Apple Search Ads Managed on D7 Revenue, Not Installs

The number that decides an Apple Search Ads account is seven-day revenue per install at keyword level, and it is readable eight to ten days after a term goes live. Cost per install is not that number: it looks best on brand and near-brand terms, which on every account I have inherited absorbed between 55 and 70 percent of spend. So month one starts with a fourteen-day brand pause and an organic baseline read. If installs hold, that budget moves to generic terms and stays there. If they fall, we now know the real price of defending the name. How the account is built: Four campaigns per storefront: Brand, Competitor, Category generic, and a Discovery campaign whose only job is to produce search terms for the other three. Discovery keeps Search Match on behind a hard negative list, and a term graduates to its own exact keyword in Generic after two installs or forty taps, whichever comes first. Bids are set from tap-through rate and conversion rate, not from a target CPI handed down by finance. Keyword and bid changes run through the Campaign Management API v5 on a schedule, which is also what makes week-over-week changes auditable. Measurement: The conversion value schema is where most app accounts quietly lose the ability to make decisions. I map fine conversion values to first-session revenue bands plus one mid-funnel event that arrives inside the postback window: for a game, session length band and first purchase; for a subscription app, trial start and paywall view. Where crowd anonymity tiers suppress the fine value, coarse values carry the revenue band and nothing else. I reconcile three sources monthly: Apple's own attribution, your MMP, and store-level installs. They will not agree. The gap is usually ten to twenty percent and the useful work is knowing which direction it leans, not closing it. Custom product pages: Keyword themes that deserve their own page get one. I write the brief — the promise, the order of the screenshots, the caption text, the first frame of the preview video — and your designer builds it. Pages are tested with enough tap volume behind each to read a difference, and a losing page is retired rather than iterated forever. Not included: App Store Optimisation for the default listing. Title, subtitle, keyword field and description are somebody else's discipline and I will not pretend the two jobs are one. No creative production. No Google App Campaigns, Meta or TikTok user acquisition — if you want one partner across all paid acquisition, I am not that partner. No buying against a target CPI with no revenue signal attached to it. Who this is not for: Apps drawing fewer than roughly four hundred store-listing installs a week. Postbacks arrive too sparse and too delayed for keyword decisions, and you would be paying me to read noise. Apps that monetise offline more than thirty days after install with no in-app event to stand in for it. Teams that want a CPI target agreed at kickoff and no further conversation: the account changes when the schema, the storefronts or the product change, and those conversations are the work. Reporting: Weekly, keyword level, in English or Portuguese, with the three-source reconciliation attached. I bring the decisions I would make and name the one I am least certain about.