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 6 of 10
“Not always do user needs that is verbally communicated actually match the behavior that you see in the product — and then looking at really where those drop-off points are all the way from acquisition throughout the subscription lifecycle.”
Stated User Needs Don't Always Match Actual Behavior — Use Both
Rachel Chukura warns against relying solely on surveys or interviews: what users say they want and what they actually do inside the product often diverge. The Weather Company triangulates across qualitative research, product analytics, and subscription analytics (via RevenueCat) to find where stated intent and observed behavior align — and to spot friction in the funnel that surveys alone would miss.
“We can reach statistical significance on an experiment within 24 hours in some cases — but we do recognize that buying behavior is often influenced by many other variables outside of very specific colors and text. Purchase behavior could be everything from what is the weather outside, what is my need right in that moment.”
Run A/B Tests Long Enough to Cover the Full Context Your App Lives In
At weather.com scale, experiments hit stat-sig quickly — but Rachel Chukura emphasizes that speed can fool you. Purchase intent is shaped by contextual factors (weather conditions, time of year, news cycles) that a single 24-hour window will not capture. Running tests long enough to see variation across conditions produces findings you can actually trust and build roadmap decisions on.
“Our big focus right now is data and experimentation — really creating a lot more rigor in how do we test and learn, whether it's the change in a color of a button or the length of our trial. Those are things that are really essential in kind of growth hacking the conversion rate and acquisition of new subscribers.”
Data and Experimentation Rigor Is the Core Growth Lever for Subscription Conversion
Rachel Chukura names data infrastructure and structured experimentation — not paywall design or pricing instincts — as the primary lever for improving trial-to-paid conversion. By investing in their data position first, The Weather Company can now run disciplined tests across the full end-to-end funnel. For any subscription app serious about conversion optimization, getting the data infrastructure right is a prerequisite for meaningful growth hacking.
“When you're pre-product market fit you should really just be focused on trying to really find the sweet spot of what you want to do and the product you want to build and an audience. You should not be really worried about what your cancel flow looks like or overthinking price.”
Pre-PMF: Relentless Product Focus Beats Any Lifecycle or Retention Tactic
Reid DeRamus warns early-stage founders against the seductive trap of optimizing cancel flows, discount offers, and win-back campaigns before finding product-market fit. If users churn in the first month at high rates, the problem is almost never the cancellation UI — it's the product. Investing in lifecycle retention tactics before PMF is a distraction from the only thing that actually matters at that stage.
“I'm having a fake conversation with a real person. So I'm having a conversation with this guy who was leaving me a one-star review saying I got this free for nine years and now I have to pay.”
War-Game the Uncomfortable Conversation Before Making the Aggressive Decision
Before making price changes or removing free access, Alex Prasad role-plays the entire customer conversation in his head — not just the angry review, but the full back-and-forth. If the company's side of that imagined conversation still feels reasonable, the decision is defensible. This mental habit of treating digital users like real people standing in front of you is his primary substitute for large-scale user research.
“We get so obsessed with well that's not scalable — but how do we know? Because we don't even know where we're going yet. Test the idea purposefully in an experimental way. What a great problem: this manual thing is working and now we just have to make it scalable.”
Build the Manual Process First — Scalability Is a Great Problem to Have Later
Alex Prasad's operating principle is to build manual, bespoke processes first and only invest in scalability once a thing is proven to work. Rejecting ideas because the path to automation isn't immediately obvious is premature optimization that kills potential. Rome wasn't built in a day — and most overnight success stories are built on years of unscalable learning.
“Changing the opacity of the X button on the paywall to 80% increases Revenue 20%. You're like okay why do I even try? I've built a long buildout feature that has very little impact even though it's beautiful.”
Test Big Swings First — X-Button Opacity Change Beat Months of Feature Work
Opal ran A/B tests on major features and minor UI details alike. The most jarring result: making the dismiss X button slightly less opaque on the paywall lifted revenue by 20%, while weeks-long feature builds moved nothing. This is an argument to ruthlessly estimate expected uplift before starting any build, and to run the quick-win tests first before long investments.
“For me now measurement is really about multiplying the sources of truth. You have to use different methodologies, different techniques to measure different KPIs and make sure that they all align and they all point to the same conclusions — that you're going in the right direction.”
Multiple Sources of Truth Beat Single-Channel Attribution Every Time
Post-privacy, no single attribution model is reliable. Deezer layers last-click MMP data, multi-touch attribution, incrementality geo-tests, brand lift studies, and media mix modeling — then looks for convergence across all signals. If only one tool shows a lift, it's noise. When three or four tools independently point the same direction, it's a real insight worth acting on.
“We ABC tested these pages against each other and none of the alternative versions won against the control version. And so yes, that's the moment that we all agreed to listen to our gut feeling — the control version was not conveying the messages that we wanted to convey and was not user-friendly at all. We implemented the branded version, and a few weeks after we saw positive results.”
When Every A/B Variant Loses, Ship the One That Serves the User Best
Deezer redesigned their logged-out homepage and ran a three-way A/B test — none of the new variants beat the control statistically. Rather than staying put, the team listened to a qualitative signal: the existing page didn't represent the brand or serve users well. They shipped the branded variant anyway. Within weeks, it outperformed the old control. Data tests can fail to detect real improvements, especially for brand-driven UX changes that take time to compound.
“When data is driven by a business strategy that informs a data strategy, things are incredibly effective and you're able to really extract a lot of value and learnings out of data. Companies that go out and hire a lot of data scientists but don't really know what they should be focused on — that can work but it so often doesn't.”
Business Strategy Must Drive Data Strategy — Not the Other Way Around
The most common data failure pattern: hire smart people, collect everything, and assume insight will emerge. Taylor Wells argues data only creates value when it's answering specific questions the business is already trying to answer. Start with the decisions you need to make, work backwards to the metrics that would inform those decisions, and only then build the data infrastructure to capture them.
“Catching something early on could be 1x of your cost and then 10x if you're fixing it by clean up downstream — because now you've got new data coming in that you'd have to continually add new iterative processes to fix — and then lastly at the last mile maybe 100x more cost if you have bad reporting that are actually telling the wrong things.”
Fix Data Architecture Early: Upstream 1x Cost, Downstream 100x Cost
Bad data compounds like debt. Taylor Wells worked on Disney+ from the ground up and saw this pattern at enterprise scale: a misnamed event or wrong tracking structure that costs an hour to fix at launch costs ten hours mid-product and a hundred hours — plus wrong business decisions — at the reporting layer. The principle applies to solo indie developers: spend half a day thinking through your event taxonomy before you ship.
“Instead of tracking 'download clicked, download started, download completed', if you're focusing generically you can say 'button clicked' and give keys to those buttons — ensure there's structure in how that's built. Now when someone comes and says 'we need to know when they hit download and then pause and then share' — well, either way those were clicks of those actions and I've mapped that behind the scenes.”
Build Generic Modular Event Tracking to Avoid Rewriting Your Data Stack Later
Disney+ built Glimpse — a generic click-stream event structure where every interaction is a 'button_clicked' with hierarchical keys, not a bespoke named event per action. When product added new features, the data infrastructure didn't break. Taylor Wells recommends this for any product: define a taxonomy of generic event types with keyed metadata, so new product features get tracked automatically without requiring new instrumentation.
“I would generally lean towards use off-the-shelf, use open source things like that early on. But with a clear evaluation structure around when you cut over to other things — and a culture that both acknowledges and accepts that it's okay that things may not be congruent over time when you switch.”
Start Off-the-Shelf but Pre-Plan Your Cut-Over Checkpoints Before You Scale
Most companies make the build vs buy decision once and treat it as permanent. Taylor Wells recommends making the decision with pre-agreed checkpoints: if we hit 10M users, we evaluate switching; if costs exceed X, we evaluate building. Disney+ had Adobe as a backup while building Glimpse internally — and within weeks of launch ($1K/day internal vs $33K/day Adobe), could confidently switch. Decision-making frameworks beat reactive pivots.
“I was tasked with writing an email each day by noon that just had three bullets — and up to two sub-bullets within each. Just an email that said: here are the three things that you should be focused on that happened the day before. And then suddenly dashboards become absolutely irrelevant frankly.”
The Best Executive Dashboard Is 3 Bullet Points in a Daily Email
Disney+ President Michael Paull found dashboards overwhelming and disconnected — too much data showing the same patterns without surfacing what mattered. Taylor Wells solved it with a daily 3-bullet email that synthesized the previous day's key signals and anomalies. The format forced prioritization: only insights that would change a decision made the cut. For small apps, the same principle applies — pick 3 metrics that actually drive action, ignore the rest.
“You have this content that is being consumed heavily by a subset of users and they are religious about it — but you've made it nearly impossible for them to pin that as a rewatch once they've completed it, or for them to find similar kid content. Your root problem is actually because of a broken process with how the profiles were set up.”
Buried High-Completion Content Is a Hidden Revenue Unlocking Opportunity
Bluey had near-100% episode completion rates and high rewatch frequency on Disney+ — but it was buried in a last row, 25 rows down, with no way to pin it after watching. Taylor Wells used completion data to argue for surfacing it prominently. The lesson: completion rate and rewatch rate reveal which content users would pay to keep accessing. If you find any feature with anomalously high engagement buried in your product, move it front and center before building anything new.
“My professor told me: strategically, you know who your customer is — they're the person that gives you money.”
Your real customer is whoever pays you — not whoever uses the product
Talking Parents initially targeted courts and judges as their customer, even sending the co-founder to courthouses around the country. Revenue only took off when they reoriented entirely around parents. The lesson scales universally: however logical the indirect channel looks, the person funding your roadmap should be the person you design for.
“We tried to build [subscription management] ourselves and it was a huge mistake. We were running around chasing our tail every time Apple would change something or Google would change something.”
Don't build subscription infrastructure in-house — the iteration cost is brutal
Vince's engineering team spent significant cycles rebuilding subscription logic every time Apple or Google changed their billing rules — cycles that could have shipped features. Once they moved to RevenueCat that overhead disappeared. The broader principle: any infrastructure that is not your core product differentiation should be bought, not built.
“We started having problems with our flagship feature. We had compared them to Twilio and decided to go with this new startup. The reality is they couldn't scale with us.”
Cheap third-party vendors that cannot scale with you become expensive emergencies
Talking Parents chose a startup video provider over Twilio to save money on their most important feature — accountable calling. When the vendor couldn't scale, their flagship product broke, forcing an emergency migration mid-growth. Vince's rule: always check whether your vendor can handle your ceiling, not just your current floor.
“What some founders do is get in a big hurry to get it to market. We made sure that what we built would scale and that we could change things out as we moved along.”
Design your codebase so individual vendors can be swapped without rebuilding
After painfully migrating from a failing transcription vendor to AssemblyAI (cutting costs in half and gaining features), Vince codified a rule: architect each integration as an isolated module. If you design the seams correctly from the start, replacing a vendor is a week's work instead of a quarter's crisis. It's the vendor equivalent of loose coupling.
“If it's an app that provides good value to users, almost everything else falls in place: they rate it well, they have affinity for it, organics are sustainable. It sounds boring but it gets back to the core.”
Acquire apps that already provide genuine utility — almost everything else follows
After 37 acquisitions, Michael's primary acquisition filter is simple: is the app genuinely useful? Apps with real utility self-generate good ratings, organic growth, and retention. He's not trying to engineer those outcomes — he's looking for products where they already exist as a natural consequence of value. The M&A screen is a product-quality filter as much as a financial one.
“Don't mess with something that's really good. We look to refine but the main proposition is already there and there's a lot of users that really enjoy it — we have to continue to deliver on that.”
When you acquire a great product, don't overhaul it — refine the edges
A common acquirer mistake is re-architecting a beloved product post-acquisition. Maple Media's rule: if the app already has loyal users, the product-market fit is proven. Day-one focus goes to implementing the Ivory platform (consistent infrastructure) and fixing presentation gaps, not redesigning the core. The value at acquisition is already there — protect it first.
“We have our own in-house platform called Ivory that allows us to manage all these apps at scale — the same codebase, same connections to analytics and advertising platforms — which is imperative when you're managing a lot of different products.”
Shared infrastructure across a portfolio removes the overhead of scaling individual apps
Without shared infrastructure, each acquired app becomes a separate engineering silo with its own analytics integrations, monetisation wiring, and reporting. Maple Media's Ivory platform standardises all of that so every app speaks the same metrics language from day one. This is the operational moat that makes running 37 apps with one team viable.
“You can model these out you can figure out if I have my retention is X and it goes down by 10% what does that do to my longtail Revenue right you can model these things out and so you just have to have a robust understanding of the data.”
Model Your KPIs: Retention, ARPU, and LTV Are the Dials — Not Individual Complaints
Before changing pricing or onboarding in response to feedback, model the downstream math. If retention drops 10%, what happens to 12-month LTV? If conversion rises 5% but retention falls 15%, is that a net gain? Consumer subscription operators who run these simulations first make far fewer reactive mistakes.
“We have given more than 20 million subscriptions like 20 million subscriptions we gifted more subscriptions than we because like now we have 6 million active subscribers maybe we have sold 10 million or so but we have given 20 million.”
Flo Gifted 20M+ Subscriptions in 50+ Countries — More Than It Has Ever Sold
In countries where women genuinely cannot afford a subscription — no credit card, or simply no disposable income — Flo unlocked the full product for free rather than leave 99%+ of users locked out. The company has gifted more subscriptions than it has ever sold: 20M gifted vs ~10M sold. There is no direct business return — only the mission alignment that makes employees and paying users more committed.
“they built a product that did very accurate gate tracking for riders... they used that one tool as a way to basically aggregate all of these equestrians in the world and then went deeper into their workflow... a safety tracking feature where basically when you turn that on if there's impact someone that you select is notified”
Wedge Deep Before Expanding — Equilab Went From Gait Tracking to Safety Alerts to Community
Start with one precise problem, get people to depend on it, then deepen into adjacent pain points within the same workflow — not copy what competitors do in other categories. Equilab's gait tracking gave them the equestrian audience; safety alerts made the app part of relationships; community tied users to a specific barn. Each step made the product stickier without changing the target user.
“for a content-based business i think there is a decision point that every content-based app needs to make which is are you a platform to enable other content providers to reach customers... the flip side being a platform is to be a content producer”
Platform vs. Content Producer — You Can't Do Both Without Diluting Your Moat
Content apps must pick a lane: Roku-style platform aggregating third-party content, or Peloton-style producer owning original content. Trying to do both forces heavy content investment while simultaneously ceding distribution to channels where users already are. Pick the model that fits your user relationship and build the moat that model enables.
“what's gonna really improve your business and kind of break through what i refer to as the carrying capacity of the business which is basically when you know churn equals new customers and you basically start leveling out to actually cut through that carrying capacity you need to drive product improvement”
To Break Through the Growth Ceiling, Drive Product Improvement — Not Just Better CRM
Every subscription business hits a 'carrying capacity' where churn absorbs all new growth. CRM tweaks, annual plan nudges, and better onboarding emails move the needle 5–10%. To actually break through the ceiling, you need product improvements — better content discovery, data accumulation fed back to users, or expansion into adjacent offerings.
“with these you know day seven row as campaigns and value optimization and event optimization campaigns facebook with all of its data and ai in an incredible targeting efficiency has kind of in some ways been doing the job of developers it's been finding those unique profiles user profiles of who's actually going to spend money”
Facebook Was Doing Your User-Profiling for You — Now You Have to Build It In-Product
For years, Facebook's targeting replaced in-product personalisation: describe your ideal user and Facebook found them via creative experimentation across hundreds of millions of profiles. Post-ATT, that black box is degraded. Developers must now rebuild in-product personalisation — show users different content based on observed behaviour — which is harder but more durable.
“if you've only you know grown via you know just sort of like organic traction and organic like magnetism and you've you've gone through like many sort of cycles of app or product iteration to sort of optimize the product for that group of people... then you've optimized for the group that's that at the greatest potential scale of your of your product is a minority”
Optimising for Organic Users Tunes the Product for the Minority at Scale
Organic users are self-selecting, high-intent early adopters who look very different from the mass-market paid users who represent 60–80% of daily installs at scale. If you've only A/B tested with organic users, you've tuned the product for a minority. Bringing in paid traffic early reveals what the majority of future users actually need.
“weekly active subscribers because we have a Cadence in there that someone is coming and doing this event... Cadence action Revenue so weekly active or this could be listening or this could be reading.”
North Star Metric = Weekly Active Subscribers — Cadence + Action + Revenue in One Number
Hannah's preferred North Star for subscription apps combines three signals: a time cadence (weekly/monthly), a core product action (listening, reading, booking), and a revenue gate (subscriber). Pure revenue metrics hide churn. Pure engagement metrics ignore monetization. Weekly active subscribers surfaces both — and tells you whether your hard-activated users are staying active enough to renew.