Product Playbooks
Decisions that shape the product itself — what to build next, when to say no, and how founders used real feedback to steer the roadmap without losing focus.
277 tactics · page 7 of 10
“the place that it had the least impact on anything including conversion to paying including hard activation rate was on first launch so as soon as someone opens the app for the first time putting it straight there and just having them press and moving on from there which was shocking.”
ATT Prompt on First Launch Had the Least Impact on Conversion — Earlier Than Expected
Conventional wisdom says delay the ATT permission prompt until after users experience value, to maximize opt-in rates. Hannah's testing showed the opposite: showing the ATT prompt on first launch had the least negative impact on conversion-to-paying and hard activation of all placements tested. Users aren't making value judgments at launch — they haven't formed any yet. The earlier position also avoids interrupting a moment that actually matters.
“I always encourage startups to really think very honestly about what your lifetime value of a customer is in terms of gross profit contribution and also be very thoughtful on your payback period.”
LTV Must Mean Gross Profit, Not GMV — And Payback Period Is What Actually Matters
Phil sees founders routinely present LTV figures that include gross merchandise value while taking a 1-10% take rate — creating numbers that look like $300 LTV when the true gross profit LTV is $3-30. Separately, a long LTV with a 24-month payback period creates a working capital crisis: you're writing a $1 check to Facebook today and waiting two years to get it back. The actionable rule: model LTV as gross profit dollars, model payback in months, and don't raise marketing spend until the math clears.
“We meant to test it in a geography or two but we unlocked it somehow across all geographies and we crashed tinder it was we had an amazing increase in daily actives as a function of that but we also accidentally crashed the system.”
Swipe Surge: Phil and Jeff Morris Accidentally Crashed Tinder Proving the Feature Worked
Swipe Surge started from a data observation: when more users are active simultaneously in a geography, everyone's match rate improves. The feature notified users when a local surge was happening. Phil and Jeff Morris accidentally launched it globally instead of as a two-geography test, crashed the servers, and got a lecture from ops — but validated the concept instantly. The story illustrates how rapid, imperfect experiments reveal high-signal product insights faster than cautious rollouts.
“About 70% of the world population do not have that ability and they need to be approached because you know all of the fitness industry is based on that pretty much that very healthy concept of yeah you get you need to get more responsible you need to get moving.”
70% of People Lack Intrinsic Fitness Motivation — Build for Them, Not the 30%
Anton observed that the entire fitness industry — from gym memberships to wellness apps — is built around intrinsic motivation: you want to be healthy, so you act accordingly. But 70% of people don't have that internal drive. Sweatcoin was designed for the majority, not the minority, using external incentives (literal rewards) to substitute for intrinsic drive. This insight shapes which product you build: if your target user lacks the intrinsic motivation your category assumes, you need to engineer an external trigger system.
“How much do you value your sweat coins at... people then would say anywhere between one and 5 cents and then we ask them how much would you sell your sweat coins from and guess what the answer was... 50 cents to a dollar.”
Users Value Their Own Coins at 10x What They'd Pay for Someone Else's
When asked to price their own sweatcoins versus what they'd pay to acquire someone else's, users revealed a 10-50x asymmetry: they'd buy at 1-5 cents but wouldn't sell below 50 cents to $1. This reflects the emotional value of effort-earned progress. It's also why sweatcoin deliberately didn't make coins fully fungible: a discoverable market price would destroy the perceived value. The more a virtual currency feels 'earned' rather than 'purchased,' the more psychologically potent it becomes.
“In order for gamification to work well there's got to be an underlying experience that's actually good for the people and they know it if the gamification is used for you to watch more ads you may be successful shortterm but long term it's not going to work.”
Gamification Only Works If the Underlying Experience Is Genuinely Valuable
Anton warns against gamification as a manipulation layer on top of a mediocre product. Duolingo works because learning a language is intrinsically valuable; Sweatcoin works because walking is genuinely beneficial. When gamification serves an underlying experience users already believe in, the game mechanics amplify that belief. When gamification is purely extractive (designed to drive ad views or increase session time for its own sake), users eventually recognize it — and the backlash is proportional to how long they felt deceived.
“What else could people pay for additional services? What we've seen is marketplaces or transactions spinning off these — if you have a really passionate user base doing camping, what about a marketplace to buy and sell used tents?”
CSS needs net revenue retention story even without SaaS expansion — consumables and marketplaces fill the gap
CSS lacks SaaS-style seat expansion, so NRR is structurally capped below 100%. Eric flags this as a known CSS limitation in public markets, but points to two paths: consumable add-ons (like Tinder boosts), and marketplace transactions that monetise existing user behaviour. The prerequisite is scale — he explicitly warns founders against adding a marketplace before $20-50M ARR, where focus still beats complexity.
“The pros, students, athletes can join the platform — we're never going to disrupt with a paywall the interaction between the two. It's a touchstone, a third rail. You change that and it's a fundamental tenant of our business.”
Identify your non-negotiable free tier first — it protects the network effect core
Every freemium strategy has at least one non-negotiable: the feature or user type that makes the whole product work. For V1 Sports, the coach-student relationship cannot be paywalled — doing so would destroy the network. Revisiting strategy every 18 months does not mean everything is on the table. Identifying the real third rails early focuses re-evaluation energy on the genuinely moveable parts.
“So many apps, if they're four, five, six years old, probably have more business strategy debt than they realise — features they give away for free that maybe are higher value, ways they do things that haven't been rethought in years.”
Business strategy debt accrues silently in apps older than four years
Business strategy debt is the less-discussed cousin of technical debt: features misclassified as free, pricing that predates current consumer behaviour, monetisation models chosen opportunistically rather than strategically. The older the app, the more this accumulates unnoticed. Even a growing app can unlock substantial upside from a single rigorous re-evaluation of the free/paid line and the underlying assumptions it rests on.
“your retention length will be really dictated by how long the user has the problem you're solving for...cell phone plans...I'll probably be with them 25 years because I have a daily need for their product...meditation apps where meditation is a thing you don't really need an app for you'll probably either like learn it or realize it's not for you relatively quickly”
Your Retention Ceiling Is Set By How Chronic The Problem You Solve Is
Dan frames churn not as a product problem but a use-case problem. Cell phone plans retain for decades because the need never goes away. Meditation apps churn quickly because the underlying need resolves — or dissolves. The highest-retention subscription products solve chronic, daily, irreplaceable problems. If your use case is episodic or short-term, you need massive acquisition volume or a retention mechanism (like a companion subscription or seasonal content) to counteract the structural ceiling.
“Everybody has access to Claude 3.7 but like you can bring something unique in terms of like brand and translation and things like that and that's what's going to drive people to convert and retain more than anything else.”
Everyone has the same AI model — brand and community are the real moat
API access is a commodity — any app can call the same foundation models. The differentiation that drives conversion and long-term retention sits in brand, community, and the specific translation layer you build between raw AI capability and a real user need. Vibe-coding a ChatGPT wrapper skips all three.
“The feedback loop was so close by like it was like you know you were in the app you can just poke over here to kind of give some feedback and that made it really nice because there was no friction people always gave really good feedback.”
Build for a community you live in — the feedback loop needs zero friction
When your users already hang out in the same spaces you do, qualitative research happens passively. Apollo's subreddit made feedback instantaneous — no surveys, no user interviews, no recruitment. Building for a community you belong to means the product-market feedback loop is always on.
“The feedback people gave over the years really shaped it into something that I think became more than just like my ideal of what a Reddit client was it kind of got shaped into this what the perfect thing for a community would be.”
Community feedback shapes the product into something you could not design alone
Apollo started as Christian's personal vision of a Reddit client but became something broader through years of community input. The product that shipped was a collective artefact shaped by thousands of voices — which is exactly why it commanded such fierce loyalty. A solo founder with an active community can build better than a team without one.
“If this is really going to unlock paid acquisition and some other things such as the marginal cost of AI features for us, then why not do it? It gives our marketing team so much more room to operate.”
Higher margins from pricing unlock the AI feature budget you could not previously afford
AI features carry real per-request costs that a flat $40/year subscription struggled to absorb at scale. The price increase to $80 did not just improve marketing economics — it funded the product roadmap. When every active user consumes more backend compute, unit economics either improve via pricing or erode via feature cuts. Pricing is now as much a product decision as a revenue one.
“So far the results have been exactly our experiment results, which hardly ever happens. Usually when you release something out in the wild it's a little bit different. That is a credit to our growth marketing team and our product team — they really tested this rigorously.”
Real-world rollout matching experiment results is the reward for rigorous testing
Most product experiments look messier in production than in a controlled test environment. The fact that Lose It!'s price doubling held up at full rollout is evidence of testing discipline: multiple platforms, multiple user cohorts, and enough duration to trust the signal before committing. Rigorous experimentation before a major price change is what lets you move decisively rather than nervously.
“The original Macintosh in 1984 was $2,500 — inflation adjusted that's like $7,200 today. It sold so poorly it was the impetus for driving Jobs out of the company. And yet. The Mac was the right idea for the future of personal computing. It needed at least four or five years before you're like holy shit this is a thriving platform.”
Vision Pro follows the 1984 Mac playbook: expensive first version building toward a mass platform
Apple's Vision Pro is not the iPhone — it is closer to the 1984 Mac: an expensive enthusiast device building a platform ecosystem years before the mass-market hardware is ready. The 1984 Mac was considered a flop; the press called the GUI a gimmick. Apple's patient pattern of replaying this arc means Vision Pro's weak early sales do not signal failure — they signal phase one of a long arc that developers building for the platform today can get ahead of.
“Early Android they just couldn't do it. There was no other platform. It was impossible to make anything that was like 'just look at it — just look at this app and look how cool it looks.' Nobody had the skill set.”
Mac-era design culture gave the iPhone App Store an ecosystem advantage no rival could match
The passionate Mac developer community of the early 2000s — Delicious Library, Panic, Omni Group, Rogue Amoeba — had spent years cultivating the craft of delightful native software. When the App Store opened in 2008 that skill set transferred immediately to iOS while Android and Windows Mobile had no equivalent talent pool. Platform quality at launch was not just about Apple's APIs — it was about the design culture a decade of Mac development had built.
“Definitely making sure that there's no paywall with that is important if you want to make hay off of organic or viral. The only ones I've ever had be successful are the ones that are like core to the product — you have to think about it early.”
Viral sharing features must be free and core to the product — paywalling them kills the loop
Every share-to-TikTok, export, or social feature that sits behind a paywall breaks the viral loop before it starts. One Second Every Day's monthly giveaway — where users share their month video for a chance to win — works because sharing is frictionless and free. The product insight: viral mechanics need to be designed into the core product early, not bolted on later as premium add-ons.
“We had TikTokers that influenced product changes — just the ability to flip and mirror their video — we made that change and people were really happy. We definitely listen to everybody on social about stuff.”
TikTok community feedback drove a product feature — mirror button shipped from user requests
One Second Every Day added a video mirror button after TikTok users kept requesting it to match a viral effect on the platform. The feedback loop — TikTok trends create product requests, product ships the change, users share the result, more users discover the app — is one of the most efficient development-distribution cycles available to small app teams.
“Very few people are going to be P95 on every single metric along the entire chain usually you're going to have a strength somewhere in the chain that you're the real outlier in that and that makes the business work even if you're below Benchmark on another metric.”
No company is P95 across every metric — find the one outlier strength and build around it
Tinder compensates for high churn by being best-in-class at conversion and multi-tier monetization. Duolingo ran below-average monetization for years while building an unassailable user base. Carter's prescription: be 'good enough' across most metrics (within striking distance of P50), and identify the one metric where performance is genuinely exceptional. Benchmarks help surface which lever is already winning.
“We came up across the use case of sleep podcast for sleep we were seeing lots of people using our app at night for sleep and it turns out podcasts can have a hypnotic effect.”
Mine your own app's usage data to discover the next market — behavioral signals beat brainstorming
Podcast App's data revealed that a significant cohort was using the app as a sleep aid — listening at night in a specific pattern. Siniawski didn't go looking for the next market; behavioral signals in the existing product pointed to it. This is a replicable discovery method: segment your retention and usage patterns by time-of-day, use case, and audience cohort to find unmet jobs-to-be-done hiding in your data.
“We have a full team that's dedicated to the value that our subscriptions are providing and they're building new features and enhancing adoption of features... and that team size is similar to the team that drives growth from better pricing discovery packaging promotions etc.”
Value creation team must be same size as growth team — one without the other creates extraction without substance
LinkedIn deliberately equalizes headcount between value creation (new features, adoption, functionality depth) and value capture (pricing, discovery, packaging, promotions). Most companies over-index on one side. Levit's observation: pure growth teams extracting value without continuous value investment will eventually exhaust subscriber goodwill and cause churn. This balance is structural, not aspirational.
“You don't come to LinkedIn for AI and the magic of AI is a bit ubiquitous right like what what is AI really... the goal is not to sprinkle AI everywhere and say we're here ai is us like what does that really mean for a human right.”
Don't brand your AI features as 'AI' — members come for outcomes, not for the technology
LinkedIn positions AI tools around member outcomes (find a job faster, write a better profile summary, surface the right leads) rather than labeling them as AI features. Levit's principle: AI is infrastructure, not identity. When a feature works, members credit the outcome — not the model. Sprinkling 'AI' everywhere dilutes brand and sets expectations that are hard to consistently meet. Build for outcomes; let the technology fade into the background.
“LinkedIn's mission is to provide an economic opportunity for every member of the global workforce and that opportunity is not paid... we wouldn't add anything that would ever make something that's related to uh finding a job or applying to a job a paid functionality.”
Free ecosystem health drives paid subscription growth — never paywall the core value prop
LinkedIn protects free job-seeking functionality as a structural commitment — it drives the supply side of the marketplace (candidates) which sustains the demand side (recruiters paying for Talent Solutions). Making job applications paywalled would kill recruiter value. Levit's principle for any freemium product: find the free experience that sustains the ecosystem, protect it vigorously, and build premium only on top of features that help users achieve more — not features that restrict baseline access.
“The thing I see most commonly is just a very chaotic approach to measurement. Your measurement model is essentially the heartbeat of the company. Everything flows from that. You need to be doing it correctly, in a way that's credible, and in a way that everyone understands.”
Measurement disorganization — not technology — is the #1 growth blocker
Seufert names measurement disorganization as the single most common failure he sees: competing tools no one knows how to reconcile, finance and UA with different definitions of success, and LTV models the product team rejected and rebuilt. The fix is a people problem — getting finance, UA, and product in a room and aligning on one operational model before touching any new technology.
“The real answer to 'what's the biggest opportunity for growth?' is just that your measurement doesn't support true growth. It's broken or flimsy. You're just trying to replicate what you've done in the past. If you don't believe your measurement can adapt to new channels, you just won't grow.”
Measurement paralysis: broken attribution freezes teams into replicating the past
Teams that don't trust their measurement framework default to running the exact campaigns they've run before — they can't evaluate anything new. Building truly robust, incrementality-focused measurement is the unlock for every other growth lever: influencer, OOH, CTV, podcasts. Without it, the growth ceiling is baked in.
“The best businesses found a really important need going unmet, invested resources in making the solution as good as it possibly could be, and then importantly were able to capture enough of that value on the back end to reinvest into the business and build that moat. Pandora's ad model was suboptimal for two reasons: worse user experience and less profit density. Spotify went all in on subscription and reinvested that revenue back into the core product.”
The Subscription Value Loop: value creation funds delivery which funds better value creation
Carter's framework distills subscription app strategy to three linked phases: create robust value, deliver it efficiently, and capture enough revenue to fund the next creation cycle. The loop is self-reinforcing — strong value creation earns better retention, better retention reduces CAC, lower CAC allows reinvestment in product. Pandora vs. Spotify is the canonical proof: inferior monetization starved product reinvestment and ultimately lost the market.
“Is it robust — do you have a value promise solving a real pain point? Is it delivered rapidly — can you get someone to the aha moment in the first 30 seconds? Is it repeatable — will users keep getting value month after month? And is it remarkable enough that people talk about it and build a community around it?”
Value creation needs four Rs: Robust, Rapid, Repeatable, Remarkable
Carter's four-part framework for evaluating the value creation step. Robust = real PMF (40% test). Rapid = aha moment in first session. Repeatable = ongoing utility that justifies recurring billing. Remarkable = organic word of mouth that compounds acquisition. Each R has its own measurement tools and common failure modes — and missing any one of them creates a predictable ceiling.
“Apple may clip off a couple percentage of their free users, but what they don't do is take the people that would pay for AllTrails and Flo. Those people do not just stay on Apple — they're trying to look for a premium experience. If you're an app that hasn't innovated, that's kind of just a me-too app sitting at the bottom of the rankings, that's the hardest spot to be.”
Build deep and verticalized — broad shallow platform features don't take paying users
Platform sherlocking is frequently overhyped. When Apple adds hiking maps or period tracking it attracts free users but almost never converts the high-intent paying audience — those users seek depth, community, and features no OS-level app will build. The defense is simple: verticalize and go deep. AllTrails adding social features, Flo moving into perimenopause, OnX adding layered maps — these are moats Apple will never bother to match.
“Consumer spend accounts for 70% of the US GDP. You know, OnX — people are spending tens of thousands of dollars a year as a hunter on these passions, these self-actualization, the love and belonging, the community. There's a lot of money going toward these. Consumer subscription businesses have an opportunity to enter those markets and capture so much more of that spend.”
Consumer spend is 70% of US GDP — the app market has barely scratched the surface
The total addressable market for consumer subscriptions is essentially the entire discretionary spend of the US consumer economy. Crowley argues most categories remain grossly under-penetrated: the analog value (in-person coaching, physical maps, print media) is still being captured digitally, and most passion-driven spending has no software subscription attached yet. The opportunity is not competition within apps — it is converting offline spend.