Founder Playbook · Sub Club by RevenueCat
14 tactics from Alper Taner
Creative Testing, Data-Driven Decisions, and Growing Mobile Apps
Watch the full episode“something becomes a problem after you systematically test it iterate on the results and try many different ways from let's say radical to iterative and you're still stuck at the same level then I think I can consider that as a potential problem”
A metric is just a fact until systematic iteration proves it a problem
A 5% install-to-trial rate (vs. a 15–20% benchmark) is not a problem — it is a fact until you have exhausted both radical and iterative approaches and are still stuck. Labeling something broken after one attempt, like $5k on Snapchat for a week, is premature diagnosis that blocks real learning.
“I believe the success of a test is at least 50% dependent on how you plan it how you strategize it how you hypothesize it and how you execute it and execution is just the what part maybe it's even sometimes less than 50%”
Planning and hypothesizing drive at least 50% of a test's success
Most teams over-index on execution and under-invest in test design. Alper argues the structure of your hypothesis — what you are testing, why, and what outcome changes behavior — matters as much as the creative or spend itself. Skipping this makes results uninterpretable and iteration impossible.
“I would more look at your cost per successful creative that is beating the BAU ads performance and it doesn't matter if you test 20 versus 200”
Track cost per winning creative, not total creatives tested
The industry race to 500 or 700 creatives per month is a vanity metric. What matters is your success rate — how many creatives beat your baseline BAU performance — and the production cost per winner. A small batch with strong hypotheses and a high hit rate beats a firehose approach where the analysis quality degrades under volume pressure.
“look at your cumulative spend and cumulative CPX metrics whether you're testing on a CPI or cost per trial or just cost per subscription and such and see when they stabilize”
Watch cumulative CPX stabilization — not a dollar threshold — to kill a test
There is no universal spend amount that signals a test is done — it depends on your CPI and funnel CPX. Instead, plot cumulative cost-per-X over time and cut when it stops fluctuating. Alper flagged accounts spending $20k to confirm what the data made obvious at $500: the signal stabilized early and was being ignored.
“just because algorithm decides to deliver one creative over others doesn't make the other creatives bad creative... that particular creative that got the 90% of the spend was maybe lucky we call it like lucky 5,000 impressions”
Pull the dominant ad to give underdelivered creatives a fair shot
Meta makes early allocation decisions within the first 5–10k impressions, meaning one ad can crowd out others not because it performs better but because it got lucky early. Removing the dominant ad from a test adset often causes suppressed creatives to suddenly bloom. Actively equalizing budget across creatives gives a cleaner read on true performance.
“let's have the same dog let's have a puppy let's have another breed let's have another animal so just to test and see if it's the animal or not”
Isolate one variable per creative phase — dog → breed → species → animal
Creative iteration should proceed in structured phases that isolate a single variable at a time rather than flooding a winning signal with hundreds of derivatives. By swapping one element per round — breed, species, count — you learn whether the category of element drives performance, not just one specific execution, before scaling production.
“look at your high LTV users like why do they stick around what do they use the most what messaging resonates with them and then because our goal in performance marketing is to find more of those high LTV users”
Use high-LTV in-app behavior as the brief for your next ad concept
First-party in-app analytics are an underused creative signal. The feature your best-retaining users engage with most is likely the hook that will resonate in ads, because UA's goal is replicating high-LTV cohorts. Most teams only look at ad-platform output to brief new creatives; layering in-app behavior data sharpens hypothesis quality significantly.
“if there is one concept with 15 variations and that is a particular the only one that has 15 variations compared other ones having just two and three you should be able to interpret that likely that creative performing well for them”
Meta ad library variation count signals competitor creative confidence
The Meta ad library reveals not just what competitors run, but how hard they are leaning into it. A concept scaled to 15 variations while others sit at 2–3 is a strong signal that concept is converting — use that as inspiration to adapt and test, filtered through your own audience data and past performance.
“I think it might matter but it matters a lot more once you get bigger because a lot of the decisions at such stage are driven by MMP/lastclick which is normal... 500k plus then these are becoming more a lot more important”
Last-click MMP works early; incrementality becomes critical above $500k/month
MMP last-click attribution is a practical starting point when budgets are small. But comparing UAC vs. Meta through last-click is misleading at scale because one is a push channel and one is pull — they do not compete for the same marginal user. Above roughly $500k/month, geo-lift, hold-out, and blackout incrementality tests are needed for sound budget allocation.
“they switched MMPs they didn't implement any events on the new MMP and the old MMP was still connected to meta they still had campaigns optimizing on meta”
MMP migration without re-implementing events silently breaks all ad optimization
A real case: a company migrated their MMP, forgot to implement conversion events on the new one, and left Meta campaigns running against stale signals from the old one — undetected for almost a month. The failure was silent because the attribution and media teams were siloed. This is a must-audit checklist item before any MMP switch.
“we have a million dollars a day budget at the campaign level... with the guard rails in you won't be spending 1 million a day of course you will only spend up to your total campaign cap”
Set campaign budget to $1M/day to remove pacing caps — guard with spend limits
At accounts already spending $50k+/day, an artificially low daily budget signals the algorithm to throttle delivery. Setting the campaign-level budget to $1M/day while placing real spend controls at the campaign-cap level removes that ceiling and lets the algorithm compete for higher-quality inventory. In one case this produced a hockey-stick volume increase while cost-per-purchase fell simultaneously.
“my apps have gotten featured a ton by Apple over the years and getting featured is fantastic because it's totally free zero cost of install but guess what those cohorts convert really poorly so do I have a conversion problem”
Cheap or featured installs create a low-conversion fact, not a conversion problem
Low trial conversion often reflects low-intent traffic, not a broken onboarding or paywall. Apple featuring delivers zero-cost installs that convert poorly — that is a traffic-quality reality, not a product failure. Before grinding on conversion optimization, check whether your cheap-install channel is structurally responsible for the low rate.
“all we did was mapping their trial event to trial and their cost per purchase/ the actual cost per trial went down by I think around 35% only with that change same creatives same campaigns just different event optimization”
Remapping trial event from purchase to trial cut CAC 35% — same creatives, same campaigns
One app with a high CAC was optimizing Meta toward purchase events — competing in an auction against e-commerce advertisers willing to pay $200 CPP for a conversion worth only $10–15 to the app. Simply remapping the optimization signal to trial start, with zero creative or campaign changes, dropped cost per trial by 35%. The mechanism: matching your bid signal to what you actually want lowers CPMs from irrelevant high-bidders.
“keep it simple keep it logical don't send all possible events because you can... when you send this mixed signals mixed funnel that doesn't make sense from the auction calculation point of view then you're also potentially losing out”
Send clean, logically ordered funnel events — do not fire everything the SDK supports
Ad platform algorithms build audience models from conversion signals. Firing every available event 'because you can' introduces contradictory funnel signals that confuse auction optimization. A clean logically-sequenced funnel (install → trial start → subscription) gives the algorithm a coherent user model; Alper has seen simpler accounts outperform complex multi-event setups because of this.