Stewart Butterfield
Co-founder of Slack and Flickr
Where to hear and read about Stewart Butterfield1 episode or test
Stewart Butterfield co-founded Flickr and Slack and sold Slack to Salesforce, and he is known for a small set of named mental models about product and leadership. The thing most worth learning from him is to make users think less and to create real value, and to treat small acts of care for the user as a lasting edge.
If you can't see almost limitless opportunities to improve, then you shouldn't be designing a product.
Mental models for building products people love ft. Stewart Butterfield
Stewart Butterfield: the story and the lessons
1Stewart Butterfield's path from Flickr to Slack
Butterfield co-founded Flickr and Slack and sold Slack to Salesforce. He rarely gives interviews. The Slack sale was one of the biggest tech acquisitions of its time.
He and Slack CTO Cal Henderson have worked together for 23 years, since Flickr. Every three to six weeks they talk about what to do next.
Slack grew out of the chat tool built inside the Glitch game. The Glitch team used IRC internally and added features only when the irritation or the gain was impossible to ignore. This went on for years without being an explicit project.
That quiet internal tool gave Slack a head start.
He pivoted only after exhausting every realistic idea. Glitch still had about $9 million in the bank and everyone liked the game and the work. Butterfield had run out of non-ridiculous long shots, so he closed it.
Closing Glitch erased a community and cost staff their jobs, some after moving their families. His rule: "Creating the distance so that you can make an intellectual, rational decision about it rather than an emotional decision is essential."
Butterfield says a pivot needs a real reason and distance from emotion. Pivoting a six-month-old app built by three people barely counts, because you are still working out what you are building.
He admits survivorship bias: it worked for him, but not everyone should pivot. He contrasts it with Canva's Melanie Perkins, rejected by about a hundred investors, who believed in the product and was refining the pitch.
After Salesforce he chose family and art over a new company. He left Salesforce two and a half years ago, and his daughter was born three days after his last day. He spends the rest of his time on philanthropy, personal art projects and supporting other artists.
He thinks phones are "kind of destroying the world" and that software helping people use them less would be a big winner. He and Henderson haven't found it.
2Butterfield's rules for comprehension and positioning
Butterfield aims to make users think less, not click less. Eight trivially easy taps beat two taps that each force a hard decision. A confusing choice makes people feel stupid and they associate that feeling with the product.
A teenager tapped Snapchat four to seven times a second for about 6 minutes, because every action was obvious. Making her tap less would have hurt the experience.
He puts 70 to 80% of product design in comprehension. Friction removal matters in registration, authentication and checkout, where intent is already high. For anything unique to a product, first-time visitors barely clear their intent threshold, so they must be told what it is and what to do next.
Taylor Swift ticket buyers will push through a slow site, but Slack visitors would not. First-time Shopify store owners often don't know reports exist.
He shows the two or three common actions and hides the rest. Uber once opened with "Where would you like to go?" plus an "other" entry that held everything else. Group items first, then trim to the top choices.
Gmail splits actions across two long, inconsistent menus, and "mark as unread" is only an unlabeled icon. Comparing options gets costlier as the count rises.
Butterfield tells teams to sell the outcome and create the market. His "We don't sell saddles here" memo went to about eight people before Slack launched, to build alignment early. It argued you must communicate what the product does for people, like horseback riding rather than saddles.
A new idea is nearly impossible to plant, but joining two existing ideas, like "Uber for pets", is easy. "You're not just creating the product, you're creating the market."
He coined the habit of checking for the owner's delusion. Owners think visitors care about their site as much as they do. Restaurant sites showed a slow photo, a Ken Burns effect and music while hiding the phone number in an image.
Visitors are late, distracted and ready to bounce in a split second. The cure is to name the bias and train yourself to look at the page as a regular person.
3How Butterfield thinks about utility curves, taste and discontent
Butterfield judges each feature by where it sits on an S-curve. Value is flat at first, climbs steeply past a quality threshold, then flattens again. Teams should ask whether a feature has had too little investment or has hit diminishing returns.
A half-built feature adds complexity but no value, so users ignore it and the team wrongly decides the idea wasn't worth doing. Rising user standards keep moving the curve.
He calls neglected core flows the place most products fall behind. Most features get built once and improved rarely. Sign-up, account creation and forgot password rarely get love, though every competitor's version keeps improving.
He cites Google Calendar's iOS time zone picker, which lists zones alphabetically so typing "east" returns four Eastern Australia entries first. Small frustrations build resentment and small delights build advocacy.
He treats taste as a trainable edge few people invest in. Like cooking, taste improves with practice, and anyone can get better at fine discernment even if some start with an advantage.
The risk on the other side is obsessing over one detail long past the point where it matters.
Butterfield demands constant dissatisfaction with his own product. In 2014, the launch year, he told MIT Technology Review that Slack was terrible. Staff printed the harsh quote on about 40 sheets and taped it to the wall.
He saw it as a prompt for perpetual improvement. Direct criticism only motivates when the team trusts each other, as in Ray Dalio's story of Michael Jordan wanting every skiing mistake named.
4How Slack tilted its umbrella
Butterfield turned courtesy into Slack's advantage. He and creative director Brandon Velestuk watched Vancouver's Yaletown sidewalk in the rain and guessed who would tilt an umbrella. About one in three did.
He told the story over and over and built it into the new-hire welcome. "Tilting your umbrella is our opportunity." Slack spread as people changed jobs and pushed their new team to adopt it.
Slack replaced mobile passwords with an emailed magic link. Owning the email address already proves identity, so Slack asked only for the email and sent a link that opened the app and signed the user in.
Butterfield says someone else invented it and Slack was the first he knows of to scale it into a standard. Typing a decent password on a phone is a terrible experience.
Slack started new accounts on noisy notifications, then calmed them. New users sent a message, saw no notification and decided the product was broken, so Slack reluctantly defaulted new accounts to every message. After about 10 messages a prompt offered the recommended quieter settings in one click.
People used to AIM, SMS and WhatsApp expected a ping for every message.
A shouty rooster taught people what @everyone costs. Outlook's priority flag was abused until it meant nothing, and @everyone was heading the same way. The rooster warned senders, for example, that the message would notify 147 people in eight time zones.
It was trivial to build, shaped communication culture and still exists.
Do Not Disturb shipped with layered overrides so no one felt blindsided. Admins got weeks of notice and an org default of 8:00 p.m. to 8:00 a.m. Users could override it and admins could reset it.
The default mattered most because most people would never turn it on. On-call engineers receiving ops alerts made the rollout delicate.
5Butterfield on organizational drift and fake work
He expects every manager to want more reports. Under Parkinson's law, headcount tracks career path, pay and authority, so wanting reports is rational. At Slack, 27-year-old product managers wanted to hire at once so they could "do strategy".
In budget season no head of engineering, sales, finance or legal proposes staying flat. Leaders have to counter this deliberately.
Butterfield says leaders must keep known, valuable work in supply. When clear work runs short, people fill the gap with hyper-realistic work-like activities such as previewing decks. Seventeen product marketers create seventeen people's worth of demand for work.
Scolding people is the wrong move; the fault lies with whoever owns priorities. "It's on you to ensure that there is sufficient supply of known valuable work to do." He says he, the board and every exec fall into it too.
He killed analysis whose best gain was smaller than its cost. Slack's thread reply box was pre-filled with an @ for the previous poster, removed by Butterfield, then restored after an A/B test showed 2.17 messages per thread versus 2.14. The test took about a dozen people and thousands of person-hours.
It was unclear that longer threads were even better. The best possible gain was smaller than the cost of finding it, so it was "guaranteed to be a loser".
Butterfield blames complexity, not villains, when things fail. A viral thread described years and many agencies needed to put up one residential stop sign. The book Why Nothing Works describes a growing number of vetoes over any course of action.
People blame villains because they understand the world through human intent. He sees the same blame-hunting inside organizations.
6Why Butterfield leads with generosity
He treats generosity as both character and strategy. He says much of it is simply how he is, thanks to how his parents raised him. Generous acts also signal he will cooperate as the game repeats, so others cooperate too.
Colleagues' examples include $500 from a cash machine before Christmas, 100% paid health insurance, a no-lockup direct listing and extended severance and job help in layoffs.
Butterfield measures success by value created for customers. He had whole companies chant that long-run success is the value created for customers. Anything that feels shady or like taking advantage of a customer should not be done.
You can point out the value you created, but "there's no substitute for actually having created it." Ethical staff also make a better workplace.
Slack's SLA refunded 100 times the money for downtime. At small scale it cost pennies. After the IPO, a many-hour outage meant about $8 million in credits, which cost no cash but wiped out revenue expected next quarter.
Credits were applied automatically, even to customers the outage didn't affect. The terms were changed after the IPO, though he says the generosity still paid off overall.
Mental models for building products people love ft. Stewart Butterfield
Butterfield shares the mental models behind Slack's product and leadership: utility curves, comprehension over friction, tilting your umbrella, hyper-realistic work-like activities and the owner's delusion. The conversation shows how small acts of craft and empathy compounded into a product people evangelized and sold for one of tech's biggest acquisitions.
Butterfield co-founded Flickr and Slack and sold Slack to Salesforce. He rarely gives interviews. The Slack sale was one of the biggest tech acquisitions of its time.
He and Slack CTO Cal Henderson have worked together for 23 years, since Flickr. Every three to six weeks they talk about what to do next.
Slack grew out of the chat tool built inside the Glitch game. The Glitch team used IRC internally and added features only when the irritation or the gain was impossible to ignore. This went on for years without being an explicit project.
That quiet internal tool gave Slack a head start.
He pivoted only after exhausting every realistic idea. Glitch still had about $9 million in the bank and everyone liked the game and the work. Butterfield had run out of non-ridiculous long shots, so he closed it.
Closing Glitch erased a community and cost staff their jobs, some after moving their families. His rule: "Creating the distance so that you can make an intellectual, rational decision about it rather than an emotional decision is essential."
Butterfield says a pivot needs a real reason and distance from emotion. Pivoting a six-month-old app built by three people barely counts, because you are still working out what you are building.
He admits survivorship bias: it worked for him, but not everyone should pivot. He contrasts it with Canva's Melanie Perkins, rejected by about a hundred investors, who believed in the product and was refining the pitch.
After Salesforce he chose family and art over a new company. He left Salesforce two and a half years ago, and his daughter was born three days after his last day. He spends the rest of his time on philanthropy, personal art projects and supporting other artists.
He thinks phones are "kind of destroying the world" and that software helping people use them less would be a big winner. He and Henderson haven't found it.
Butterfield aims to make users think less, not click less. Eight trivially easy taps beat two taps that each force a hard decision. A confusing choice makes people feel stupid and they associate that feeling with the product.
A teenager tapped Snapchat four to seven times a second for about 6 minutes, because every action was obvious. Making her tap less would have hurt the experience.
He puts 70 to 80% of product design in comprehension. Friction removal matters in registration, authentication and checkout, where intent is already high. For anything unique to a product, first-time visitors barely clear their intent threshold, so they must be told what it is and what to do next.
Taylor Swift ticket buyers will push through a slow site, but Slack visitors would not. First-time Shopify store owners often don't know reports exist.
He shows the two or three common actions and hides the rest. Uber once opened with "Where would you like to go?" plus an "other" entry that held everything else. Group items first, then trim to the top choices.
Gmail splits actions across two long, inconsistent menus, and "mark as unread" is only an unlabeled icon. Comparing options gets costlier as the count rises.
Butterfield tells teams to sell the outcome and create the market. His "We don't sell saddles here" memo went to about eight people before Slack launched, to build alignment early. It argued you must communicate what the product does for people, like horseback riding rather than saddles.
A new idea is nearly impossible to plant, but joining two existing ideas, like "Uber for pets", is easy. "You're not just creating the product, you're creating the market."
He coined the habit of checking for the owner's delusion. Owners think visitors care about their site as much as they do. Restaurant sites showed a slow photo, a Ken Burns effect and music while hiding the phone number in an image.
Visitors are late, distracted and ready to bounce in a split second. The cure is to name the bias and train yourself to look at the page as a regular person.
Butterfield judges each feature by where it sits on an S-curve. Value is flat at first, climbs steeply past a quality threshold, then flattens again. Teams should ask whether a feature has had too little investment or has hit diminishing returns.
A half-built feature adds complexity but no value, so users ignore it and the team wrongly decides the idea wasn't worth doing. Rising user standards keep moving the curve.
He calls neglected core flows the place most products fall behind. Most features get built once and improved rarely. Sign-up, account creation and forgot password rarely get love, though every competitor's version keeps improving.
He cites Google Calendar's iOS time zone picker, which lists zones alphabetically so typing "east" returns four Eastern Australia entries first. Small frustrations build resentment and small delights build advocacy.
He treats taste as a trainable edge few people invest in. Like cooking, taste improves with practice, and anyone can get better at fine discernment even if some start with an advantage.
The risk on the other side is obsessing over one detail long past the point where it matters.
Butterfield demands constant dissatisfaction with his own product. In 2014, the launch year, he told MIT Technology Review that Slack was terrible. Staff printed the harsh quote on about 40 sheets and taped it to the wall.
He saw it as a prompt for perpetual improvement. Direct criticism only motivates when the team trusts each other, as in Ray Dalio's story of Michael Jordan wanting every skiing mistake named.
Butterfield turned courtesy into Slack's advantage. He and creative director Brandon Velestuk watched Vancouver's Yaletown sidewalk in the rain and guessed who would tilt an umbrella. About one in three did.
He told the story over and over and built it into the new-hire welcome. "Tilting your umbrella is our opportunity." Slack spread as people changed jobs and pushed their new team to adopt it.
Slack replaced mobile passwords with an emailed magic link. Owning the email address already proves identity, so Slack asked only for the email and sent a link that opened the app and signed the user in.
Butterfield says someone else invented it and Slack was the first he knows of to scale it into a standard. Typing a decent password on a phone is a terrible experience.
Slack started new accounts on noisy notifications, then calmed them. New users sent a message, saw no notification and decided the product was broken, so Slack reluctantly defaulted new accounts to every message. After about 10 messages a prompt offered the recommended quieter settings in one click.
People used to AIM, SMS and WhatsApp expected a ping for every message.
A shouty rooster taught people what @everyone costs. Outlook's priority flag was abused until it meant nothing, and @everyone was heading the same way. The rooster warned senders, for example, that the message would notify 147 people in eight time zones.
It was trivial to build, shaped communication culture and still exists.
Do Not Disturb shipped with layered overrides so no one felt blindsided. Admins got weeks of notice and an org default of 8:00 p.m. to 8:00 a.m. Users could override it and admins could reset it.
The default mattered most because most people would never turn it on. On-call engineers receiving ops alerts made the rollout delicate.
He expects every manager to want more reports. Under Parkinson's law, headcount tracks career path, pay and authority, so wanting reports is rational. At Slack, 27-year-old product managers wanted to hire at once so they could "do strategy".
In budget season no head of engineering, sales, finance or legal proposes staying flat. Leaders have to counter this deliberately.
Butterfield says leaders must keep known, valuable work in supply. When clear work runs short, people fill the gap with hyper-realistic work-like activities such as previewing decks. Seventeen product marketers create seventeen people's worth of demand for work.
Scolding people is the wrong move; the fault lies with whoever owns priorities. "It's on you to ensure that there is sufficient supply of known valuable work to do." He says he, the board and every exec fall into it too.
He killed analysis whose best gain was smaller than its cost. Slack's thread reply box was pre-filled with an @ for the previous poster, removed by Butterfield, then restored after an A/B test showed 2.17 messages per thread versus 2.14. The test took about a dozen people and thousands of person-hours.
It was unclear that longer threads were even better. The best possible gain was smaller than the cost of finding it, so it was "guaranteed to be a loser".
Butterfield blames complexity, not villains, when things fail. A viral thread described years and many agencies needed to put up one residential stop sign. The book Why Nothing Works describes a growing number of vetoes over any course of action.
People blame villains because they understand the world through human intent. He sees the same blame-hunting inside organizations.
He treats generosity as both character and strategy. He says much of it is simply how he is, thanks to how his parents raised him. Generous acts also signal he will cooperate as the game repeats, so others cooperate too.
Colleagues' examples include $500 from a cash machine before Christmas, 100% paid health insurance, a no-lockup direct listing and extended severance and job help in layoffs.
Butterfield measures success by value created for customers. He had whole companies chant that long-run success is the value created for customers. Anything that feels shady or like taking advantage of a customer should not be done.
You can point out the value you created, but "there's no substitute for actually having created it." Ethical staff also make a better workplace.
Slack's SLA refunded 100 times the money for downtime. At small scale it cost pennies. After the IPO, a many-hour outage meant about $8 million in credits, which cost no cash but wiped out revenue expected next quarter.
Credits were applied automatically, even to customers the outage didn't affect. The terms were changed after the IPO, though he says the generosity still paid off overall.
Slack