Snap
The company behind Snapchat, a camera-first messaging app popular with teens.
Snap's 2018 redesign flattened DAU, then a focus on Android performance for existing users drove years of growth. Snapchat Plus reached 25 million subscribers and Instagram copied it, while Sellis calls missing camera ad monetization a top failure. Lesson: lift engagement of core users first, start subscriptions early on messaging utilities, and expect features to be cloned.
- 1Redesign flattens DAU2018 redesign stalled daily users
- 2Speed up AndroidFaster app for people already using it
- 3Lift DAU/MAUOne daily user ≈ 30 monthly users
- 4Growth restartsA run lasting multiple years
From Sellis, Snap's first PM, on a podcast
What was said about Snap
1Growing through close friends and existing users
Snapchat grew by connecting people to their closest friends, not everyone. The old belief was that the biggest network is the stickiest. Snapchat instead linked people to a best friend, partner or spouse, where Spiegel says most of the value sits.
Spiegel credits the young app store with helping, but says this different way of growing did the real work. "It wasn't about having the most friends. It was about having your close friends."
After the 2018 redesign, Android performance for existing users restarted growth. When the 2018 redesign flattened DAU, Snap focused on making the app faster for people already using it, and that started a growth run that lasted multiple years.
Sellis argues one daily user equals about 30 monthly users, so moving DAU/MAU a few basis points beats acquiring new MAU and hoping they retain. At Discord the same approach, focusing on friends who play multiplayer games, moved users from about 10 active days a month to 12 or 13.
Define core product value as a phrase whose words become retention metrics. Snap's core product value was "the fastest way to share a moment with your best friends," and each word was turned into a metric tied causally to retention.
Sellis gave this onboarding talk to every new Snap hire for seven years, and the phrase was used to judge roadmap items and onboard data scientists. Discord's version was talking and hanging out with friends before, during and after games.
2Why Snap builds ecosystems and hardware instead of relying on software
- 1Camera off the phoneFirst step away from phones
- 2Add depth cameraSecond camera for depth
- 3Add a displayContent shown in the glasses
- 4Developer OSReleased for developers in 2024
- 5Consumer Specs launchShared AR anchored in the world
From Evan Spiegel interview
Software features get cloned, so Snap invested in ecosystems that cannot be. Snap learned 15 years ago that features get copied quickly even with patents, so it built creator relationships and an AR platform where developers have made millions of lenses.
Spiegel lists Stories, swipe navigation, the camera as the main screen, face swapping and AR glasses among Snap's firsts. He says network effects are necessary but not enough, and that being copied beats building things nobody wants to copy.
Specs is the end of a step-by-step hardware roadmap toward shared AR. Snap went from a camera off the phone, to a second camera for depth, to a display, then released an operating system for developers in 2024, with the consumer Specs launch this year.
Spiegel wants content anchored in the world for hanging out with friends, not phone notifications on your face, and calls phone AR "like a keyhole." Sellis describes reaching glasses by backcasting: imagining the best way to share a moment with no tie to today's tech.
Distribution, not product, now separates consumer winners. Spiegel says the only recent consumer hits, TikTok and Threads, solved distribution: TikTok spent billions subsidizing viewers and creators, and Threads borrowed Meta's reach.
People download far fewer apps than when Snapchat launched. He expects new platforms like glasses to reopen distribution, with most upside for those who already have it.
3Snapchat Plus and the ads problem
Ads struggle in a camera-first messaging app with a young audience. Snap's ads were held back by under-17 tracking rules that make ROI harder to measure, an app that opens to a camera with no natural ad slot, and a messaging core.
Sellis calls his failure to drive camera monetization one of his biggest at Snap. "It is just not a feed-based product." He also notes a billion monthly users did not produce a truly massive business.
Subscriptions on messaging apps build slowly, so launch them early. Snapchat Plus passed 25 million subscribers and a $1 billion revenue run rate, after which Instagram launched its own Plus.
Sellis says penetration grows slowly and roughly linearly because users must get used to daily use first, and wonders if launching Plus four years earlier would have eased pressure on ads. "Time is your enemy with subscription products."
Snap still has to prove a profitable core before Specs can win. With over 1 billion monthly users and over $6 billion a year in revenue, Snap is not yet net income profitable, and Spiegel calls this year a crucible moment.
4The small flat design team
Run a tiny flat innovation team beside the large structured company. Snap serves almost a billion people with a design team that swings between nine and 12 people with no fancy titles, and leaders keep it in dialogue with the larger organization.
Spiegel cites Loonshots: scale needs hierarchy and rigor, which makes people chase promotions and avoid risk. "The dialogue between those two teams is where a lot of that innovation happens."
Review hundreds of unfiltered ideas with the CEO every week. Spiegel spends a couple of hours weekly with designers looking at hundreds of ideas, and anyone can add anything to the list with no gate.
New designers present on day one so preciousness fades early. The habit comes from art school's constant output and brutal critique: "If you want to have a good idea, you have to have lots of ideas."
Listen deeply to users, but don't build what they literally ask for. Users asked for a send-all button; Snap built Stories instead, with no likes, 24-hour expiry and chronological order.
Spiegel rejects surveys and suggests one or two hours with a person on how they use technology and feel about their products.
Make design approval a deliberate bottleneck to keep the app cohesive. Nothing ships at Snap without design approval, which slows shipping and annoys people but avoids an app that feels built by separate teams owning pages.
This holds even as AI lets designers ship code.
Hire designers on portfolio range, then rotate them across products. Snap ignores pedigree, hires most designers straight from school, and looks for range, since a uniform style signals an artist rather than a designer.
Candidates pick one piece and tell why they made it and what they learned; Spiegel interviews every designer. Designers rotate through areas like chat, the map and the AR platform so they don't get stuck or bored.
Snap had designers do PM work and delayed hiring PMs. Snap reportedly had about 200 employees before its first PM, because Spiegel didn't want designers reduced to producing visuals for a PM's direction.
PMs became important later for coordinating legal, trust and safety, and data science. "My view is not that we don't need PMs. My view is that designers should do that work."
5How Snap uses AI agents
Guardrails let designers ship AI-written code at billion-user scale. Automated code review at Snap has caught close to 10,000 bugs, and a shake-to-report system has agents debug issues and suggest fixes.
Designers can ship code with AI but are not required to.
Map AI work to the jobs to be done for users and advertisers. Snap listed jobs such as getting people to download the app, add close friends and use lenses, or onboarding advertisers and configuring campaigns, then chose where to deploy agents.
The list also tracks progress against business outcomes for each job.
Chain a full launch workflow into one agent. Snap's go-to-market agent takes a product idea and writes the spec, finds who must sign off, runs legal and trust and safety risk analysis, and writes launch materials like the blog.
Visual creation is being added, and Snap uses Claude for much of this work. Spiegel also built a Glean agent that combs dashboards, documents and leaders' weekly updates to flag what he might miss.
Spiegel expects human comfort, not technology, to set AI's pace. He thinks tech leaders wrongly assume people will adopt new tools blindly and expects significant societal pushback on AI.
Evidence is limited to a passing mention of unfavorable polls. "Humanity dictates how technology is adopted."
6Leadership lessons from Sellis and Spiegel
A shared mission plus named owners replaces most collaboration. Sellis, Snap's first PM for seven years, wants a single directly responsible owner per problem and PMs who decide as founders would have.
He argues collaboration makes a team as slow as its slowest node, and compares the model to a terrorist organization: an ideology everyone understands and a clear structure of who decides.
Give strong performers more load and make room for spiky talent. Sellis hands consistent performers more decisions until he sees bad ones, and says a team that cannot absorb spiky talent like Nikita Bier is failing.
He admits he has no good way to help people improve, and the approach skips the cost of burnout. He also argues the median PM is weak because the best leave and pay keeps weaker ones in the role, without data.
Spiegel learned communication by doing all-hands until he enjoyed them. Early on he avoided all-hands and Q&A and preferred email, until a board member told him it was his job.
He now frames the CEO role as explaining things so people move in the same direction. "Just doing it is really the only way."
90 minutes of unfiltered product advice from Snap and Discord’s product chief | Peter Sellis
Sellis was its first PM for seven years. After the 2018 redesign flattened DAU, a focus on Android performance for existing users drove years of growth, and its core product value was the fastest way to share a moment with your best friends. Ads were held back by a young audience, a camera-first screen and a messaging core, and he calls missing camera monetization one of his biggest failures.
After the 2018 redesign, Android performance for existing users restarted growth. When the 2018 redesign flattened DAU, Snap focused on making the app faster for people already using it, and that started a growth run that lasted multiple years.
Sellis argues one daily user equals about 30 monthly users, so moving DAU/MAU a few basis points beats acquiring new MAU and hoping they retain. At Discord the same approach, focusing on friends who play multiplayer games, moved users from about 10 active days a month to 12 or 13.
Define core product value as a phrase whose words become retention metrics. Snap's core product value was "the fastest way to share a moment with your best friends," and each word was turned into a metric tied causally to retention.
Sellis gave this onboarding talk to every new Snap hire for seven years, and the phrase was used to judge roadmap items and onboard data scientists. Discord's version was talking and hanging out with friends before, during and after games.
Specs is the end of a step-by-step hardware roadmap toward shared AR. Snap went from a camera off the phone, to a second camera for depth, to a display, then released an operating system for developers in 2024, with the consumer Specs launch this year.
Spiegel wants content anchored in the world for hanging out with friends, not phone notifications on your face, and calls phone AR "like a keyhole." Sellis describes reaching glasses by backcasting: imagining the best way to share a moment with no tie to today's tech.
Ads struggle in a camera-first messaging app with a young audience. Snap's ads were held back by under-17 tracking rules that make ROI harder to measure, an app that opens to a camera with no natural ad slot, and a messaging core.
Sellis calls his failure to drive camera monetization one of his biggest at Snap. "It is just not a feed-based product." He also notes a billion monthly users did not produce a truly massive business.
Subscriptions on messaging apps build slowly, so launch them early. Snapchat Plus passed 25 million subscribers and a $1 billion revenue run rate, after which Instagram launched its own Plus.
Sellis says penetration grows slowly and roughly linearly because users must get used to daily use first, and wonders if launching Plus four years earlier would have eased pressure on ads. "Time is your enemy with subscription products."
A shared mission plus named owners replaces most collaboration. Sellis, Snap's first PM for seven years, wants a single directly responsible owner per problem and PMs who decide as founders would have.
He argues collaboration makes a team as slow as its slowest node, and compares the model to a terrorist organization: an ideology everyone understands and a clear structure of who decides.
Give strong performers more load and make room for spiky talent. Sellis hands consistent performers more decisions until he sees bad ones, and says a team that cannot absorb spiky talent like Nikita Bier is failing.
He admits he has no good way to help people improve, and the approach skips the cost of burnout. He also argues the median PM is weak because the best leave and pay keeps weaker ones in the role, without data.
How to win when software is not a moat | Evan Spiegel (Snapchat CEO)
Runs a public company serving almost a billion people alongside a flat 9 to 12 person design team that reviews hundreds of ideas weekly. It is not yet net income profitable, calls this year a crucible moment, and uses AI agents and guardrails such as automated code review that caught close to 10,000 bugs.
Snapchat grew by connecting people to their closest friends, not everyone. The old belief was that the biggest network is the stickiest. Snapchat instead linked people to a best friend, partner or spouse, where Spiegel says most of the value sits.
Spiegel credits the young app store with helping, but says this different way of growing did the real work. "It wasn't about having the most friends. It was about having your close friends."
Software features get cloned, so Snap invested in ecosystems that cannot be. Snap learned 15 years ago that features get copied quickly even with patents, so it built creator relationships and an AR platform where developers have made millions of lenses.
Spiegel lists Stories, swipe navigation, the camera as the main screen, face swapping and AR glasses among Snap's firsts. He says network effects are necessary but not enough, and that being copied beats building things nobody wants to copy.
Specs is the end of a step-by-step hardware roadmap toward shared AR. Snap went from a camera off the phone, to a second camera for depth, to a display, then released an operating system for developers in 2024, with the consumer Specs launch this year.
Spiegel wants content anchored in the world for hanging out with friends, not phone notifications on your face, and calls phone AR "like a keyhole." Sellis describes reaching glasses by backcasting: imagining the best way to share a moment with no tie to today's tech.
Distribution, not product, now separates consumer winners. Spiegel says the only recent consumer hits, TikTok and Threads, solved distribution: TikTok spent billions subsidizing viewers and creators, and Threads borrowed Meta's reach.
People download far fewer apps than when Snapchat launched. He expects new platforms like glasses to reopen distribution, with most upside for those who already have it.
Subscriptions on messaging apps build slowly, so launch them early. Snapchat Plus passed 25 million subscribers and a $1 billion revenue run rate, after which Instagram launched its own Plus.
Sellis says penetration grows slowly and roughly linearly because users must get used to daily use first, and wonders if launching Plus four years earlier would have eased pressure on ads. "Time is your enemy with subscription products."
Snap still has to prove a profitable core before Specs can win. With over 1 billion monthly users and over $6 billion a year in revenue, Snap is not yet net income profitable, and Spiegel calls this year a crucible moment.
Run a tiny flat innovation team beside the large structured company. Snap serves almost a billion people with a design team that swings between nine and 12 people with no fancy titles, and leaders keep it in dialogue with the larger organization.
Spiegel cites Loonshots: scale needs hierarchy and rigor, which makes people chase promotions and avoid risk. "The dialogue between those two teams is where a lot of that innovation happens."
Review hundreds of unfiltered ideas with the CEO every week. Spiegel spends a couple of hours weekly with designers looking at hundreds of ideas, and anyone can add anything to the list with no gate.
New designers present on day one so preciousness fades early. The habit comes from art school's constant output and brutal critique: "If you want to have a good idea, you have to have lots of ideas."
Listen deeply to users, but don't build what they literally ask for. Users asked for a send-all button; Snap built Stories instead, with no likes, 24-hour expiry and chronological order.
Spiegel rejects surveys and suggests one or two hours with a person on how they use technology and feel about their products.
Make design approval a deliberate bottleneck to keep the app cohesive. Nothing ships at Snap without design approval, which slows shipping and annoys people but avoids an app that feels built by separate teams owning pages.
This holds even as AI lets designers ship code.
Hire designers on portfolio range, then rotate them across products. Snap ignores pedigree, hires most designers straight from school, and looks for range, since a uniform style signals an artist rather than a designer.
Candidates pick one piece and tell why they made it and what they learned; Spiegel interviews every designer. Designers rotate through areas like chat, the map and the AR platform so they don't get stuck or bored.
Snap had designers do PM work and delayed hiring PMs. Snap reportedly had about 200 employees before its first PM, because Spiegel didn't want designers reduced to producing visuals for a PM's direction.
PMs became important later for coordinating legal, trust and safety, and data science. "My view is not that we don't need PMs. My view is that designers should do that work."
Guardrails let designers ship AI-written code at billion-user scale. Automated code review at Snap has caught close to 10,000 bugs, and a shake-to-report system has agents debug issues and suggest fixes.
Designers can ship code with AI but are not required to.
Map AI work to the jobs to be done for users and advertisers. Snap listed jobs such as getting people to download the app, add close friends and use lenses, or onboarding advertisers and configuring campaigns, then chose where to deploy agents.
The list also tracks progress against business outcomes for each job.
Chain a full launch workflow into one agent. Snap's go-to-market agent takes a product idea and writes the spec, finds who must sign off, runs legal and trust and safety risk analysis, and writes launch materials like the blog.
Visual creation is being added, and Snap uses Claude for much of this work. Spiegel also built a Glean agent that combs dashboards, documents and leaders' weekly updates to flag what he might miss.
Spiegel expects human comfort, not technology, to set AI's pace. He thinks tech leaders wrongly assume people will adopt new tools blindly and expects significant societal pushback on AI.
Evidence is limited to a passing mention of unfavorable polls. "Humanity dictates how technology is adopted."
Spiegel learned communication by doing all-hands until he enjoyed them. Early on he avoided all-hands and Q&A and preferred email, until a board member told him it was his job.
He now frames the CEO role as explaining things so people move in the same direction. "Just doing it is really the only way."
Evan Spiegel, Snapchat: Building a Multi-Billion Dollar Company
Snap built a kind, uncompromising culture, a small design team reviewing hundreds of ideas weekly, and a 12-year hardware bet funded by Snapchat's cash flow. It is shifting ads from large brands to small and medium advertisers and expects to operate completely differently within 18 months thanks to AI.
How to win in the AI era: Ship weekly, embrace technical debt, ruthlessly cut scope, and more
Ran product with a 10 to 12 person design team and no PMs until monetization, even at 5,000 to 6,000 employees. It prototyped inside the live app, tested in Australia and high schools, and its CEO approved every UI change. Its private, safe sharing mission ruled out public short-form video.
Billion dollar failures, and billion dollar success | Tom Conrad (Quibi, Pandora, Pets.com, Zero)
Tom spent two years as VP of product executing Evan Spiegel's vision. Strong investor backing let Evan buy game-changing technology and build speculative features that a normal roadmap would prune, and many bets paid off.
Mastering paid growth | Jonathan Becker (Thrive Digital)
Thrive won Snap after answering a confusing RFP with a bespoke deck telling executives what they needed, then asking them to hire only if they agreed. Snap called about two hours later and the relationship lasted a couple of years.
More in Social & Dating
Social & DatingA social network built around connecting people and sharing.
58 sources · 128 tips
Social & DatingA photo and video sharing social app built around creators, Stories, Reels and feeds.
38 sources · 79 tips
Social & DatingTikTok
A short-form video app where creators post content and can earn through a revenue-sharing program.
35 sources · 75 tips
Social & DatingA professional networking platform for jobs, hiring and career content.
15 sources · 55 tips
Social & DatingTinder
A dating app with tiered subscriptions plus boosts, Super Likes and other in-app purchases.
11 sources · 47 tips
Social & DatingDiscord
A chat and voice platform for gaming friends and communities.
10 sources · 29 tips
Social & DatingSnapchat
A camera-first messaging and social app built around close friends, Stories and AR lenses.
10 sources · 23 tips
Social & DatingA social network for short public posts and real-time conversation.
11 sources · 29 tips