Founder Playbook · Sub Club by RevenueCat
16 tactics from Andy Carvell
Proven Growth and Retention Strategies for App Developers
Watch the full episode“The impact that you can drive with notifications is reach times relevance times frequency. Not all notifications are equal and the really killer ones that are going to supercharge your business have high reach, high relevance and high frequency — and then you're in that golden quadrant.”
Notification impact = Reach × Relevance × Frequency (the RRF formula)
Distilled from SoundCloud's 500M-pushes-a-month system, the RRF formula scores every notification on three axes: reach (opt-in rate × addressable segment), relevance (proxy: CTR), and frequency (cadence before opt-outs). Boost relevance via personalization and you can push frequency higher without bleeding reach.
“If it's highly relevant, users will not just tolerate but actually welcome a high volume of notifications. If you're able to increase relevance by personalization, they're much less likely to turn them off, and you can send more.”
Relevance is what users will tolerate frequency for — even welcome it
The unlock isn't 'send less' — it's 'earn the right to send more' by personalizing. Click-through rate is the proxy for relevance: when CTR is high, opt-out pressure stays low and you can scale frequency without burning the audience. That's how SoundCloud got to 500M pushes/month without nuking opt-ins.
“In the early days we would run experiments in places like Pakistan — we'd just not send them to Berlin where all the engineers were based. We'd get a feel for the uplift, then take it to the lawyers. Only when you come back with data showing retention moved 5 percentage points do they give you a bit more leeway.”
Test bold notifications in Pakistan, not Berlin — escape the engineer bias
SoundCloud's engineering team in Berlin was convinced aggressive notifications would destroy the product. Andy's workaround: ship bold experiments in rest-of-world markets the engineers didn't see, prove uplift with hard numbers, then earn permission to roll out. Internal taste is not user taste.
“CRM is not the most effective lever for engaging users — actually that's product. But CRM is a great way to circumvent a six-month product backlog and an engineering backlog and rapidly iterate on ideas. It's got built-in measurement and segmentation.”
CRM lets you circumvent a six-month engineering backlog
Product is the strongest retention lever, but customer-engagement platforms (Braze, Iterable, Leanplum) let growth teams overlay messaging and in-app experiences on top of the product without waiting on shipping. Day-zero cohorts refresh daily — even small apps can iterate every day or two on first-run flows.
“I went to unsubscribe from their email because they were sending me two a week. The unsubscribe flow said 'hey, do you want to hear from us twice a week, once a week, or once a month?' I said once a month. I wish more newsletters would do that.”
Give users frequency control instead of a binary unsubscribe
The Twelve South example: offering a frequency downgrade saved a high-intent customer who would otherwise have churned the channel entirely. Most teams ship a binary unsubscribe and bleed audience needlessly. Let users self-select cadence and keep the ones who actually want to hear from you.
“You have greater surface area or greater reach earlier on in the funnel because more users are still around. That's a big lever on how much impact you'll drive with any particular notification — the size of your target audience that's addressable.”
Reach is biggest at the top of the funnel — optimize there first
Optimizing the last step of a funnel feels sophisticated but rarely pays. The first or second touchpoint matters more because cohorts haven't decayed yet — same logic applies to which lifecycle moments deserve investment. Pick the high-reach moments and invest relevance there.
“Twitter's not real life. The developers and the communities and the styles we have as people who make software are very different from the median consumer — which is not even a thing. You're dealing with this huge distribution of users and they have different appetites for marketing.”
Twitter is not real life — developers are not the median consumer
Indie-Twitter consensus says push and email are evil. The actual user base of a consumer app is much broader and many users genuinely want to hear from apps they care about. Don't let engineer-tribe taste set your retention strategy — let the data set it, and remember there is no median consumer.
“Sometimes people misinterpret that as 'oh I have to do all of these things, I have to tick all of these boxes in order to be successful.' No, absolutely not. You have to play to your strengths and to your company stage and your priorities. If you try to overreach and do everything, you're going to do it all really badly.”
The Mobile Growth Stack is a menu, not a checklist
Andy's framework encapsulates everything that could form part of a mobile growth strategy — but its purpose is to help teams pick, not pile on. Stage-appropriate focus beats comprehensive coverage; early stages reward big swings on PMF, not sophisticated experimentation infrastructure.
“When you're early you take big swings that don't need sophisticated measurement to see the results. You don't need sophisticated A/B testing, you don't need sophisticated analytics — you need to take big swings that give you obvious results. As you grow you can start taking smaller swings that require more sophistication.”
Early stage: take big swings, no sophisticated measurement required
Each stage rewards a different size of bet. Founders who install Braze, set up A/B testing harnesses, and instrument funnel analytics before they have PMF are over-engineering the wrong layer. Pre-PMF, iterate the product weekly and watch raw cohort numbers — the big swings are obvious if they work.
“What an early stage app team needs to work on is product-market fit, whether they think they do or not. They're probably two or three years away from product-market fit. That means they need to iterate on the product and they need to iterate on the marketing and meet there somewhere in the middle.”
Most early-stage apps are 2-3 years from product-market fit
Andy refuses early-stage clients at Phiture because they arrive thinking they're ready for hyper-growth and almost never are (Flappy Bird is the rare exception). The honest expectation: years of product and marketing iteration until the retention curve flattens, not weeks of growth hacks.
“It all starts with understanding the user journey, and understanding users probably better than you currently do. That for me always starts with asking them questions rather than diving straight into the analytics and looking at funnels.”
Interview users before you open the analytics dashboard
Tech teams default to dashboards because the data is there. Andy's instinct flip: quant tells you what users do, but qualitative interviews surface what they're thinking, hoping, and expecting at each touchpoint — the why behind funnel drop-offs that analytics alone won't explain.
“At Elevate before we even launched we had an Android app — it was great because we could release a new version every week and do a beta list. We curated a list, we measured each cohort. We weren't A/B testing — we just watched and iterated on the app until we felt like that number was pretty good, and then we launched.”
Iterate weekly on Android beta cohorts before you ever launch (Elevate)
Pre-launch, Elevate shipped a weekly Android beta to a curated TestFlight list and watched raw cohort numbers like signups and activations. No A/B testing, no heavy tracking — just iterating until the funnel looked healthy before going public. Big swings, obvious results.
“A great lifecycle program starts outside of the app. It starts with a need or a want, then the user discovers the app and they're all going to go through the app store. You start the user journey there — your ad creative and your app store page set the expectation you then need to deliver on in the very first session.”
Lifecycle starts in the App Store — qualify users before they install
Lifecycle and onboarding aren't separate from acquisition. The ASO page sets the expectation; activation has to deliver on it in session one. Misalignment between the promise on the store page and the actual first run is the cheapest place to bleed retention — and the easiest to fix.
“Even though they were headquartered in Berlin, they built it global from the start. When I came in to help them level up on mobile, we made sure that we translated the app store page into all the languages — when I came in it was just in English.”
Localize the store page even before localizing the app
SoundCloud's organic growth lever wasn't a German-first launch — it was translating the store listing into every language so any market could catch fire. You don't know which market will resonate; remove the localization friction at the cheapest layer (the listing) before you invest in full app translation.
“In-app messaging is a killer channel for engaging users who are in the app. It bridges the gap between classic product and classic marketing because you can overlay and augment new experiences on top of what's built in the product. You can make them look super native.”
In-app messaging bridges product and marketing — iterate without a release
In-app messages let the growth team prototype product changes — onboarding tweaks, upsells, new flows — without a release cycle, while still feeling native. Day-zero cohorts refresh daily so even small apps with a few hundred downloads can iterate every couple of days on first-run flows.
“You can't really plan for virality but you can at least sow the seeds and see if they germinate. If you don't cultivate the right conditions for virality to occur it never will. If you have the ability to share content, put those share features in — then at least you can see the natural predilection of users to share.”
Plant the seeds of virality even if you can't plan for it
Andy's pragmatic take on viral loops: build the cheap share/referral surfaces during the growth phase so you can observe whether organic sharing happens. Even small amounts lower blended CAC; occasionally you'll catch fire. Don't engineer virality, but don't preclude it either.