After a decade running app campaigns across the GCC, I've noticed something: the budget rarely leaks in exotic ways. It leaks in the same seven places, over and over, across fintech, marketplace and consumer apps alike. None of them show up as a dramatic failure. They show up as a CPI that looks fine while revenue quietly underperforms. Here are the seven, and what to do about each.
1. Optimising to installs instead of value
The most common one. Campaigns are pointed at cheap installs because installs are easy to count and easy to feel good about. But an install isn't a customer. If your platforms are optimising toward "install" rather than the event that actually matters, subscription, purchase, qualified signup, the algorithm will happily find you the cheapest, lowest-intent users on earth. Fix: define the real value event and optimise to it, even if volume drops at first.
2. Treating iOS measurement like it's still 2020
Since Apple's privacy changes, iOS attribution runs through SKAdNetwork, and most teams have it half-configured, or configured once and never revisited. The result is a measurement layer that quietly lies to you. Fix: get your SKAdNetwork conversion values mapped to real value tiers, and stop making iOS decisions on numbers you haven't validated.
3. Creative concentration and fatigue
I'll often pull an advertiser's live ads and find twenty variations of the same two or three angles. Spend is high, the creative is tired, and there's nothing fresh in the pipeline. Performance doesn't fall off a cliff, it erodes. Fix: run a small number of genuinely distinct angles, and build a repeatable system that produces the next batch before the current one fatigues.
4. Discount-led creative that trains the wrong users
Leading every ad with a discount is tempting, it works in the short term. But if every creative shouts "50% off," you train an audience of deal-seekers who churn the moment the discount ends. Fix: test value-led and results-led creative against your discount control, and watch retention, not just install cost.
5. No view of lifetime value by channel
Plenty of teams know their blended CAC. Far fewer can tell me which channel brings users who are still around in ninety days. Without LTV by channel, you scale on cost, which means you often scale the channel that's quietly bringing your worst users. Fix: connect post-install value back to channel, and let the money follow the users who stay.
6. Treating Android and iOS as one thing
They behave differently, on cost, on measurement, on user quality, and on store conversion. Averaging them hides problems. I regularly find one platform subsidising the other while a leaky store listing or a misconfigured build drags the whole account down. Fix: read and budget the two platforms separately.
7. Confusing platform-reported numbers with the truth
Meta, Google and TikTok will each happily claim the same conversion. Add them up and you've "generated" more revenue than you actually made. Without an independent MMP and some sense of incrementality, you're optimising toward whoever over-reports the most. Fix: trust your MMP over self-reported platform numbers, and test incrementality on your biggest spend.
Running the ads is no longer the hard part. The platforms automate that. Knowing what to feed them, and being able to read the truth afterwards, is where budgets are won or lost.
None of these require a bigger budget to fix. They require pointing the budget you already have at the right users, and being able to prove it. If you're not sure which of these is costing you, that's exactly what a free App Growth Review is for.