits down with Harry Stebbings to share tactical lessons from scaling Duolingo from 3 to 200 million users, alongside a detailed masterclass on mastering PR, storytelling, and early-stage startup team building.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Harry holds 17.3% of the talking time here. How this is scored →
Harry as informed peer 3.8 Guest teaching 4.7 Guest disagreement 1.7 Harry pushing back 2.9
speaking balance: gold is Harry, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 1:01:45 Explicit disagreement with previous host guests
Gina directly rejects advice given by past 20VC guests, forcefully arguing that hiring a growth leader early is a mistake because leaders are too removed to execute scrappy experiments.
Hardest push from Harry ▶ 44:25 Harry's outright refusal of PR misquote resolution framingHarry forcefully refuses Gina's suggestion that journalists can be persuaded to fix misquotes by reaching out kindly, speaking from direct experience that journalists simply do not care.
Biggest teaching moment ▶ 11:31 Explaining why harmful experiment results are inflection pointsGina educates Harry on growth theory, showing why a statistically significant negative result is actually a key inflection point worth doubling down on rather than avoiding.
Harry holds his own ▶ 53:40 Harry breakdown of PR agency economics vs internal hiringHarry demonstrates clear domain mastery by breaking down PR agency retainer costs and contrasting them with the ROI of hiring a hungry junior internal rep chasing journalists directly.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Harry as informed peer | Guest teaching | Guest disagreement | Harry pushing back | Why |
|---|---|---|---|---|---|---|
| Fumbling in Her Early Twenties & Accidental Path to Tumblr | 1 | 2 | 1 | 1 | Gina shares her early career missteps and how she unexpectedly got recruited by Tumblr. Harry acts primarily as a supportive listener, briefly stepping in to ask how Tumblr sourced her while she was planning to farm. | |
| Lessons from Tumblr: Everyone Is Just Figuring It Out | 3 | 5 | 2 | 4 | Gina discusses early growth work and explains how harmful experiment results signal an inflection point worth testing further. Harry challenges the logic of doubling down on failed experiments before Gina clarifies the growth mechanics. | |
| The Failure and Future Success of Duolingo's Badge System | 4 | 5 | 1 | 2 | Gina walks through the failed initial rollout of Duolingo's badge system. Harry demonstrates product intuition by correctly diagnosing why an MVP badge for simply signing up failed to motivate users. | |
| The Importance of Retention and the DAU North Star | 4 | 4 | 1 | 2 | Harry cites Meta CMO Alex Schultz on retention being king to frame a question about North Star metrics. Gina details how Duolingo prioritized DAUs across product and PR. | |
| PR Tactics: Uncovering and Refining Your Story | 3 | 4 | 1 | 4 | Harry interrupts Gina to demand granular tactical detail on how founders should codify their PR story. Gina breaks down how to observe audience reactions to find story hooks. | |
| The One-Liner and Leveraging Authority to Secure Press | 3 | 5 | 2 | 3 | Harry asks why journalists should care about brand stories when everyone thinks theirs is unique. Gina outlines how to construct authority signifiers and artificial scarcity in pitch emails. | |
| PR Pitching: Exclusives, Embargos, and Building Relationships | 4 | 6 | 2 | 3 | Gina details PR strategies around exclusives versus embargos and explains why startup founders can break traditional agency PR rules like multi-pitching within an outlet. | |
| Timing Your Pitch and Navigating Paywalls vs. Breadth | 5 | 4 | 1 | 3 | Harry demonstrates personal PR expertise by sharing tactics on building Twitter rapport with targeted journalists. Gina breaks down when to prioritize paywalled tier-one publications over broad distribution. | |
| Content Control, PR Power Dynamics, and Handling Embargo Breaches | 5 | 6 | 3 | 5 | Harry refuses Gina's claim that friendly outreach can fix journalist misquotes, citing his own experience that reporters don't care. Gina counters by explaining how agency leverage and cultural dynamics alter reporter responsiveness. | |
| PR Metrics and the Psychology of Sharing & Humble Brags | 3 | 5 | 1 | 2 | Gina details how to measure PR efficacy with DAUs and explains the social psychology of crafting personalized share text that acts as a humble brag for users. | |
| PR Firms vs. In-House Teams | 6 | 4 | 2 | 6 | Harry forcefully attacks the agency model, citing pricing ($3k-4k/mo) and arguing that hiring a hungry 23-year-old intern is vastly superior. Gina agrees mostly but pushes back gently regarding non-communicator engineer founders. | |
| PR Pitfalls, Steady-State PR, and Re-defining Media | 4 | 5 | 1 | 1 | Gina redefines modern PR beyond legacy journalists to include anyone commanding a large audience, referencing Harry's podcast audience as a prime example. | |
| The Saturation of Growth Tactics and Pitfalls of Paid Ads | 5 | 5 | 3 | 3 | Harry pushes on the lack of creativity in current growth tactics using a Mr. Beast hypothetical. Gina rejects rigid marketing frameworks, crediting her philosophy background for her intuitive approach. | |
| Building Scrappy Growth Teams and Conducting Effective Postmortems | 3 | 6 | 3 | 2 | Gina directly contradicts previous 20VC guests by arguing against hiring senior growth leaders early in a startup's lifecycle. She also outlines best practices for postmortems. |
Statements from this episode (29)
Gotthilf: Leaders at successful early-stage startups rarely know what they are doing
“One of the things I learned at Tumblr that I think a lot of people in early stage startups that succeed learn is that no one knows what they're doing at most points in time in life and in business.”
Gotthilf: Tumblr failed to pay her for six months over wire issues
“I wasn't paid at Tumblr for six months, because they couldn't figure that out, and I was like, I was living with my parents, like in my childhood bedroom at the time when I was 25, and like working out of like my childhood bed, literally.”
Gotthilf: Duolingo's initial growth team was just her and one engineer
“It was just me and one engineer. That's all they gave me and one backend engineer.”
Gotthilf: Statistically significant failed experiments reveal high-leverage user moments
“If you run an experiment with statistical significance and it harmed your product, that means that you found a moment that really matters to users. And if you found a moment that really matters to users, you can manipulate it in multiple ways for positive and …”
Gotthilf: Educational apps inherently struggle with user retention compared to games
“Games get retention, right? And education apps do not get retention, right?”
Gotthilf: Badges became Duolingo's single most effective growth experiment
“It ended up being the most effective thing that we did and collected collection of experiments where we were, we could impact not just growth numbers, but every single metric across the company, like monetization, et cetera.”
Gotthilf: Early-stage startups cannot easily run statistically significant tests
“It's really hard really early on in early stage startups, which is where I'm focusing my time now, because you don't have enough people for statistical significance.”
Gotthilf: Retention equals product-market fit and determines startup success
“Retention is everything. To me, retention is almost like product market fit. If you have retention, it means that people actually like the thing that you made, and it's actually useful for them. And that determines whether something's going to succeed or is no…”
Gotthilf: Daily Active Users was always Duolingo's sole North Star metric
“The north star metric for Duolingo at all times was daily active users. That's it. DAU.”
Gotthilf: PR was the primary growth lever for Tumblr LatAm and Duolingo
“And it honestly was the most important lever that helped Tumblr grow in Latin America and that helped Duolingo grow.”
Gotthilf: Product builders are often unable to tell their own product's story
“And often when you are the person who built the thing, you're not the person who can actually tell the story about why it's important.”
Gotthilf: Duolingo relied on shared oral knowledge rather than written brand docs
“Yeah, I didn't, with Duolingo, we never wrote it down on a Google Doc. It's just, it was like shared knowledge. I worked very closely with Luis Van An, the CEO and co-founder of Duolingo in the early days, and so I just heard him tell the story over and over, …”
Gotthilf: Duolingo's government partnership in Colombia failed to generate real user adoption
“I worked with the Colombian government, the Mintic, which is the ministry of technology there to get Duolingo into public schools. You would think, wow, Gina, that's amazing. You got Duolingo into public schools. You've got the whole population. No, it actuall…”
Gotthilf: Startups should pitch least important press targets first to iterate
“Make a list of all of this PR sources that you might want to be featured on and start with the least important ones when you're pitching. Because as you pitch, you need to learn what's working and what's not working.”
Gotthilf: Teasing short pitches builds rapport that increases journalist response rates
“If the person has responded to you once, now they have a relationship with you, even if it's a very marginal relationship, but before they had zero relationship to you. So it's very easy to ignore an email that you've never responded to before, but if you've r…”
Gotthilf: Coordinated press embargoes create a surround-sound effect for startups
“Embargo means that you tell journalists they can only publish something at a specific time on a specific date. And if you're able to do that, then you can get all of the new sources to publish something at the same time. And that's magical because you create a…”
Gotthilf: Duolingo rejected paywalled press exclusives to maximize broad reach
“We did not give exclusives, especially to publications that had a paywall, because that would really limit our breadth, and we were going for breadth.”
Gotthilf: Leverage early TechCrunch press coverage to convince other journalists to publish
“If you're able to get something on TechCrunch, even if it wasn't your, you know, core consumer possibility, then very quickly tell other publications that are more, that feel more insecure than TechCrunch, like editors and journalists, and the link to TechCrun…”
Gotthilf: Most non-top journalists copy-paste press releases due to workload pressure
“I would say that most journalists, not at the top publications, but you would be surprised with most journalists. Tend to be fairly lazy. And I don't fault journalists. They have way too many things to do. They're trying to click, you know, get clicks. They ar…”
Gotthilf: Dictating coverage to journalists is like micromanaging software engineers
“You can't have control and tell a journalist, I want you to write this or that, because I actually will rub them in an incredibly the wrong way, much like when you're working with an engineer and you tell them like, here's the solution I want you to build vers…”
Gotthilf: Duolingo measured PR success directly using daily active users
“So for Duolingo, it was DAUs. Like if we were actually able to get new users that became active users, that was our success metric, because then we knew that we were reaching the right audience and that, you know, the story that was told was the story that we …”
Gotthilf: Latitud grew by giving accepted founders pre-written humble-brag copy
“With latitude, one of our biggest growth levers was when we started admitting startup founders, so we work with early stage startup founders in Latin America and we started admitting them, we created this token that said like, You know, I am, I was selected to…”
Gotthilf: Personally wrote 100+ press emails per Duolingo global launch
“When I was doing press for Duolingo and we were doing these global launches, I was personally writing over a hundred emails at a time, like, to get, like, and it had to be within a window of time because this launch was coming up.”
Stebbings: 99% of PR firms are crap
“I get 50 PR firms a week in my inbox at least. They are generally crap, and I actually just 99% dislike PR firms.”
Stebbings: Cheap PR agency fees equal hiring full-time junior talent
“Even if you get a cheap PR fund, they'll still be, at best, three, four grand a month, at best. You can hire a 23, 24 year old who's hungry as hell, chasing journalists on Twitter DM, 24 hours a day for the same amount.”
Gotthilf: PR is reaching anyone with a large audience, not just journalists
“I think when people think about PR, they think about journalists and they think about like big media. But to me, PR is reaching anyone who has a big audience and getting them to share that with their big audience.”
Stebbings: 20VC averages 1.3 million plays per episode
“This podcast gets 1.3 million plays per show”
Gotthilf: Facebook paid ads are a palliative, addictive growth tactic
“I think this one died a death from day one, but people don't realize it is paid marketing and especially like paid ads on Facebook. Unless you're really smart about how you're doing it and you know exactly like, you know, how you're going to scale this over ti…”
Gotthilf: Hiring senior growth leaders too early harms early-stage startups
“So when you're, so I think that's number one, the number two, I would say, especially for early stage is, and it's something I will disagree with some other guests you've had on your podcast is hiring a leader too soon, because when you hire a leader, they mig…”