Jason Cohen
Four-time founder (WP Engine, A Smart Bear) and investor in 60+ startups
Where to hear and read about Jason Cohen1 episode or test
Jason Cohen is a four-time founder (including WP Engine and A Smart Bear), an investor in over 60 startups and a long-time blogger. The thing most worth learning from him is to diagnose stalled growth in a fixed order, starting with customers who leave, because fixing the earliest broken step beats tuning later ones.
Too expensive is never ever ever the reason.
The surprising advice from a founder who built 2 unicorns | Jason Cohen (WP Engine)
Jason Cohen: the story and the lessons
1Who Jason Cohen is and how he works
Cohen has built four companies and invested in over 60 startups. He is a four-time founder with two unicorns, including WP Engine, and his view comes from one bootstrapped company, one VC-funded company and about 60 investments with mixed outcomes.
He shares a 2017 Central Texas Entrepreneur of the Year award with WP Engine CEO Heather Bruner, whom he calls a late-joining co-founder.
Cohen publishes rarely and only when a piece is his best. He has written about 300 to 350 posts in 18 years, and 150 to 200 of them he is proud of. Some years he posted once or twice, others about 40 times.
He ignores the advice to post on a schedule, and running two unicorns at once left no room for volume. He admits the high bar can tip into paralysis and says he still struggles with that.
His book Hidden Multipliers argues small decisions compound into big gains. Cutting monthly cancellations from 5% to 4% sounds small but has a large effect on revenue and profit. Onboarding customers up front is another multiplier.
His next book covers decisions under uncertainty, such as quitting, moving or marrying. He says it is a relief to put the tools down, since market research won't say whether to sell.
Cohen builds theories on real data and tests them with AI. He uses Gemini to turn chart images into tables that paste into Google Sheets, so he can check a theory against many real examples.
With ten examples of something, a theory that was once too hard to test becomes testable. He uses this often for his articles and book.
2Cohen on finding and fixing churn
Cohen starts every growth stall with logo churn, the worst problem. Once a customer leaves you can't save them or expand their revenue, and churn often brings negative reviews. A canceling customer already found you, accepted the price, had budget and onboarded.
He says fixing later funnel steps while this one is broken is the same mistake as tuning the bottom of a funnel that breaks higher up.
Cohen computes a customer ceiling to make churn feel real. Maximum customers equals new customers per month divided by monthly cancellation rate. 100 new a month at 5% churn caps the company at 2,000.
Marketing grows only as fast as you improve it, while cancellations grow automatically with size, so they eventually overtake it. He calls anything above 3% monthly cancellation terrible, but a 2% rate is already good for small business customers.
Cohen distrusts drop-down cancellation reasons and asks open questions. At Smart Bear the first item in the list was picked most. After he randomized the order, every option was picked about equally, so it was noise.
Asking 'what made you cancel' instead of 'why did you cancel' doubled usable responses at Groove, from 10% to 20%.
Cohen says 'too expensive' is never the real reason a customer cancels. The customer already saw the homepage and pricing page and bought. The real cause is often something like a missing integration, such as a tool that works with Jira but not Linear.
He treats a stated reason as a proximate cause of death and digs further back until it is actionable. He rejects the idea of a single root cause and says five whys wrongly implies one sits at the bottom.
Cohen talks to struggling customers before they cancel. Warning signs include never uploading data, too many or too few support calls, and long gaps without logins. These customers can still be saved and still talk to you.
Without much data, form a theory, act and adjust. When comparing good and bad customers, look for what good customers share that bad ones don't.
Cohen bets on onboarding when the best churn fix is unclear. Most companies lose more customers in the first day, 30 days or 90 days than later, and early leavers are the most unprofitable because acquisition cost is never repaid.
He compares it to a YouTube retention curve: if 50% of viewers drop in the first 30 seconds, moving to 55% staying could lift the 20% still watching at the end to 25%.
Cohen uses AI for themes in feedback but reads the details himself. He calls an LLM an averaging machine. He asks it to list every specific detail under each theme, with the customer who said it.
The themes are predictable, but the details trigger action. Most companies have few enough cancellations to just read them.
3How Cohen thinks about pricing and positioning
Cohen says most prices are too low because they were guessed once. He cites Patrick Campbell's 4,200 startup data points: prices are way too low because founders guessed and never changed them. Fear of raising prices is emotional, not economic.
Pricing is art as much as science, and a price set years ago may no longer fit a market that has changed.
Cohen holds that price selects the market, so raising it often costs no signups. A founder selling to enterprise at $300 a year moved to $300 a month, a 12x increase, and still got one or two signups a week.
A thousand-person company sees a very cheap product as immature. Cohen told the founder to raise again, since a 12x rise with no visible change means the price is not near its ceiling. He describes demand as a mesa, not the textbook slope.
Cohen prices against growth delivered, not money saved. A savings pitch caps price at a fraction of the savings: a customer spending $40K a month on ads who halves costs saves $20K, and pays about a quarter. Selling 'double your leads' lets the same product charge $40K instead of $5K, 8x more.
He says the story is simplified and a bit exaggerated to make the point. Price also means structure, such as per site, per usage or per seat.
Cohen treats a price change as a strategy change. Bigger buyers may need SOC 2, governance features, new integrations or professional services, and a company's edge may only matter in its current market.
Buffer could sell upmarket but chose to stay a company for the little people. Jen Abel's point that enterprises buy most easily at $75K to $150K a year shows how the bucket shapes the sale.
Cohen's test for any pricing move is whether the customer would call it better. A change must be good for the company and good for the customer. Internal proposals often call a price rise 'better' for customers just to justify it.
He says investor pressure can push policies that work short term but are a bad long-term strategy.
4Cohen on expansion, channels and knowing when to stop growing
Cohen pairs net revenue retention with logo churn to stay honest. Median NRR at IPO for SaaS companies is 119%, and only about two public SaaS companies sit below 100%. But logo count has no upgrades, so it only falls.
A 20% loss needs a 25% gain to recover. If logos fall too fast, too few customers remain to upgrade.
Cohen raises expansion by creating value and splitting it with the customer. Add features and tiers, or price by usage or seats so revenue rises with value. If the price doubles, the customer should feel about five times the value.
Where no single metric exists, he uses proxy metrics or interviews. He says of the split: 'It's not a charity.'
Cohen turns to referrals or a second product when customers won't expand. Referrals triple when the base triples, like cancellations. AG1 sells a sleep supplement to existing buyers, and Duolingo offers gems and app colors as small ways to pay more.
In B2B referrals may not work, since a mid-size company doesn't refer another company. The question stays the same but its form changes.
Cohen assumes every acquisition channel saturates on an elephant curve. It is an S-curve whose tail sags. Magazines reported rising circulation, then went out of business, and conferences reported great attendance, then folded.
The people running a channel won't say it is shrinking. If you can't say which channels are saturated, he suggests the answer may be all of them.
Cohen restarts growth with a new kind of channel or a bet anchored in strength. Constant Contact held in-person city workshops for small businesses using power-user agencies, and HubSpot's agency channel became about 50% of revenue after four or five years.
For a new product or market, he says to keep one foot planted in something you already do well. The foot that moves is the risky one, so the upside must be big.
Cohen asks whether growth is the right goal at all. Growth can be redefined, for example from revenue to profit. Bootstrapped companies like 37signals hit a limit, and their founders are paid millions a year in dividends.
He says 'if you're not growing, you're dying' is most true of the person, not the company, since years of the same work mean no learning or career growth. His aim is to grow as much as possible within limits you are proud of.
Cohen says quitting a company can't be settled by expected value. Nobody knows the probability curves, and even with a probability you sell the company once or you don't.
Lenny cited Matt MacInnis's rule that a startup not clicking after four or five years is probably time to leave. Cohen treats it as a one-time personal choice.
Cohen thinks routine A/B testing of small details mostly wastes time. It can't judge strategy, nobody A/B tests whether Uber is a good idea, and with rare real wins even a 95% accurate tool gives more false positives than wins.
Teams stack winners for a year and find conversion unchanged. Shopify's holdout group showed about a third of found effects disappeared later, and Lenny argued it can still pay off at millions of users.
The surprising advice from a founder who built 2 unicorns | Jason Cohen (WP Engine)
A five-step diagnostic for stalled growth, in order: logo churn, pricing and positioning, net revenue retention, saturated marketing channels, and whether you need to grow at all. It matters because small fixes at the first broken step beat tuning later steps, and one repositioning example raised price 8x for the same product.
Cohen has built four companies and invested in over 60 startups. He is a four-time founder with two unicorns, including WP Engine, and his view comes from one bootstrapped company, one VC-funded company and about 60 investments with mixed outcomes.
He shares a 2017 Central Texas Entrepreneur of the Year award with WP Engine CEO Heather Bruner, whom he calls a late-joining co-founder.
Cohen publishes rarely and only when a piece is his best. He has written about 300 to 350 posts in 18 years, and 150 to 200 of them he is proud of. Some years he posted once or twice, others about 40 times.
He ignores the advice to post on a schedule, and running two unicorns at once left no room for volume. He admits the high bar can tip into paralysis and says he still struggles with that.
His book Hidden Multipliers argues small decisions compound into big gains. Cutting monthly cancellations from 5% to 4% sounds small but has a large effect on revenue and profit. Onboarding customers up front is another multiplier.
His next book covers decisions under uncertainty, such as quitting, moving or marrying. He says it is a relief to put the tools down, since market research won't say whether to sell.
Cohen builds theories on real data and tests them with AI. He uses Gemini to turn chart images into tables that paste into Google Sheets, so he can check a theory against many real examples.
With ten examples of something, a theory that was once too hard to test becomes testable. He uses this often for his articles and book.
Cohen starts every growth stall with logo churn, the worst problem. Once a customer leaves you can't save them or expand their revenue, and churn often brings negative reviews. A canceling customer already found you, accepted the price, had budget and onboarded.
He says fixing later funnel steps while this one is broken is the same mistake as tuning the bottom of a funnel that breaks higher up.
Cohen computes a customer ceiling to make churn feel real. Maximum customers equals new customers per month divided by monthly cancellation rate. 100 new a month at 5% churn caps the company at 2,000.
Marketing grows only as fast as you improve it, while cancellations grow automatically with size, so they eventually overtake it. He calls anything above 3% monthly cancellation terrible, but a 2% rate is already good for small business customers.
Cohen distrusts drop-down cancellation reasons and asks open questions. At Smart Bear the first item in the list was picked most. After he randomized the order, every option was picked about equally, so it was noise.
Asking 'what made you cancel' instead of 'why did you cancel' doubled usable responses at Groove, from 10% to 20%.
Cohen says 'too expensive' is never the real reason a customer cancels. The customer already saw the homepage and pricing page and bought. The real cause is often something like a missing integration, such as a tool that works with Jira but not Linear.
He treats a stated reason as a proximate cause of death and digs further back until it is actionable. He rejects the idea of a single root cause and says five whys wrongly implies one sits at the bottom.
Cohen talks to struggling customers before they cancel. Warning signs include never uploading data, too many or too few support calls, and long gaps without logins. These customers can still be saved and still talk to you.
Without much data, form a theory, act and adjust. When comparing good and bad customers, look for what good customers share that bad ones don't.
Cohen bets on onboarding when the best churn fix is unclear. Most companies lose more customers in the first day, 30 days or 90 days than later, and early leavers are the most unprofitable because acquisition cost is never repaid.
He compares it to a YouTube retention curve: if 50% of viewers drop in the first 30 seconds, moving to 55% staying could lift the 20% still watching at the end to 25%.
Cohen uses AI for themes in feedback but reads the details himself. He calls an LLM an averaging machine. He asks it to list every specific detail under each theme, with the customer who said it.
The themes are predictable, but the details trigger action. Most companies have few enough cancellations to just read them.
Cohen says most prices are too low because they were guessed once. He cites Patrick Campbell's 4,200 startup data points: prices are way too low because founders guessed and never changed them. Fear of raising prices is emotional, not economic.
Pricing is art as much as science, and a price set years ago may no longer fit a market that has changed.
Cohen holds that price selects the market, so raising it often costs no signups. A founder selling to enterprise at $300 a year moved to $300 a month, a 12x increase, and still got one or two signups a week.
A thousand-person company sees a very cheap product as immature. Cohen told the founder to raise again, since a 12x rise with no visible change means the price is not near its ceiling. He describes demand as a mesa, not the textbook slope.
Cohen prices against growth delivered, not money saved. A savings pitch caps price at a fraction of the savings: a customer spending $40K a month on ads who halves costs saves $20K, and pays about a quarter. Selling 'double your leads' lets the same product charge $40K instead of $5K, 8x more.
He says the story is simplified and a bit exaggerated to make the point. Price also means structure, such as per site, per usage or per seat.
Cohen treats a price change as a strategy change. Bigger buyers may need SOC 2, governance features, new integrations or professional services, and a company's edge may only matter in its current market.
Buffer could sell upmarket but chose to stay a company for the little people. Jen Abel's point that enterprises buy most easily at $75K to $150K a year shows how the bucket shapes the sale.
Cohen's test for any pricing move is whether the customer would call it better. A change must be good for the company and good for the customer. Internal proposals often call a price rise 'better' for customers just to justify it.
He says investor pressure can push policies that work short term but are a bad long-term strategy.
Cohen pairs net revenue retention with logo churn to stay honest. Median NRR at IPO for SaaS companies is 119%, and only about two public SaaS companies sit below 100%. But logo count has no upgrades, so it only falls.
A 20% loss needs a 25% gain to recover. If logos fall too fast, too few customers remain to upgrade.
Cohen raises expansion by creating value and splitting it with the customer. Add features and tiers, or price by usage or seats so revenue rises with value. If the price doubles, the customer should feel about five times the value.
Where no single metric exists, he uses proxy metrics or interviews. He says of the split: 'It's not a charity.'
Cohen turns to referrals or a second product when customers won't expand. Referrals triple when the base triples, like cancellations. AG1 sells a sleep supplement to existing buyers, and Duolingo offers gems and app colors as small ways to pay more.
In B2B referrals may not work, since a mid-size company doesn't refer another company. The question stays the same but its form changes.
Cohen assumes every acquisition channel saturates on an elephant curve. It is an S-curve whose tail sags. Magazines reported rising circulation, then went out of business, and conferences reported great attendance, then folded.
The people running a channel won't say it is shrinking. If you can't say which channels are saturated, he suggests the answer may be all of them.
Cohen restarts growth with a new kind of channel or a bet anchored in strength. Constant Contact held in-person city workshops for small businesses using power-user agencies, and HubSpot's agency channel became about 50% of revenue after four or five years.
For a new product or market, he says to keep one foot planted in something you already do well. The foot that moves is the risky one, so the upside must be big.
Cohen asks whether growth is the right goal at all. Growth can be redefined, for example from revenue to profit. Bootstrapped companies like 37signals hit a limit, and their founders are paid millions a year in dividends.
He says 'if you're not growing, you're dying' is most true of the person, not the company, since years of the same work mean no learning or career growth. His aim is to grow as much as possible within limits you are proud of.
Cohen says quitting a company can't be settled by expected value. Nobody knows the probability curves, and even with a probability you sell the company once or you don't.
Lenny cited Matt MacInnis's rule that a startup not clicking after four or five years is probably time to leave. Cohen treats it as a one-time personal choice.
Cohen thinks routine A/B testing of small details mostly wastes time. It can't judge strategy, nobody A/B tests whether Uber is a good idea, and with rare real wins even a 95% accurate tool gives more false positives than wins.
Teams stack winners for a year and find conversion unchanged. Shopify's holdout group showed about a third of found effects disappeared later, and Lenny argued it can still pay off at millions of users.