Lukas App Growth

📈 Meta ads for subscription apps
💰 $20M managed · from -$15K/mo to $60K/mo
👇 Book a call

[Lukas App Growth @LukasAppGrowth]

@justddev one thing to watch: onboarding usually moves how many people reach the paywall, not the paywall's conversion itself. so read it as 'did more people get to the paywall', not just paid conversion, otherwise a real win upstream can look flat if you only stare at the final number

[Lukas App Growth @LukasAppGrowth]

@letannacook they copied the copyable half (idea + wording). the half you're building right now, an audience watching you build, is the one they can't clone. devs with your exact app and no distribution still lose to you. the build-in-public IS your moat

[Lukas App Growth @LukasAppGrowth]

@nicktheriot_ simple wins because it forces you onto the one thing that actually scales, the message. the 'smart' guy pours his iq into settings, which is exactly the noise you're describing. it's not simple beats smart, it's message beats settings

[Lukas App Growth @LukasAppGrowth]

smart bet, quarterly + annual only makes annual the obvious pick since there's no monthly to make it look expensive. one thing to watch though: you're not just changing price, you're raising the commitment gate at the exact moment cold paid traffic is least sure. your conversion rate on high-intent users probably won't budge, but you might quietly lose the impulse buyer who'd have started monthly. watch new subs per 1000 installs, not just conversion rate, that's where the loss (or the $8k) actually shows up

[Lukas App Growth @LukasAppGrowth]

the 'right format = virality' part is real. the 'virality = conversion' part is where it breaks for most people, it worked for you because the looksmaxxing format's audience happens to BE your buyer. the formats that go most viral usually convert worst because they pull everyone, not buyers. it's not virality = conversion, it's virality among the right people = conversion. you nailed the match, most who copy this get the views and none of the revenue

[Lukas App Growth @LukasAppGrowth]

the weekly is a smart anchor, $6.99/wk makes an annual look like a steal, but you're undercutting your own move by dropping the annual to $29.99. the whole point of adding an expensive weekly is it lets you hold or RAISE the annual, not lower it. also you're testing two changes at once (billing period + annual price), so a winner won't tell you which one moved it. i'd test the weekly against the $39.99 annual, not a discounted one, and let the anchor do its job

[Lukas App Growth @LukasAppGrowth]

the number that jumps out is buried in your own screenshot: 4m downloads, under $5k. that's the real lesson and it's the opposite of 'build a game'. riding a trend clearly solves distribution, but 4m people wanted it and almost none paid. downloads aren't the business, 'build a game' gets you installs not revenue. the hard part just moved from getting users to making a casual player actually pay

[Lukas App Growth @LukasAppGrowth]

the 'for app' list is spot on. one cut on the marketing list though: people, locations, gender, ages, ethnicities aren't separate tests anymore, they're outputs of the angle. you test the angle and the pain point, the right people fall out of who leans in. testing demographics directly in the ad set just splits your already-thin app signal. shorten it to angles + pain points, the creative does the rest of that list for you

[Lukas App Growth @LukasAppGrowth]

Trying to test every single ad in ABO is costing you thousands a month.

Upload 10 ads and Meta spends on the 2-3 with the strongest early signals. If those dont beat your CPA, theres a 95% chance the rest wont either. Write off the batch, move on.

And an ad Meta ignored in testing is basically cursed, it'll never want to spend on it. Forcing it isnt worth the squeeze. Trust the algo, focus on volume.

I help subscription apps scale Meta ads. Link in bio 👇

[Lukas App Growth @LukasAppGrowth]

agree on the example, but i'd frame the lesson slightly differently: the stickers aren't really 'product', they're a marketing feature disguised as one. a better journal editor is pure product and it still makes $0. the stickers win because they're screenshot-able and demoable in a 70k-view tiktok, that's distribution built into the feature. so it's not 'focus on product', it's 'build the one feature that markets itself'. that's what separates the $50k app from the 99%, not product quality, a product decision that doubles as a distribution decision

[Lukas App Growth @LukasAppGrowth]

the onboarding change working is a real signal, 0→5 trials from 'build trust before the paywall' is exactly right. one thing to watch now: those are trials, not payers yet, and the same principle you just used moves one step later. you built enough trust to get the card in, now the trial has to deliver the promised value fast (the aha) or they cancel before it charges. onboarding gets them to start, activation gets them to stay. watch how many of the 5 convert, that tells you if the trial delivers what the onboarding promised

[Lukas App Growth @LukasAppGrowth]

this is the point most people refuse to believe, a slideshow doing modest numbers can out-earn a viral one, because reach to the wrong people converts like reach to nobody. the 'right research' you're pointing at is really finding the exact pain and words of your buyer so the slideshow self-qualifies before they ever tap. watch conversion-per-view, not views, a small post with high conversion-per-view is the one you run as paid and scale. $527 in 5 days with 0 audience is the proof, you found the angle not the algorithm

[Lukas App Growth @LukasAppGrowth]

analytics like this are the piece most founders running ads completely skip, they judge everything off meta's dashboard which under-counts through skan. pairing real in-app revenue analytics with a 'how did you hear about us' survey is how you actually see your true cac, because meta always shows it worse than reality. most people optimize ads blind to their own numbers, this is the fix

[Lukas App Growth @LukasAppGrowth]

1.3m views on autopilot is the easy part, and honestly the trap. views aren't revenue, and an ai persona posting without a human reading what landed optimizes for reach not installs. automation multiplies whatever hook you feed it, so if the hook doesn't convert you just generated 1.3m views of the wrong people faster. the money machine isn't the mcp, it's whether one of those posts actually makes a stranger install and pay

[Lukas App Growth @LukasAppGrowth]

this is a real gap most app founders wing, good tool. one thing to pair with it: the index gets the number right, but price is the easy 20% of localizing. the same purchasing-power gap that makes $6.99 wrong for vietnam also means the plan you highlight should differ (weekly as hero in low-ppp markets, annual anchor in high-ppp), and a ppp-correct price behind an english store listing and paywall still converts badly. localize the number, but the packaging and paywall copy have to move with it or you've only done the easy part

[Lukas App Growth @LukasAppGrowth]

the 'don't look like an ad' point has one wrinkle specific to apps vs dtc: with a physical product a viewer can infer what it is from the shot, so a street interview or documentary can stay subtle. an app can't be inferred, so the same format still has to land the 'oh, there's an app for that' beat or you get cheap engaged views that never install. don't-look-like-an-ad works for apps only when it's wrapped around the problem AND resolves into the app moment, otherwise you're buying awareness for a product nobody knows exists

[Lukas App Growth @LukasAppGrowth]

you said it better than i did. one thing that makes the fast-signal read actually work: scale in isolation. if you bump the winner inside a blended campaign, rising cpm and cost per trial-start get muddied by everything else and you can't tell it's THIS ad hitting its ceiling vs account-wide noise. isolate the scale test (own ad set, small bump) so the fast signal is clean enough to act on before revenue confirms. clean signal is the whole reason the fast metrics beat waiting for roas

[Lukas App Growth @LukasAppGrowth]

the awareness-level mapping is the part 99% skip, and one extension makes it pay for apps: map the destination across awareness too, not just the creative. a product-aware person can go straight to the store, but a problem-aware person (the biggest pool for most apps) needs a pre-sell first, a quiz or advertorial, or they hit a cold paywall and bounce. most founders map angles across awareness then send everyone to the same store page, so the low-awareness traffic they worked to reach leaks at the door. the angle and the destination move together

[Lukas App Growth @LukasAppGrowth]

the 4→20 progression is the right call, that's how you find winners faster. one thing to lock in before the volume hits: at 20/day you'll drown in view-winners, and the format that gets views is rarely the one that drives installs. so when you 'double down on best performing', define best as downstream (profile visits → installs → trials), not views. otherwise volume just makes you very good at entertaining people who never open the app. track the format that converts, not the one that pops