How do you set up Google app install campaigns that scale?

How should you structure app campaigns for installs?

Keep the structure boring: one campaign per operating system per market tier. For most Gulf apps that means an Android campaign for Saudi and the UAE, a second Android campaign for the rest of the GCC, and separate iOS campaigns mirroring the same split. Never let two campaigns compete for the same OS and geography — they bid against each other and split the conversion data both need to learn.

Resist the urge to fragment further. App campaigns learn from conversion volume, and every split halves the signal. The one split that does pay is language-led: an Arabic-asset campaign and an English-asset campaign for the same tier can each find their own audience pockets, but only once you have the conversion volume to feed both.

Google UAC best practices: the asset stack decides everything

With no keywords or audiences to set, assets are your only steering wheel. Treat the stack as a portfolio the machine assembles per placement:

• Five headlines (30 characters) covering five distinct angles — price, speed, trust, selection, outcome — not five phrasings of one line.
• Five descriptions (90 characters) that each stand alone; Google mixes them freely with any headline.
• At least ten images across landscape, square and portrait — text-light, because they run on tiny placements.
• At least five videos, portrait-first; a 15-second demo with Arabic voiceover consistently earns Gulf inventory the English asset never reaches.
• Check asset ratings after two weeks and replace 'Low' performers one or two at a time — swapping the whole stack at once resets learning.
• Match asset language to campaign language: Arabic campaigns deserve natively written Arabic copy, not translations.

Should you bid tCPI, in-app actions, or tROAS?

Climb the ladder in order. Start with target CPI (installs) to build volume and exit learning — this is where every new app campaign should live for its first weeks. Move to in-app action bidding (registration, first order, purchase) once that event fires at least ten times a day; below that, the algorithm optimizes on noise. Reserve tROAS for apps with high, stable purchase volume and consistent revenue values — it is the most demanding strategy on the platform and starves without data.

Two budget rules keep the machine sane: give install campaigns a daily budget around 50 times the target CPI, and action campaigns roughly ten times the target CPA. Underfund either and the campaign oscillates between overspending mornings and going dark by afternoon — which reads as instability in every metric you track.

Why does your MMP show different numbers than Google Ads?

The console and your MMP will never match, and the gap can be multiples — in either direction, on installs and especially on in-app purchases. Google is a self-attributing network: it claims any install it touched within its windows, counts view-through and engaged-view conversions, and models conversions where privacy hides them. Your MMP applies its own windows, last-click logic and cross-channel deduplication to the same events.

The decision rule that keeps you honest: the MMP is the source of truth for cross-channel budget allocation, and the console is a directional signal for in-channel optimization — asset ratings, bid pacing, week-over-week trends. Never add platform-reported conversions across channels; the sum will exceed reality, sometimes absurdly. And if the gap suddenly widens, audit the conversion import first — a broken event link silently starves the bidder.

How do you scale Google app campaigns without breaking them?

Scaling UAC is a patience game — the algorithm punishes sudden moves. The working rules:

• Raise budgets no more than about 20% at a time, then hold for two to three days before the next step.
• Move tCPI or tCPA targets in 10–15% steps — never budget and bid in the same move.
• Scale horizontally by opening the next market tier as its own campaign, not by stretching one campaign across the whole region.
• Expect CPI to drift upward as cheap inventory exhausts; judge scale on cost per retained user, not on holding a vanity CPI.
• Plan around the calendar: Ramadan evenings and White Friday weeks inflate CPMs across the Gulf — either pre-scale before the surge or budget through it deliberately.
• Refresh videos every few weeks at high spend; a fatigued video stack is the most common reason a scaling UAC quietly stalls.