Scaling Your Subscription App with Meta Ads – Marcus Burke, Independent Consultant

At InnoGames from ~2012-2013, scaled Meta by jumping on each new product: app event optimization, then value optimization. With these on, no manual targeting was needed; pushed out many creatives and let Meta find the buyers.

Run one campaign with one ad set rather than e.g. 10 campaigns x 50 ad sets, so all signal trains one algorithm. Add only the splits you need (e.g. by country, since the US is worth more than Brazil). Add campaigns/ad sets as spend grows. This also helps pass the SKAN 3 privacy threshold of 88 installs per day per SKAN campaign ID (Meta may create hidden IDs per placement). Don't layer interest targeting.

Pick the conversion event closest to business value; for a subscription app that is the trial start. Put the trial in onboarding so it fires within the SKAN window, and make it the highest-value event in the SKAN conversion schema. Delayed-monetization products (e.g. strategy games) struggle because big purchases fall outside the window.

SKAN 3 postbacks arrive 2-3 days late, so for the first two days a new campaign may show e.g. $5K spent and zero installs. Match postbacks to the spend from 2-3 days earlier, and wait at least a week before evaluating so the algorithm has had data to work with. Don't shorten a trial (7 to 3 days) just to get faster reads if the longer one converts 5-10% better.

Start broad otherwise: choose core markets, no interest targeting, but exclude users under 25 (sometimes under 30, depending on the app). Under-25s have low CPMs and give cheap trials, so Meta loves spending on them, but trial-to-paid is very poor. Broaden to 25+ or lower only once you have conversion data by age. Older users convert better because they have more money.

Start with manual campaigns: separate ad sets per country, younger age groups excluded. Be skeptical of 'recommended' Advantage+ automation, which maximizes Meta's reach and works best when the conversion event is close to revenue (a purchase, not a trial). Later in the account's life, layer on fully automated campaigns with a lower cost-per-trial target to reflect the lower conversion of the broader audience.

Don't compare creatives only on overall cost per trial (e.g. €10 vs €15); break results down by placement, age and gender in Ads Manager. Audiences differ: Instagram Reels is younger, lower intent and TikTok-like; Instagram Feed is higher quality but pricier; Facebook is older with more purchasing power. Use each app yourself and make ads that match that placement's native content.

Use Meta as the experimentation channel for angles, copy and styles, then carry winners into the rest of the funnel. Often as simple as reusing the winning ad as the App Store preview video or a screenshot, so people who click through see they landed in the right place.

If the best audience is e.g. women 35-44 on Facebook with a certain creative, tailor onboarding: show reviews from that demographic. If you drive UGC-style Reels traffic (portrait, <30s, person talking to camera), it skews young and low-intent, so a high price like $80-90/year will struggle; sync paywall and pricing with the placements and ads you run.

Because Meta passes no user-level data and SKAN often underreports ~50% of trials, add onboarding questions: age, gender, and 'Where did you hear about us?'. Use answers as samples, not one-to-one attribution (users name their first touch, e.g. a friend). Compare value by channel (Meta vs TikTok vs organic) and by age band against Meta's age/gender reporting, e.g. if 45-54 is 5x more valuable than 18-24, check whether the 18-24 cost per install is really 5x lower.

Three inputs: (1) incrementality: switch ads on/off and watch the baseline (e.g. $2K/day on Meta lifts baseline 50%), easiest for small apps with little organic; (2) the ATT opt-in sample matched to channels; (3) the 'Where did you hear about us?' survey. From these, set cost-per-trial targets per channel (e.g. Meta worth 2x TikTok) and by country and age. Then check actual payback against the model (e.g. at $50K/month, is payback really day 8?) and revisit after ~3 months.

You don't need an MMP to run Meta: implement the Meta SDK, set the SKAN model there and track the trial event. For a small developer, extra tooling is another breaking point and becomes a cost center once free tiers run out. Use subscription analytics (e.g. RevenueCat) for trial-to-paid by country to feed the model.

Never upload untested creatives into the scaled campaign: any change resets the learning phase and causes swings. Run a separate testing campaign that mirrors the scale setup: same optimization event (trial, not install) and one of your live core markets (the cheapest of them), not cheap unrelated markets like Thailand or Mexico where nobody knows the brand. Move only proven winners into the scale campaign. Fewer, more deliberate creative bets than pre-ATT.

Use the placements yourself (including Facebook, which skews older with more purchasing power) and study what non-ad content people engage with, not just competitor ads. Follow a broad spectrum of news sites outside your bubble on Instagram and Facebook (e.g. Fox News) to see headline styles, how clickbaity they are and image styles, then tailor ads to feel native.

Aggregated Event Measurement gives instant, modeled (IP-matching) postbacks and helps high-CPI advertisers who can't pass SKAN privacy thresholds. But it's a black box Apple could shut down. Keep the account foundation and evaluation on SKAN, which is future-proof and improving (SKAN 4 lowers the threshold to ~20 installs/day and adds two more postbacks), and use AEM campaigns for extra scale when they perform.

Meta is usually one of the first channels to launch: it has volume and represents the audience you'll target later, and good creative is a competitive edge; it also tracks iOS better than Google Ads. Apple Search Ads buys high-intent traffic (someone searching 'weather app'), useful for validating funnel and pricing without worrying about ad quality, but it's competitive and doesn't scale.

Posts tactical Meta ads, web-to-app, paywall and onboarding advice on LinkedIn daily; plans to add digital products such as an email course.