What it Takes to Succeed with Paid User Acquisition — Thomas Petit, App Growth Consultant

Client set a rule before working with Thomas: any channel that cannot return at least 2x straight from the money spent is not run. Thomas warned it would shut off many channels; on the one channel he manages (Apple Search Ads) they reach the goal but cannot scale beyond a certain point. Same ceiling seen with Google Ads for another client.

A marketplace had many sellers and no buyers in one category and the reverse in another. Paid acquisition was aimed specifically at rebalancing those categories, judged on keeping existing users from churning rather than on direct return.

VC-backed startup that knew paid would be its main channel started UA before building an organic base. Goal set up front: at least 500 new users per day to iterate on onboarding and paywall and get retention data, with an accepted loss budget and a three-month deadline to learn retention, which messaging and creative work (for ads and the store listing), and to build the team, metrics and tools for scaling. Organic was built in parallel.

Bootstrapped, very profitable app with strong ASO. Model: take revenue from 7-day free trial conversions, add an assumed 40% renewal of annual subscribers at year one, and allow paid UA as long as it is confidently profitable by day 366. Later renewals (60-70% of renewers) are upside. Relies on having enough cash in the bank to finance acquisition before money comes back.

Size budgets by event volume: each campaign needs roughly 10-20 optimization events (e.g. trials) per day. At ~$500/day (~$15K/month) a US subscription app might get 100-200 installs and 10-20 trials. So aim for five-digit monthly budgets, about $10-20K per platform, and optimize for downstream events (trials) rather than installs, clicks or impressions, because platforms keep high-converting inventory for advertisers optimizing on deeper events. Narrow manual targeting on Meta at ~$20/day is technically possible but penalized with poor delivery and high prices, and privacy thresholds leave both you and the platform blind. $50K/month is where 2-3 channels and budget splits become possible.

Three checks before any paid UA: (1) the real goal, dug into with repeated 'why' - minimum return per dollar and time frame, or non-return goals like velocity, critical mass, market share or marketplace balance; (2) cash availability - only gamble what you can afford to lose, since learning periods and payback delays are common; (3) current revenue per install by country and platform, noting the source mix (featured traffic lowers it, high-intent search traffic inflates it). Rules of thumb: under ~$1 per install in the US on a subscription app won't work; $0.20 has no chance; ~$1 can work for some games or cheap Android markets like Pakistan, Indonesia, Egypt.

With $5-10K to test, skip algorithmic platforms. Start with Apple Search Ads (no optimization after the install, so small budgets work) and influencers (can buy $100 deals). Often before the app is ready, run Google web search ads on chosen keywords with $100-500 budgets and a zero return goal, sending traffic to a simple standalone landing page built in a builder like Unbounce or Instapage (Google forbids sending these campaigns straight to the store) that explains the app and redirects to the store. Use the learnings for ASO, imagery and ad messaging.

Measurement depends on the channel: influencers need little (a link system); Apple Search Ads uses its free attribution API (built into MMPs and tools like RevenueCat); Google Ads works best with Firebase; Meta via its SDK or an MMP. Get an MMP from the start if the app will clearly run multiple channels at scale, so data is unified and new networks (some, like TikTok and formerly Pinterest, require one) plug in without custom work. Always double-check platform-reported results against App Store Connect / Play Console analytics and in-house product analytics (Amplitude, Mixpanel, Firebase).

Treat blended metrics as the only truth: track the trend of blended cost per subscriber and blended installs, trials and paying subscribers as spend goes up or down, not the absolute value. Run the analysis per country + platform segment (e.g. Brazil Android, US iOS), bundling small countries into groups by region or similar conversion. Clean out days with featuring, seasonality or product changes. Make budget allocation decisions from blended trends and channel-level tweaks (creative, ad group) from network-reported data. At small scale do it by pivoting in Excel or Looker Studio; at larger scale use MMM tools like Meta's Robyn, Google's LightweightMMM, Incremental (used 1+ year), Recast or metric.works.

Don't test 20 variants of one concept. Test radically different concepts, even unpolished: person vs animation, provocative vs bland, strong colours, lots of text, long vs short, open on the app vs on a person. Put ~20 ads in and let the platform pick: the ad getting most spend after 100-200 impressions is the winner (right ~99% of the time). Where a concept gets traction, iterate: reorder scenes, lengthen, swap the person, change one element at a time. Keep off-brand winners running but try to realign them with brand messaging. Don't trust intuition about which ad will win.

Use several production sources at once: an in-house designer, multiple creators, agencies, and founders shooting selfie videos on an old phone (often the best performer at several companies). No single person's range can keep finding winners. Run broad targeting (country, 18+, iPhone) and let the variety of creatives find different audience pockets; never give all creative to one person or one creator.

Ads that just say what the app does rarely stop the scroll. Find what the audience is interested in (e.g. funny or hard facts about rain for a weather app) and use it as the hook in the first 1-3 seconds, then transition to the product, even abruptly or with an end card. Always make it clear it's an app so the store visit isn't a surprise. Avoid clickbait that drives engagement but not installs: platforms eventually punish ads whose impression-to-install-to-trial ratio is poor.

Motivation app (Monkey Taps) ran an ad that was just a screenshot of one of its notifications, showing it was an app without explaining why it was better. Thomas tried the same trick for other apps with similar value props and never made it work.

Blinkist ads showing a famous CEO and claiming they read huge numbers of books, making viewers want to be as smart and successful, then transitioning naturally to book summaries as the easy solution.