You hand over the bidding and keep one number. That number is the whole campaign.
Every other campaign type asks you what a tap is worth. Maximize Conversions inverts it: you give Apple a target cost per install, and it sets every bid itself, optimising toward that figure as a weekly average rather than a per-tap cap.
That inversion is the appeal and the risk. You stop managing bids; you also stop being able to correct one. The target is the only lever you have, and it is not a cap — individual installs will cost more and less than it.
Set it too low and the campaign barely spends, because Apple will not buy traffic it does not believe can hit the number. Set it too high and it spends comfortably on installs you cannot afford. There is no bid to adjust afterwards; there is only this.
So it should come from your own arithmetic, not from a plausible-looking default: what an install earns you, less what you must keep, and then some room, because paid installs convert worse than the blend your revenue figure is averaged over. Working out that number is the same exercise as pricing a tap.
Apple's own guidance is a daily budget that allows for at least five conversions a day. Below that, the algorithm does not have enough events to model anything, and the campaign will spend a fortnight and still have no reliable cost per install to report.
| Your target | Apple’s learning floor | Comfortable |
|---|---|---|
| EUR 2 | EUR 10/day | EUR 60/day |
| EUR 5 | EUR 25/day | EUR 150/day |
| EUR 10 | EUR 50/day | EUR 300/day |
The floor is Apple's. The comfortable column is thirty times the target — ours, not Apple's, and worth naming as such: it is room to learn quickly rather than the least it can learn from at all.
As a first campaign for a new app, usually not. Exact and Search Match teach you which searches convert, at a price you control, and that knowledge is what makes a good target possible later. Handing bidding to an algorithm before you know your own numbers means you cannot tell whether a disappointing fortnight was the algorithm, the target, or the app.
It also cannot be run on the same budget logic as the others: its budget comes from the target, not from a share of a pot, so it adds to the day’s exposure rather than dividing it. Apple has no account-level cap — the sum of your campaign budgets is what you are committing.
A campaign type where you set a target cost per install and Apple sets every bid itself, optimising toward that figure as a weekly average rather than a per-tap cap. You give up bid control and keep one number.
Apple's guidance is a daily budget that allows at least five conversions a day, so five times your target. Below that the algorithm has too few events to learn from and the fortnight ends with no reliable cost per install.
You can, but each change makes Apple rebuild its model against the new figure and the learning period starts again. A campaign adjusted every few days never finishes learning. Set it and leave it for a fortnight.
Usually not. It needs conversions to learn from and a target that comes from knowing what an install is worth, and a new app has neither. Exact and Search Match teach you both at a price you control.