John Zeratsky
Where to hear and read about John Zeratsky1 episode or test
John Zeratsky is a co-creator of the Design Sprint, and with Jake Knapp and Eli Blee-Goldman he built the Foundation Sprint at Character Capital. The thing most worth learning from him is to spend about 10 hours up front settling who the customer is and how the product is radically different, then test that hypothesis weekly before building.
Every new product has at its core a hypothesis. There is a founding hypothesis. It's just usually not explicit.
Rapidly test and validate any startup idea with the 2-day Foundation Sprint
John Zeratsky: the story and the lessons
1How John Zeratsky came to the Foundation Sprint
Zeratsky co-created the Design Sprint with Jake Knapp and has run it widely. He and Knapp have worked with teams at Google, Microsoft, YouTube, Slack and Uber. The Foundation Sprint appears in their book Click, and its excerpt is among the most popular posts in Lenny's newsletter.
The framework draws on the Design Sprint and work with over 300 teams.
A one-week prototype that rescued a stalled Google project inspired the Design Sprint. Knapp's side project had gone nowhere for three years and its Stockholm office faced closure. A pitch, a PRD and a design could not win over Larry, Sergey and Eric, so the team built a prototype for fellow Googlers, and it became Google Meet.
The aim was a prototype so real that people react as if it were real and want it.
About 200 teams at Google Ventures showed the Design Sprint was missing a first step. Over five years the week was refined into map, sketch, decide, prototype and test. Established companies already knew their positioning, but pre-product startups did not.
The same questions kept returning: what problem, who is the ideal customer, how are we different, who are the real competitors.
Zeratsky, Knapp and Eli Blee-Goldman built the Foundation Sprint at Character Capital. They developed it in 2021 and 2022. Character Capital runs one right after each investment, and now runs more Foundation Sprints than design sprints.
Founders who run it say it compresses three to four months of work into three to four weeks.
2How Zeratsky's Foundation Sprint works
Zeratsky has the core team clear its calendar for about 10 hours. Co-founders, or in a larger company the leads of product, engineering, design and marketing, follow a scripted sequence in two blocks of four to six hours. Making key decisions together gives everyone one picture of the project instead of three.
It can take eight hours if the team is fast and twelve if it is larger.
Teams write silently, vote, and let one decider lock each answer. Everyone writes alone, the team votes, then a designated decider, usually the CEO, makes the call and the group moves on. Silence and structure get through many decisions quickly and save energy for the ones that matter.
When teams look up from writing, they see answers they did not expect, which is where early value shows.
The basics cover customer, problem, competition and advantages, answered concretely and fast. Competition includes workarounds like art fairs, not just startups, because if a problem is worth solving people already cope with it somehow. Advantages come from the team's insight, motivation and capabilities.
Teams are often surprised that their answers matched only about two of six items, and that the problem was less clear than they thought.
Zeratsky has teams list several approaches, pick one and name a backup. Latchet weighed an app, a newsletter platform, a Shopify plugin and a full stack build, and chose the social sales app with the full stack as backup. A backup makes failure less scary and pivoting faster.
Even a team locked on one path is asked what else could work and whether a dismissed option deserves another look.
Magic lenses turn a vague debate over approaches into one picture. Each approach is plotted on 2x2 charts for customer, pragmatic, growth, money and differentiation, plus custom lenses such as founder conviction. If one option sits top right on most lenses the choice is easy.
If nothing wins every lens, that is reassuring because no perfect choice exists and the team only has to pick the lens that matters most. About four hours of this replaces weeks or months of building and learning.
The sprint ends in one explicit founding hypothesis sentence. The form is: if we solve this problem for this customer with this approach, they will choose it over the competitors because of two differentiators. Latchet's bet was that artisans would choose a cooperative, easy social sales app over Shopify and Etsy.
Until the hypothesis is explicit, different people hold different hidden versions and nobody can test its variables.
Zeratsky holds back ideation so the team does not lock onto its first idea. The longer a team waits to sketch a solution, the likelier it lands on one that works. Bob Baxley calls the first drawing the "primal mark", since every later decision responds to it.
3Zeratsky on differentiation
Zeratsky treats differentiation as the main event because new products get ignored. People guard their attention, so a product needs a promise clear and strong enough to be tried. The promise is the customer's, not the technology or market story told to investors.
Teams that skip it end up painting a message onto the product at the end, when nobody built the thing known to matter from the start.
He wants the product alone in the top right and rivals in Loserville. The team scores itself and each competitor on a few scales, picks the strongest two, and builds a 2x2. Both must hold: the team can deliver it and customers care.
Latchet chose helps you grow and cooperative against Etsy, Shopify and art fairs. Mellow chose mobile first and works out of the box.
Teams warm up on seven classic scales, then write custom differentiators. The classics are fast to slow, smart to not so smart, easy to hard to use, free to expensive, focused to one size fits all, simple to complicated, and integrated to siloed. Custom ones offer customers a new lens on the world, such as networked versus siloed.
Mellow wrote about 25 custom differentiators, such as purposeful, magical and human, before choosing two. Honest scoring dropped the ones it could not deliver.
Zeratsky says to score honestly and accept you will not win everywhere. Driving home one or two things customers care about, where you are radically better, improves the odds of product market fit. Mellow was not best on every scale, which is normal.
Scoring across the whole range also shows where you will win and gives comfort that trailing on some scales is fine.
He advises against leading with price, except for AI cost cuts. Price is rarely the most important differentiator, and big rivals can run loss leaders. It can work for AI products that solve problems software could not touch before, such as BindWell designing precision pesticides without a large team of R&D chemists.
Because AI output may be a bit worse than a human process, the rule of thumb mentioned is about 10x cheaper.
Differentiators become principles that make a one-page mini manifesto. Latchet's principle is that sellers should help each other and earn more. Google's "fast is better than slow" was used in meetings to pick the approach that rendered 12 milliseconds faster.
The manifesto describes how the team believes it will win and is the decision guide, while the basics page holds the core facts. Day one reaches it in about four hours.
The biggest hits he has seen from the inside made one promise that set them apart. Gmail beat Hotmail and Yahoo with massive storage and great search. Slack promised fun and better teamwork, ChatGPT gave a direct answer with zero clicks, and Reclaim used AI to prioritize your calendar and was acquired by Dropbox last year.
Those teams had no Foundation Sprint. Some had the wrong promise and changed it after experiments, but they stayed keyed in on differentiation.
4Testing the hypothesis and using AI, according to Zeratsky
Weekly design sprints test one risk at a time and track a scorecard. Each week the team maps risks, sketches, builds prototypes and tests with around four new customers. The scorecard colors each part of the hypothesis red, yellow or green: right customer, real problem, right approach, chosen over competition, and differentiation.
Latchet's first scorecard was mostly red, week two had some yellow, and every box was green by week three. The scorecard is new and not in the original book.
He brings a specific hypothesis and prototype to customer conversations. Founder Maruthi John had 50 conversations in a month. He said the first one with a hypothesis and prototype was "night and day".
Customers also see real competitors side by side with the prototypes. Interviews only simulate the real world, but a product that clicks is a strong signal to build.
Editing the hypothesis after each sprint records what the team learned. Mellow's board shows small changes to customer, problem, approach and competition after sprint one. Its first scorecard was mostly yellow rather than red.
Another tweak followed the next sprint, so the team can see how each lesson changed its strategy.
Zeratsky says to outsource prototyping to AI but not the thinking. Two Character Labs teams vibe coded right away and got generic results, because models are trained on existing products. In sprint two they stepped back, thought it through, and only then prototyped.
Detailed paper sketches work as prompt engineering. Teams also decide which parts are real, which are vibe coded and which are a static mockup or video.
He argues that slowing down to think makes building faster. Building has momentum of its own, and AI makes progress easy to see, even in the wrong direction. A launched generic product only yields data from the few who found it, not from those who dismissed it.
His advice is to think hard first about what makes the product unique, then go as fast as possible, like Frodo choosing the path to Mordor before marching.
Founders with clear product market fit can skip the sprint. Some founders in the labs started selling quickly and were told to stop running experiments. Base44's founder sold for $80 million in six months with no such process.
Zeratsky treats that as selection bias, not a template: with evidence and conviction, run with it, otherwise the sprint improves your odds.
Rapidly test and validate any startup idea with the 2-day Foundation Sprint
The episode walks through the Foundation Sprint, a roughly 10 hour, two-day process that settles customer, problem, competition, differentiation and approach before any building starts, then tests the resulting hypothesis in weekly design sprints. Founders who run it say it compresses three to four months of work into three to four weeks, and the example startup Latchet went from a scorecard full of red to all green in three weekly sprints.
Zeratsky co-created the Design Sprint with Jake Knapp and has run it widely. He and Knapp have worked with teams at Google, Microsoft, YouTube, Slack and Uber. The Foundation Sprint appears in their book Click, and its excerpt is among the most popular posts in Lenny's newsletter.
The framework draws on the Design Sprint and work with over 300 teams.
A one-week prototype that rescued a stalled Google project inspired the Design Sprint. Knapp's side project had gone nowhere for three years and its Stockholm office faced closure. A pitch, a PRD and a design could not win over Larry, Sergey and Eric, so the team built a prototype for fellow Googlers, and it became Google Meet.
The aim was a prototype so real that people react as if it were real and want it.
About 200 teams at Google Ventures showed the Design Sprint was missing a first step. Over five years the week was refined into map, sketch, decide, prototype and test. Established companies already knew their positioning, but pre-product startups did not.
The same questions kept returning: what problem, who is the ideal customer, how are we different, who are the real competitors.
Zeratsky, Knapp and Eli Blee-Goldman built the Foundation Sprint at Character Capital. They developed it in 2021 and 2022. Character Capital runs one right after each investment, and now runs more Foundation Sprints than design sprints.
Founders who run it say it compresses three to four months of work into three to four weeks.
Zeratsky has the core team clear its calendar for about 10 hours. Co-founders, or in a larger company the leads of product, engineering, design and marketing, follow a scripted sequence in two blocks of four to six hours. Making key decisions together gives everyone one picture of the project instead of three.
It can take eight hours if the team is fast and twelve if it is larger.
Teams write silently, vote, and let one decider lock each answer. Everyone writes alone, the team votes, then a designated decider, usually the CEO, makes the call and the group moves on. Silence and structure get through many decisions quickly and save energy for the ones that matter.
When teams look up from writing, they see answers they did not expect, which is where early value shows.
The basics cover customer, problem, competition and advantages, answered concretely and fast. Competition includes workarounds like art fairs, not just startups, because if a problem is worth solving people already cope with it somehow. Advantages come from the team's insight, motivation and capabilities.
Teams are often surprised that their answers matched only about two of six items, and that the problem was less clear than they thought.
Zeratsky has teams list several approaches, pick one and name a backup. Latchet weighed an app, a newsletter platform, a Shopify plugin and a full stack build, and chose the social sales app with the full stack as backup. A backup makes failure less scary and pivoting faster.
Even a team locked on one path is asked what else could work and whether a dismissed option deserves another look.
Magic lenses turn a vague debate over approaches into one picture. Each approach is plotted on 2x2 charts for customer, pragmatic, growth, money and differentiation, plus custom lenses such as founder conviction. If one option sits top right on most lenses the choice is easy.
If nothing wins every lens, that is reassuring because no perfect choice exists and the team only has to pick the lens that matters most. About four hours of this replaces weeks or months of building and learning.
The sprint ends in one explicit founding hypothesis sentence. The form is: if we solve this problem for this customer with this approach, they will choose it over the competitors because of two differentiators. Latchet's bet was that artisans would choose a cooperative, easy social sales app over Shopify and Etsy.
Until the hypothesis is explicit, different people hold different hidden versions and nobody can test its variables.
Zeratsky holds back ideation so the team does not lock onto its first idea. The longer a team waits to sketch a solution, the likelier it lands on one that works. Bob Baxley calls the first drawing the "primal mark", since every later decision responds to it.
Zeratsky treats differentiation as the main event because new products get ignored. People guard their attention, so a product needs a promise clear and strong enough to be tried. The promise is the customer's, not the technology or market story told to investors.
Teams that skip it end up painting a message onto the product at the end, when nobody built the thing known to matter from the start.
He wants the product alone in the top right and rivals in Loserville. The team scores itself and each competitor on a few scales, picks the strongest two, and builds a 2x2. Both must hold: the team can deliver it and customers care.
Latchet chose helps you grow and cooperative against Etsy, Shopify and art fairs. Mellow chose mobile first and works out of the box.
Teams warm up on seven classic scales, then write custom differentiators. The classics are fast to slow, smart to not so smart, easy to hard to use, free to expensive, focused to one size fits all, simple to complicated, and integrated to siloed. Custom ones offer customers a new lens on the world, such as networked versus siloed.
Mellow wrote about 25 custom differentiators, such as purposeful, magical and human, before choosing two. Honest scoring dropped the ones it could not deliver.
Zeratsky says to score honestly and accept you will not win everywhere. Driving home one or two things customers care about, where you are radically better, improves the odds of product market fit. Mellow was not best on every scale, which is normal.
Scoring across the whole range also shows where you will win and gives comfort that trailing on some scales is fine.
He advises against leading with price, except for AI cost cuts. Price is rarely the most important differentiator, and big rivals can run loss leaders. It can work for AI products that solve problems software could not touch before, such as BindWell designing precision pesticides without a large team of R&D chemists.
Because AI output may be a bit worse than a human process, the rule of thumb mentioned is about 10x cheaper.
Differentiators become principles that make a one-page mini manifesto. Latchet's principle is that sellers should help each other and earn more. Google's "fast is better than slow" was used in meetings to pick the approach that rendered 12 milliseconds faster.
The manifesto describes how the team believes it will win and is the decision guide, while the basics page holds the core facts. Day one reaches it in about four hours.
The biggest hits he has seen from the inside made one promise that set them apart. Gmail beat Hotmail and Yahoo with massive storage and great search. Slack promised fun and better teamwork, ChatGPT gave a direct answer with zero clicks, and Reclaim used AI to prioritize your calendar and was acquired by Dropbox last year.
Those teams had no Foundation Sprint. Some had the wrong promise and changed it after experiments, but they stayed keyed in on differentiation.
Weekly design sprints test one risk at a time and track a scorecard. Each week the team maps risks, sketches, builds prototypes and tests with around four new customers. The scorecard colors each part of the hypothesis red, yellow or green: right customer, real problem, right approach, chosen over competition, and differentiation.
Latchet's first scorecard was mostly red, week two had some yellow, and every box was green by week three. The scorecard is new and not in the original book.
He brings a specific hypothesis and prototype to customer conversations. Founder Maruthi John had 50 conversations in a month. He said the first one with a hypothesis and prototype was "night and day".
Customers also see real competitors side by side with the prototypes. Interviews only simulate the real world, but a product that clicks is a strong signal to build.
Editing the hypothesis after each sprint records what the team learned. Mellow's board shows small changes to customer, problem, approach and competition after sprint one. Its first scorecard was mostly yellow rather than red.
Another tweak followed the next sprint, so the team can see how each lesson changed its strategy.
Zeratsky says to outsource prototyping to AI but not the thinking. Two Character Labs teams vibe coded right away and got generic results, because models are trained on existing products. In sprint two they stepped back, thought it through, and only then prototyped.
Detailed paper sketches work as prompt engineering. Teams also decide which parts are real, which are vibe coded and which are a static mockup or video.
He argues that slowing down to think makes building faster. Building has momentum of its own, and AI makes progress easy to see, even in the wrong direction. A launched generic product only yields data from the few who found it, not from those who dismissed it.
His advice is to think hard first about what makes the product unique, then go as fast as possible, like Frodo choosing the path to Mordor before marching.
Founders with clear product market fit can skip the sprint. Some founders in the labs started selling quickly and were told to stop running experiments. Base44's founder sold for $80 million in six months with no such process.
Zeratsky treats that as selection bias, not a template: with evidence and conviction, run with it, otherwise the sprint improves your odds.