Fireflies.ai: The Inbox Nobody Is Watching
Why it matters: Every recap is either an invitation or an irritation
HypothesisThe first recap a non-user receives decides which. A useful recap from a meeting they cared about pulls them in; a recap from a meeting they barely attended pushes them away for good.
In 30 seconds
- The company: Fireflies.ai, an AI notetaker that joins meetings, records them and emails a recap. It's used by 20M+ people across 500,000+ organisations and is valued at $1B.
- What everyone measures: sign-ups, meetings recorded, seats sold.
- What they miss: after every meeting, Fireflies emails a recap to people who never signed up, including clients and other companies. That inbox is Fireflies' biggest free acquisition channel, and its biggest source of annoyance.
- My metric: of the people who receive a recap without an account, how many sign up within 30 days, against how many unsubscribe.
The thinking, drawn
3 diagrams
How I'd Test It
- Measure both sides by recipient type: internal colleague, external client, external partner. My guess is clients unsubscribe most, while colleagues of an existing user convert best.
- Test a smarter first recap: for non-users, lead with the one thing they'd care about — their own action items — rather than a full summary.
- Success means more 30-day sign-ups per 1,000 non-user recaps, with unsubscribes flat or lower.
The people a product reaches without their consent are still users. They just haven't decided yet whether they like you.
The full deep dive
For interviewers who want every detail and source.
More on the signal
Average maths from public numbers: 20M users across 500,000 organisations is about 40 users per organisation. Each of those users' meetings sends recaps to people outside the organisation, so the non-user audience is likely several times the user base. Estimate Fireflies doesn't publish recap volume.
What's Public
- Fireflies reports 20M+ users across 500,000+ organisations, including 75% of the Fortune 500. It has been profitable since 2023.
- Growth is product-led: one person signs up free, then their team follows.
- By default, the recap goes to everyone on the invite, including external guests such as clients.
- Free users get unlimited transcription credits if they keep recaps going to everyone. In effect, Fireflies pays free users to distribute it.
- Fireflies' own help centre has an article for non-users asking why they got a recap, and how to stop it.
Why It Could Be Wrong
If most sign-ups come from colleagues hearing about Fireflies at work, not from recaps, the recap inbox matters less than I think. Tagging sign-ups by their first touch would show that quickly.
PM Deep Dive
The growth model: bottom-up, land and expand
Fireflies grows product-led: one person signs up free, records meetings, and their colleagues and clients see the result. Each recap is an exposure event, a free impression on someone who didn't choose to see it. That's what makes it powerful, and what makes it fragile: the same email that recruits one person annoys another.
Metric tree
- North Star: weekly meetings where at least one recap is opened.
- Input metrics: recaps sent, recap open rate, non-user 30-day sign-up rate, users added per organisation (expansion).
- Guardrails: unsubscribe rate, spam complaint rate, and meetings where the notetaker is removed.
The technical guardrail most teams miss: email deliverability
Inbox providers such as Gmail and Outlook track how recipients treat a sender. If enough people mark recaps as spam, the sender's reputation drops and recaps start landing in spam for everyone, including paying users. So spam complaints aren't just a growth metric; they can quietly break the core product.
Trust and consent
Recording rules vary by country and state, and some require everyone on a call to consent. A notetaker that joins meetings by default carries trust risk for the host as well as for Fireflies. That's another reason to treat non-user reactions as a first-class metric.
Experiment design
- Hypothesis: if a non-user's recap leads with their own action items, more will sign up and fewer will unsubscribe, because the email becomes useful to them personally.
- Randomisation unit: the meeting host, so every recap from one host follows the same version.
- Primary metric: 30-day sign-ups per 1,000 non-user recaps.
- Guardrails: unsubscribe and spam complaint rates must not rise.
- Segmentation: read results separately for internal colleagues and external guests, since the right recap for each is likely different.