Price Power

16: How to Build a Subscription App in the AI Era w/ Alice Muir

Alice Muir, independent subscription consultant who's worked with Headspace, VSCO, Adobe, SoundCloud, and MyFitnessPal, explains why "higher engagement equals lower profit" is the new reality for AI apps, how to use strategic friction without choking activation, and why most consumers don't actually care that your app is AI-powered. What you'll learn: * How strategic friction worked for MyFitnessPal, and what they got wrong by gating barcode scanning too late * Why "protect the learning actions, charge for the outcomes" beats free-vs-paid debates * How to handle the 90% of installs that never subscribe when free users now cost real money * When hybrid monetization actually makes sense and when it just adds complexity * Why weekly subscriptions are a proxy for usage-based pricing on novelty AI apps * Why margin-qualified acquisition matters more than CAC alone for AI products * How Flibbo used persona tiers on the paywall to get users to self-identify their willingness to pay * Why a fitness app got a 6x paywall lift by removing strikethroughs, countdown timers, and stacked offers * What the Subscription Stack framework needs to add for the AI era * The back-of-the-napkin math founders should run before shipping any AI feature * Why consumers may actually be turned off by "AI-powered" positioning Key Takeaways: * Protect learning actions, charge for outcomes. Users should learn what your product does for free. The thing that completes their job is what they pay for. * GPU cost is a CAC line item, not a margin problem. If everyone you acquire gets one free generation, that compute cost belongs in your acquisition budget. Run worst-case-scenario math before you ship. * Highest engagement is now lowest profit. Traditional subscription thinking inverts when each interaction has a real cost. The Subscription Stack engagement layer needs a full revision for AI apps. * Weekly subscriptions are a usage proxy. AI apps see strong initial conversion and terrible retention because users have high intent for short bursts. Annual plans for a song generator are a fantasy. * Self-identification at the paywall beats the questionnaire. Flibbo put basic, pro, and max personas on the paywall. Users picked the one that matched their use case. * Simpler paywalls outperform copycat paywalls. Stripping countdown timers, strikethroughs, and stacked plan tiles from a fitness web-to-app funnel produced a 6x lift. * Consumers don't care about AI as a feature. They care about the outcome. "AI-powered coach" can read as cheap, not premium. Lead with benefit, not technology. * Don't add AI just to add AI. If a feature doesn't measurably improve retention or activation, you're paying GPU costs to compress your margin. Links & Resources * Alice Muir on LinkedIn: https://www.linkedin.com/in/alicemuir/ * Subscription Stack framework: https://phiture.com/resources/subscriptionstack/ * Andrew Chen on consumer reactions to AI: linkedin.com/posts/andrewchen_when-consumers-dont-care-that-youre-building-activity-7358342997639360512-wqKM * Thomas Petit RevenueCat article on hybrid monetization: https://www.revenuecat.com/blog/growth/ai-hybrid-monetization/ Timestamps   00:00 – Intro 01:25 – Baby raves in Berlin and the new May Day 02:58 – How the playbook changed: from acquisition-first to retention-first 07:25 – Strategic friction and the MyFitnessPal example 10:43 – Hard paywalls vs letting users discover value 11:55 – Protect learning actions, charge for outcomes 17:25 – The 90% problem: monetizing low-intent users 21:36 – When hybrid monetization actually makes sense 26:15 – Apple tax, GPU costs, and the AI app profitability squeeze 28:55 – Why weekly pricing fits novelty AI apps 33:53 – Margin-qualified acquisition for AI apps 39:08 – Flibbo's self-identifying paywall personas 43:55 – The 6x paywall win: stripping out the fluff 47:56 – Revisiting the Subscription Stack for the AI era 51:55 – Switching models to protect margin 53:38 – What founders should get right before adding AI 56:58 – Hot take: consumers don't care about AI

15: How to start with Signal Engineering w/ Shumel Lais

Shumel Lais, co-founder of Day30 and previously founded Appsumer (acquired by InMobi), explains why most subscription apps feed ad platforms the wrong goal, how precision and recall reshape signal selection, and what a realistic measurement maturity ladder looks like in 2026. Shumel walks Jacob through the five stages of measurement maturity, from apps that just compare App Store Connect revenue to ad spend, through MMP attribution and cohorted reporting, up to incrementality testing for the largest spenders. He breaks down why signal engineering only makes sense once you have the right foundation in place, shares the 10-conversions-per-campaign-per-day rule of thumb for when to go further down funnel, and unpacks the restaurant booking app mistake that first put him onto the precision/recall framework. What you'll learn: * Why optimizing to cost-per-trial leaves money on the table for most subscription apps * How Meta's 7-day visibility window forces the signal engineering problem * Why recall, not precision, is the metric most marketers overlook * Why the restaurant booking app example was Shumel's own mistake, and what it taught him * How Meta's event-day reporting can hide renewals inside new purchase counts * Why server-side events struggle more with matching than client-side events * How to decide between revenue-value signals and binary convert/no-convert signals * Why subscription apps are years behind gaming on analytics maturity * The 10 conversions per campaign per day floor before attempting signal engineering * When LTV curves become reliable enough to extend payback from 30 days to 6+ months Key Takeaways: * Signal engineering is closing the gap between what the platform can see and what you actually care about. Meta sees 7 days. You care about month 3 revenue. * Recall is the metric most teams forget to measure. Precision tells you if the users firing your signal convert. Recall tells you what share of your actual converters it captures. A signal with 90% precision and 40% recall tells the algorithm that 60% of your good users are bad. * There are five levels of measurement maturity, and most apps skip steps. ASC comparison → platform attribution → MMP → cohorted reporting → incrementality. Signal engineering is a level 3 or 4 exercise. Attempting it earlier wastes the effort. * The 10-conversions-per-campaign-per-day rule. Below that, Meta cannot learn from a more selective signal. Above 30 to 40 per day, you are leaving performance on the table by not going further down funnel. * Meta reports on event day, not install day. Renewals fire as purchase events, so Meta can claim credit for users who were already paying. Without install-cohorted MMP visibility, you are paying to acquire users you already had. * Speed of signal affects matching quality and algorithm learning. Events sent within 24 hours have more matching parameters, and they let Meta decide if a user is good without waiting 7 days for the purchase to come through. * The restaurant booking app was Shumel's own mistake. Before Day30, he optimized toward behaviors that correlated with bookings but were not causal. Performance did not move. The fix was cohorts, observation windows, and a binary prediction statement. * Measurement problems are not an excuse anymore. In 2026, the tools exist and the playbooks exist. Hiding behind attribution gaps is a choice, as is hiding behind blended CAC when direct CAC is uncomfortable. Links & Resources * Day30: https://day30.ai [https://day30.ai] * Shumel Lais on LinkedIn: https://www.linkedin.com/in/shumellais/ [https://www.linkedin.com/in/shumellais/] Timestamps 00:00 Shumel's background and early mobile agency days 00:56 The signal engineering framing and how Day30 landed on it 03:30 A basic example: trials vs trials plus behavior 05:56 Why signal engineering exists (attribution gap, not just subscriptions) 08:45 Signal volume as the second dimension after precision 09:30 Defining recall and the photo storage app example 15:58 When to send revenue values vs binary convert/not-convert 16:41 The restaurant booking app mistake and causation vs correlation 19:33 Experiments are still the only real proof 20:00 Measurement maturity level 1: no MMP, just ASC 22:37 Do you actually need an MMP to start? 23:39 Level 3: why MMP matters (Meta's event-day reporting trap) 25:37 Level 4: cohorted metrics and aligning on day-30 ROAS 26:30 Level 5: incrementality and MMM for the largest spenders 27:35 The 10 conversions per campaign per day threshold 29:30 Why the MMP matters for signal engineering (measurement, not the signal itself) 31:03 MMP vs Conversions API for sending signals 33:04 SDK vs server-side: matching and speed 36:43 Payback periods and when to extend them 40:32 Simple inputs for a basic predictive LTV model 42:52 If you're running Meta to CPT today, what do you change first 44:41 The quantity vs quality of signal tradeoff 46:48 Hot takes: no more hiding behind attribution 48:02 Favorite pricing and packaging tactics seen recently 50:08 Day30's free signal audit offer

14: Fix Activation Before Growth w/ Daphne Tideman

Daphne Tideman, growth advisor and consultant for subscription apps, explains why most retention problems are actually activation problems, how to distinguish vanity activation metrics from ones that predict real retention, and why the aha moment should start in your ads, not just your product. Daphne walks through her evolution from treating activation as a simple funnel step to seeing it as a layered, behavioral process spanning the first 7 to 30 days. She shares real examples from growth audits where onboarding completion rates looked great but users vanished by day two, and breaks down the "time to first value" vs. "time to core value" framework for thinking about activation in stages. She also makes a case for monthly subscriptions as a faster learning tool for startups, and explains why revenue is a terrible North Star metric. What you'll learn: * Why onboarding completion is often a vanity metric that hides activation failures * How to identify whether your retention problem is actually an activation problem * Why "any action vs. no action" comparisons overstate the value of weak activation metrics * How to build mini aha moments into onboarding before the paywall * How to use the "time to first value" vs. "time to core value" framework * Why monthly subscriptions can help startups learn faster about activation * How to test whether an activation metric is predictive or just correlated * When user interviews beat quantitative analysis for defining activation * Why extending onboarding can drop completion rates but improve retention * How to diagnose activation vs. retention vs. acquisition problems * Why revenue as a North Star metric leads teams to extract value instead of create it Key Takeaways: * Onboarding completion is a vanity metric. An app had over 90% onboarding completion on both platforms, but most users were gone by day two. The onboarding was too short and easy to click through. When they extended it and built in value-delivering steps before the paywall, completion dropped but retention improved. * Your retention problem is probably an activation problem. For most apps, losing users in the first 30 days isn't a retention failure. It's an activation failure. Daphne argues we even mislabel it: "day two retention" and "day seven retention" describe periods when you're still activating users, not retaining them. True retention problems show up when users were active early but trickle off later. * Activation should start in the ad. Showing the job to be done and the transformation in your ad creative builds trust before users even open the app. A coding app's best performing ad showed someone coding in a lift, making viewers think "I could find time for that too." * Correlation isn't causation in activation metrics. Any action will always look better than no action. The real work is finding which behaviors, at what volume and timing, predict retention across cohorts and channels. * Mini aha moments beat one big moment. Instead of trying to engineer a single big aha moment (which is often technically difficult), build multiple smaller moments of perceived value. These can be as simple as a personalized plan, a visual showing the outcome, or a first small win before the paywall. * Monthly plans help you learn faster. For startups without much data, monthly subscriptions force users to make a renewal decision every month, which generates faster signal on who is truly activated vs. who is coasting on inertia. * Revenue is a terrible North Star metric. It pushes teams toward extracting value from users rather than creating it. Activation and usage metrics better align the team's incentives with user outcomes. Links & Resources * Daphne Tideman's Growth Ways newsletter: https://growthwaves.substack.com/ * Daphne Tideman on LinkedIn: https://www.linkedin.com/in/daphnetideman/ 00:00 Intro and Daphne's path from e-commerce to app growth consulting 01:20 How activation thinking evolves from 2D to 3D 04:20 Common activation mistakes: oversimplifying and picking the wrong metric 05:50 Why standard metrics weren't predicting retention 07:20 Onboarding completion as a vanity metric: 90% completion, gone by day two 10:20 Activation vs. monetization: which to fix first 13:20 Building mini aha moments into onboarding and ads 17:50 User interviews and the role of emotions in activation 20:20 Your retention problem is actually an activation problem 23:20 Time to first value vs. time to core value framework 27:20 How to test whether an activation metric is real or vanity 29:20 Starting with user interviews vs. data when you lack scale 31:50 Correlation vs. causation: finding the right activation threshold 34:20 Learning from failed experiments 36:50 Diagnosing activation vs. retention vs. acquisition problems 39:20 Why activation problems are more common than retention problems 42:20 Matching subscription models to use cases 44:50 Biggest activation mistake apps make right now 45:50 Lightning round: pricing wins, hot takes, and best activation results

13: The Four Horsemen of Churn w/ Dan Layfield

Dan Layfield, author of Subscription Index and former product lead at Codecademy and Uber Eats, explains why churn is the silent ceiling on subscription growth, how to diagnose which type of churn is killing your business, and the pricing trick that can double your LTV overnight. Dan walks through his four horsemen framework: payment failures, activation issues, pricing and plan mix, and voluntary cancellation. He shares the bottom-up optimization approach he uses with every company, starting with Stripe settings that take 10 minutes to fix. What you'll learn: * Why your Stripe retry settings are probably wrong and how to fix them in 10 minutes * How to calculate your growth ceiling using churn rate and acquisition numbers * Why payment receipts might be reminding users to cancel every month * How to price annual plans based on your monthly retention data * How to build cancellation flows that save 20% of churning users * Why activation experiments are tricky and often produce duds * Why quality problems are the easiest growth fixes Key Takeaways: * Churn dictates your ceiling. New users divided by churn rate equals your max subscribers. 1,000 new users with 20% churn = 5,000 subscriber ceiling. Lowering churn raises that ceiling proportionally. * Start at the bottom of the funnel. Stripe settings, dunning emails, card updaters can be fixed in minutes and win back 5% of churn. Do these before tackling bespoke activation problems. * Annual pricing should match monthly LTV plus one or two months. If average retention is five months, price annual at six months. Looks like a steep discount but doubles LTV. * Turn off monthly email receipts. Netflix, Spotify, and Amazon don't send them. That monthly reminder is a monthly prompt to cancel. * Cancellation flows should solve the underlying problem. Pausing works when the need is temporary. Downgrading works when they're paying for unused features. Links & Resources * Subscription Index: https://subscriptionindex.com [https://subscriptionindex.com] * Dan Layfield on LinkedIn: https://www.linkedin.com/in/layfield/ Timestamps 00:00 Intro and Dan's path from JP Morgan to Codecademy 04:00 Freemium conversion benchmarks: sub-1% vs. good (3%) vs. great (7%) 06:30 The growth ceiling formula 08:00 The four horsemen of churn 12:00 Bottom-up optimization: start with Stripe settings 13:30 Cancellation flow tactics: pause, discount, upgrade/downgrade 19:30 Payment failure quick wins: smart retries, card updater, dunning emails 22:30 The annual pricing trick that doubled LTV at Codecademy 30:00 Activation and the Reforge framework 37:30 Onboarding should show value, not just explain device setup 42:30 Ethical cancellation flows and click-to-cancel legislation 49:30 Screenshot audit: where to start when you're stuck 52:30 Turn off monthly receipts: the easiest churn win 53:30 Lightning round

12: Price Testing for Subscription Apps with Michal Parizek

Michal Parizek, pricing and growth lead at Mojo, explains how to predict long-term revenue from short-term price test data, why Apple's automatic regional pricing is wrong for most apps, and how to sequence pricing, packaging, and paywall tests for maximum impact. Michal walks through the 13-month revenue projection model he built at Mojo, which uses seven-day cancellation rates as a proxy for annual renewal rates. He shares how his team raised yearly prices by 50% in the US and Germany with minimal conversion drop, how they tested free trial lengths and found almost no difference between three-day and seven-day trials, and why the ratio between monthly and yearly plan prices matters more than the absolute price point. What you'll learn: - How to use seven-day cancellation rates to project 13-month revenue - Why Apple's exchange-rate-only pricing leaves money on the table - How to sequence price tests: price first, then packaging, then paywall design - Why the monthly-to-yearly price ratio drives plan share more than absolute price - How hiding the monthly plan pushed yearly share from 60% to 80% - Why free trials still matter for new users, despite advice to remove them - How three-day trials performed as well as seven-day trials at Mojo - Why your first price test should have big price gaps, not small ones - How traffic source mix can distort price test results - Why a 100% price increase was a short-term winner but long-term loser Key Takeaways: - Seven-day cancellation rate is a reliable early signal. 20-30% of cancellations happen in the first seven to ten days. Measure that rate per variant, project renewal rates from it, and you can evaluate a price test without waiting months. Mojo validated this against real data and it held. - Apple's regional pricing is just exchange rate math. No purchasing power, no local context. Look at your top five markets individually, compare conversion funnels by country, and cross-reference competitor pricing. - Pricing and packaging beat paywall design in impact. Changing price points, plan structures, and introductory offers had more effect than design or copy. Start with pricing, then plan mix, then layout. - The monthly-to-yearly price ratio drives plan selection. Changing only the monthly price shifted yearly subscriber share significantly. The perceived deal relative to monthly is a strong behavioral lever. - Don't remove free trials for new users without testing. Mojo tried it based on popular advice and saw revenue decline. Test it for your app. - Start price tests with big jumps. Test $40 vs $60 vs $80, not $50 vs $48 vs $52. Find the zone first, refine later. - Revisit cohorts months after shipping. Mojo's 100% price increase looked great short-term but cancellation rates spiked. The 13-month projection caught it. Links & Resources - Michal Parizek's Botsi blog post: https://www.botsi.com/blog-posts/pricing-experiments-the-backbone-of-mojos-monetization-success - Michal Parizek on LinkedIn: https://www.linkedin.com/in/michalparizek/ Timestamps 0:00 Intro 1:03 Using seven-day cancellation rates to predict 13-month revenue 3:25 Building the report template and data pipeline 6:13 Validating the renewal rate prediction model 10:03 Benchmarks for new apps without renewal history 12:09 Why Apple's automatic price tiers are wrong 13:33 How to research and set regional prices 17:10 Relationship between pricing, packaging, and paywall design 21:15 Sequencing: price first, then packaging, then design 23:55 Why paywall layout tests that touch plan visibility are most impactful 26:41 Free trial strategy and length testing 31:03 Paid trial options as an emerging trend 33:16 The biggest mistake: not having enough data volume 35:56 Raising prices 50% in the US and Germany 38:46 Start with big price gaps, refine later 40:11 Don't be afraid to test prices