The surprising advice from a founder who built 2 unicorns | Jason Cohen (WP Engine)
These notes were generated by AI and may contain inaccuracies.
A lot of product teams and founders build something that starts showing success, then growth stops. Series of questions to diagnose why growth is slowing, in order of priority.
First question: Are customers leaving? (logo churn). Think about the gauntlet customers went through: finding out about the product (hard and improbable), not bouncing off homepage, reaching pricing page without being scared off, having budget and buying it. After all that, clearly wanting it to work, they leave. Emotionally, that's terrible - fundamentally not fulfilling the promise made (product or communication issue).
Churn is worst problem: nothing can be done once it happens (no saving them, no increasing revenue). Correlated with negative reviews hurting growth.
Math: Cancellations grow faster than marketing, creating hard ceiling on company size. Triple customers overnight: marketing delivers same new customers (AdWords, SEO don't care about size), but cancellations triple (e.g., 5% of tripled base). Cancellations grow automatically with size; marketing only grows as fast as you improve it (hard, linear).
Maximum size formula: New customers per month / monthly churn rate = maximum customers. Example: 100 new customers/month at 5% churn = max 2,000 customers (100 / 0.05). As you approach, growth slows (in ≈ out). Applies to logo churn or revenue churn (dollars in / churn rate).
Marketing includes all growth work: PLG, marketing, sales (PLG still needs initial marketing).
Anything above 3% per month cancellation is terrible.
Customers don't want to tell you why they cancel (already mentally out the door). Dropdown forms unreliable: at Smart Bear, "too expensive" topped list (first option); randomized list made all equal - complete noise.
Ask open-ended: "What made you cancel?" (not "Why did you cancel?"). Groove case study: "Why?" got 10% usable responses; "What made you?" got 20%.
"Too expensive" never true reason: they already saw pricing, decided to buy (action proves not too expensive). Something else happened (didn't fulfill promise, lost budget - different issue).
Analogy to healthcare: approximate cause (stopped breathing) vs. real (undiagnosed diabetes). Dig deeper; complex systems have many interlocking causes, not one root cause (root cause analysis incorrect).
Jason Cohen: four-time founder (two unicorns, including WP Engine), investor in 60 startups, writer at smartbear.com for almost 20 years (150-200 proud posts, 300-350 total over 18 years). Only publishes best work (not regular; some years 1-2 posts, others 40). Built unicorns while writing.
Framework for stalled growth: practical, actionable.
Upcoming book: Hidden Multipliers (pre-order at hiddenmultipliers.com).
Quality > consistency. Rarely writing means each must be more awesome (pressure). Tension: perfection motivates but causes paralysis.
Many companies experiencing slowed growth: economy, AI threat, size (can't 2x forever), algorithm changes, or gradual deceleration (running through mud).
Jason's experience: built four companies (two unicorns: one VC-funded, one bootstrap), invested in 60 (some failed, some successful).
Framework: series of questions in order (fix first before later ones, like funnel leak).
Second question previewed: pricing positioning (prices too low, guessed, unchanged; raise prices, signups often don't change). Large companies think $2-100/month "can't be good enough".
Other questions: saturated channels? Need to grow? ("If not growing, dying" - investor pressure?).
Tenweb: AI website builder (2M+ sites in 3 years). VIP coding platform like Lovable for WordPress (front/back-end, any complexity: e-commerce, portfolios, blogs; WordPress admin + thousands plugins). API for SaaS, agencies, hosting. 10web.io/lenny (code lenny: free credits, 30% off).
Strella: AI customer research platform. Automates interviews/analysis (fast, continuous). AI moderator probes deeper, surfaces patterns (hours vs. weeks). Used by Amazon, Duolingo for Figma testing, concept validation, journey research. str.io/lenny.
Becoming famous is amazing. Being famous is a mixed bag. Losing fame is miserable.
Undiagnosed diabetes is much more of a cause than stopped breathing. When customers say "it's too expensive," dig deeper: they wanted this stuff, but it didn't work with Linear, only with Jira, revealing a lack of integration. Decide whether to build that integration based on frequency and other factors—not just dismiss as expensive.
Avoid simplistic root cause thinking. Five Whys sometimes implies a single root at the bottom; instead, generate an array of potential factors. For "project ended," don't assume nothing can be done. If software was more successful, the project might not have ended—indicating product wasn't useful enough. Consider if targeting small businesses or consumers leads to higher churn due to high variance and frequent project endings. Is it your fault for picking the wrong ideal customer profile or target segment? Saying "nothing we can do" closes doors on improvements. Market segment choice heavily impacts retention since behaviors differ.
Ask customers in trouble but not yet canceled to learn more and potentially save them. Signals include: never uploaded data, calling tech support too much or too little, not engaged, didn't log in for a while. Correlate signals with cancellation using data, or guess, act, and adjust with more data.
Most cancellations happen in first day, 30 days, or 90 days. Small onboarding changes yield large effects, unlike later stages. YouTube video retention drops sharply in first 30 seconds (e.g., 50% fall-off), then flattens. Shifting from 50% to 55% retention in first 30 seconds can boost end retention from 20% to 25%—a 20-30% increase.
In SaaS, early churn is unprofitable due to acquisition costs not recouped. Improving onboarding pays off enormously in revenue and profit.
If you don't know what to do, focus on onboarding. Even if you do know, onboarding is still a good bet.
Calculate maximum customers: new customers divided by cancellation rate. Everyone chases new customers, but this caps growth.
- Use free-form "What made you cancel?" instead of multiple-choice.
- AI good for themes, bad for actionable details (it's an averaging machine). Use AI to extract themes, then specific details with customer links for human review. Still best to read responses manually, especially with low volume.
- Dig beyond surface reasons (not just pricing or project ended).
- Catch customers early via signals before cancellation.
- Prioritize onboarding/activation.
For SMB, 2% monthly cancellation is good but worth improving unless at floor.
Newer companies often price too low: guessed, copied competitors, or adjusted arbitrarily. Patrick Campbell (4,200 startup data points): "Your prices are way too low because you just guessed and you haven't changed them."
Raising prices doesn't always reduce demand per textbook curves. Often, signups stay same or increase. Example: Enterprise/government SaaS at $300/year (1-2 signups/week). Changed to $300/month (12x increase)—still 1-2/week. Advised to raise again since no change observed.
Pricing selects the market. Low prices signal low quality to mid-size/enterprise (e.g., $2-100/month looks immature, poor support). They won't buy. Raising prices enters better market, improves metrics like retention. Demand curve slopes up into viable range, plateaus, then down—like a mesa, not steady decline.
Pricing is art and science; world changes over 5-10 years. Not just webpage number—structure and positioning matter.
Company "Double Down" halves AdWords costs. Customer spends $40K/month on AdWords.
- Cost-saving pitch: "Cut AdWords cost in half." Saves $20K, but customer pays ~$5K/month to net $15K savings.
- Growth pitch: "Double the leads per month." Customer willing to spend $40K to double leads (same budget).
Growth (market share, valuation) beats cost-saving for CEO. Same product charges 8x more via positioning.
What does the CEO want to hear more? Growth or saves money? Growth is healthier competitively.
Pitching to double leads per month: If willing to spend $40k for current leads, willing to spend $40k to double them. Double Down gets $40k instead of $5k for cheaper leads, resulting in 8x the money. CEO reaction: doubled leads at same ROI and CAC. Leads to "how can we do more of that?" Same product, but structured differently.
Pricing is not just the number on the page—it's positioning, budget alignment, structure (per site, usage, seat). Same price/product can seem fair or expensive based on structure. Lesson for product managers: sell more of what the company values (growth, retention, competitiveness). Growth is obvious. Avoid capping at savings/ROI/time efficiency; value what they value for higher cap (order of magnitude higher).
When asking if pricing is correct, consider structure, positioning. Growth slowing suggests room for improvement.
Pricing correct isn't the number (20, 25, 100)—it's market, selling approach, price story, positioning of problem solved.
Jen Ael (podcast guest) advice: enterprises sweet spot $75-150k contracts for SaaS. Aim product there vs. $1k-2k/month; everything easier.
Pricing not independent knob—tied to strategy. Raising prices means new customers demand SOC2, governance, integrations, professional services. Next market may not value distinctions/competitiveness. Cultural fit: Buffer stays for "little people," not upmarket.
If cancellations overtake marketing, combat with growth from remaining customers (upgrades, usage-based) proportional to size.
NRR (Net Revenue Retention): Revenue from existing cohort now vs. one year later (cancels/downgrades down, upgrades up). Can exceed 100% if upgrades > cancels/downgrades.
NR (Net Retention) lacks upgrades, only down—most important, keeps honest. NR positive but N down too fast = not enough base left to upgrade.
Investment analogy: $100 down 20% to $80, up 20% to $96—not back to 100%. 20% loss needs 25% gain to recover. NR assumes 20% cancel offset by 20% upgrade = only 96%; N reveals truth.
Public SaaS: 100+ companies, ~2 have NRR <100% (horrible financials/valuations). Median IPO NRR 119%. Mandatory for growth.
Plug logo churn first, then NR: add features/tiers, usage/seat pricing that auto-increases with value.
Customer must feel value exceeds price increase (price doubles, value 5x). Measure customer-perceived value (northstar if quantifiable; proxies otherwise; qualitative if needed). Generate more value, then split (keep customer forefront, not charity).
Land-and-expand: can't 10x from $10k to $100k without 10x value perception—stuck at reference point. Avoid coercion into anti-customer policies; test if truly good for customer.
Sequence power: 1) logo retention (PMF), 2) pricing/positioning (right market/charge), 3) growth/expansion.
Applies B2B SaaS primarily, but mechanics universal. Consumer/small biz harder NR, but try (e.g., Duolingo gems/app colors). Alternatives: second product to same segment (e.g., AG1 sleep supplement), referrals (consumer: invite friend; scales with base).
If stalled, channels may be saturated. Avoid flogging AdWords/SEO teams. Limits exist: search inventory, one position per keyword.
Channels decline over time: starts S-curve (unlock to 100 leads/month), taps out into optimization, then sags (elephant curve: trunk then butt).
Marketing channels follow an elephant curve pattern: initial trunk-like growth surge, then a sag or butt as performance declines. Marketers have stories about this. Reasons include audience saturation from repeated exposure—e.g., 'rule of seven' (seeing an ad seven times before acting), but after 20 views, people tune out. With magazine ads, you'd get a surge then drop-off as people said 'we've seen it.' Channels decline quietly: magazines claimed rising circulation yet went out of business; conferences reported great attendance but failed.
This happens across AdWords, Facebook ads, SEO, affiliates. AI disrupts everything unpredictably—stories vary, future unknown, but not all channels will keep growing. Can't rely on marketing forever; finite channels, all sag eventually. Don't flog marketing if saturated—growth slowing means channels are full, even sagging.
Do you know which channels are saturated? If no, assume all are. Adding one feature and flogging AdWords won't work. Common trap: add feature, expect marketing to push it. Need different approaches.
- Try new channels: direct to SEO/social/indirects, or vice versa. Optimized ads inform SEO content, and vice versa.
- Experience: products selling well direct often fail in social/SEO; SEO traffic doesn't convert well to ads (no data, but feeling).
Interviewer: Always get some wins from channels, but one drives most growth (sales, word of mouth, ads); others layer on top. At scale, add all channels affordably, but get creative: new product, market, or redefine channel.
Constant Contact (email marketing for small businesses): Held in-person workshops in cities teaching email marketing using their tool. Attendees (restaurants, dentists) became customers. Cost-effective despite physical effort, low $20/month price. Used power-user agencies as instructors, who also became customers.
HubSpot: Tested selling through agencies (not direct)—became 50% of revenue after 4-5 years.
WP Engine: Many sites sold through WordPress agencies.
New channels via humans/ecosystems can dramatically boost growth.
Time for next product? Stay in target market, leverage company strengths/assets. Framework: plant one foot in strength/asset, move other foot (risky part) into upside bet—makes it smart. Forces drastic bets when marketing exhausted.
Interviewer praises methodical questions for undoing stalled growth, generating ideas. Assume working channels (SAPI, TikTok) will slow as others pile in—elephant curve dips as competitors discover.
Explore new channels before apex. People know but don't act.
Most companies fail: something works, then stops, growth stalls permanently. Links in show notes: Emily Kramer on ecosystems as growth channel (influencers, content, partners); Whiz growth head: 'Why start with zero when you can start with 10,000?' Upcoming post on ChatGPT app store for submitting apps—huge potential channel.
After fixing logo retention, pricing, NRR, marketing/channel saturation—ask: is stalled growth a problem? Define 'grow': this product/market/company? Maybe no—need different product/market. Or shift from maximizing revenue to profit.
Examples: 37signals, bootstrap companies hit limit, pay founders millions in dividends—okay to stay small, avoid disliked organization/market.
Philosophically: why grow? 'If not growing, you're dying'—investor pressure? Even bootstraps: stagnation bad for founder fulfillment (same thing daily?), employees (no career growth, innovation?).
Some okay with stasis (e.g., CPA with clients). Question mode of constant growth/promotion—does it make you happy/fulfilled? Stuck in rut?
For product builders/innovators/founders: likely not rote for 20 years. 'Not growing, you're dying' applies to you the person (shark must move), not just company. Retirees decline without purpose.
Context-specific: if stagnant after exhausting options, drastic change—leave, pivot company, sell, change jobs.
Don't wait—check fulfillment annually. True at all scales; public companies stagnate at natural size. Philosophical but important.
Interviewer: Bootstrap founders can be happy with flat revenue. Psychologically hard, but question ego/habit. Avoid regrettable 'growth at all costs.' Set boundaries: grow proudly (e.g., no crappy content). Early stage: looser standards to survive (copy styles), tighten as successful.
Product okay not growing—side-run or sunset, pursue else. Matt McInness (CPO Rippling): quit after 4-5 years if not clicking (most won't succeed long-term).
Interviewer book on uncertainty decisions (quit, move, marry, launch)—probability/expected value fails (unknown probs, binary life outcomes). Drop those tools; market research won't answer selling company.
Hidden Multipliers (pre-order at hiddenmultipliers.com): Little decisions/actions with huge impact (e.g., cancellation 5% to 4%, onboarding focus). Turns 'I know' into action via deep examples/steps. Covers revenue/profit boosters—things never considered or not acted on.
Jason emphasizes that the common thread throughout growth advice is ensuring customers get real value. The product must promise the right thing, deliver on it, enable easy onboarding, help customers realize the value, and measure it to confirm increases. This north star—creating value in the customer's language and way—guides pricing, retention, and other elements. While platitudes like this aren't actionable alone, the detailed discussion provides depth.
Host notes that value might exist but could be targeting the wrong customers or market, or poor positioning. Jason agrees: many ways to get it wrong, like failing to communicate value clearly on the homepage so visitors instantly recognize it. Solving problems is tough because everything must align.
Jason shares using AI, especially Gemini, to convert chart images from the internet into copy-pasteable tables for Google Sheets. This enables quick analysis, model-building, or application of data. Useful for his book and articles where he incorporates real data. Now, gathering 10 examples to test theories is feasible, which was previously too hard.
Host praises the tip, noting it's cool for reversing infographic generation (e.g., with Gemini or Midjourney) by feeding charts back to AI for text extraction.
Jason believes A/B testing doesn't work well for most things, especially strategy, vision, or insights critical to success—you don't A/B test if Uber is a good idea. Even for details, common practices fail:
- Testing mundane changes like verbs or ads leads to picking 'winners' (e.g., 7th or 8th variant), but stacking them over a year yields no improvement in metrics like conversion rates.
- Tools lack statistical accuracy; even 95% accurate tools produce more false positives than true effects when positives are rare.
Result: Mundane A/B testing is mostly false positives and a waste of time unless highly sophisticated. Sophisticated teams exist, but for others, it's like poker—if you don't spot the sucker, it's you.
Host acknowledges personal success with A/B at scale (millions of users, where 1% gain = millions in revenue), but Jason notes most aren't at that scale. He references host's Shopify CTO interview: even their 100-person team finds ~1/3 of effects disappear; they use holdout groups to verify persistence, and most don't last.
Recommended Books
On Writing Well by William Zinsser: Aim to write like this on best days; core lesson is relentless cutting—always more to cut. Anecdote: Zinsser to amateur writer complaining 'writing is hard': 'I never knew writing could be easy.' Echoes Hemingway: 'Writing is easy. You just sit at the typewriter and bleed.'
Crossing the Chasm by Geoffrey Moore: Everyone knows the diagram, few read the book. Details market definition, actions, and models. Early companies seek peers in the market saying 'This is great' (not just any early adopter). Markets respect internal opinions, so case studies don't cross markets.
Favorite Recent Movie/TV Show
ER (1994, 15 seasons): Holds up after 25 years; Jason watches all with 16-year-old daughter (now on season 13), who calls it great TV despite long episodes and endless seasons.
Host recommends The Pitt on Netflix (recent awards winner); fun fact: cousin appeared as young patient, now a fancy actress.
Favorite Recent Product
Whisper Flow for dictation: Excellent with keyboard shortcuts, works across software.
Anker products (power stations, docks, chargers): Super high quality; everything charges twice as fast.
Host loves new Anker charger with display showing percentage, wattage per port.
Shoutout: Newsletter insiders (annual subscribers via Lenny's Product Pass at lennispodcast.com) get 1 year free Whisperflow; currently 19 products, more coming.
Favorite Life Motto
'Be yourself. Everyone else is taken.' Attributed to Oscar Wilde (unverified, but fits). Deep truth: Avoid mimicking others online; fans love the authentic you.
Award Story
Ernst & Young Entrepreneur of the Year 2017 for Central Texas, co-won with Heather Brunner (WP Engine CEO). Calls her 'late-joining co-founder' (like Reid Hoffman called Jeff Weiner at LinkedIn)—joined 4 years in but transformed success, culture. After 11 years, everyone sees her as co-founder. Rare co-winner.
Host praises conversation's value, anticipates listener ideas for products.
Jason: Find free articles at asmartbear.com (links to Twitter etc.); non-commercial, no ads/courses. Book Hidden Multipliers at hiddenmultipliers.com is best work (physical, so paid). Ordering helps; free content repays his past generosity.
Keep Lenny's Podcast in your library
Save the videos and channels worth coming back to, and find them again in one place.
