Branded, generic, and discovery campaigns working together, with a keyword waterfall as the engine.
The company I was working for decided to launch a mobile app. When I joined the project, the app itself wasn't ready yet, the development team was still building it.
There was no product to promote yet. Spending on paid acquisition before there was an app to download would just have been burning budget with nothing to show for it. The pressure to "do something" was there, but the right move was to not spend anything yet.
Instead of putting money into the market too early, I used the waiting period to properly learn the channel. I took a dedicated Apple Search Ads course to understand budgeting, account structure, and the mechanics of how the platform actually works, rather than launching on assumptions once the app was ready.
Once the app was ready to launch, I built the account around four campaign types, based on keyword research done up front:
Branded campaigns, to protect the brand's own name in search
Generic campaigns, split into:
A placement based campaign
Discovery campaigns, used specifically as a keyword mining tool
The core of the approach was a waterfall between the discovery campaign and the generic campaigns.
The discovery campaign worked similarly to a dynamic search ads campaign: instead of targeting fixed keywords, it surfaced whatever relevant search terms were actually driving downloads. Those keywords were then reviewed, and the strongest performers were moved into the generic campaigns as dedicated, targeted keywords.
Once a keyword was promoted into a generic campaign, it was added back into the discovery campaign as a negative keyword. That stopped the discovery campaign from competing with the generic campaign over the same term, and forced it to keep surfacing new keyword opportunities instead of just repeating what already worked.
This meant constant negative keyword maintenance, checking for overlap between campaigns and updating the negative lists, was the ongoing engine behind the account rather than a one time setup task.
The instinct when told to launch an app is to start spending on day one. Waiting instead, and using that time to genuinely learn the channel through a proper course rather than trial and error on a live budget, meant the account was structured correctly from the start instead of needing to be rebuilt later.
The waterfall structure itself is the more transferable lesson: a discovery or broad campaign isn't just a top of funnel play, it's a live keyword research engine, as long as negative keyword hygiene is maintained so it doesn't just end up competing with your own targeted campaigns.
This work sits under App Growth within Industry Growth, supported by the core Performance Marketing discipline used to structure and manage the campaigns.
See also: Google Ads revenue growth, the same client's paid search work on a different channel.
No. Spending on paid acquisition before there's a product to download burns budget with nothing to show for it. The more effective use of that period is preparation, such as formal training on the ad platform's budgeting and account structure, so the account is built correctly once the app is ready to launch.
It functions as a keyword mining tool, similar in principle to a dynamic search ads campaign. Instead of targeting fixed terms, it surfaces whatever relevant search terms are actually driving downloads, which can then be evaluated and promoted into more targeted campaigns.
To prevent the discovery campaign from competing with the more targeted campaign over the same keyword. Without that step, the two campaigns would bid against each other instead of the discovery campaign continuing to surface new, untapped keyword opportunities, a structure often called a keyword waterfall.
By device, country, and language, so campaigns match how an app is actually being searched for and used in each market, rather than running a single undifferentiated structure globally.
Ongoing. It only works if negative keywords are actively maintained and checked for overlap between campaigns on a continuing basis, rather than configured once and left alone.