Using Creative Assets to Influence Network Distribution in Google App Campaigns

One thing I’ve learned from working across different industries, business models and digital channels is that best practices are useful - but they are still only a starting point.

What matters in the end is what the data tells you about your own campaign.

The observations below come mainly from my experience running B2C Google App Campaigns in a specific market. They won’t necessarily translate directly to every app, audience or business model. But the testing approach behind them is much more transferable: understand what you can control, change variables deliberately, give the system enough time to react, and evaluate what happens beyond the first conversion.

This became particularly important for me with Google App Campaigns.

Google App Campaigns - Universal App Campaigns (UAC) - automate much of what advertisers would traditionally control manually. Google can distribute ads across properties such as Search, YouTube, Google Display Network, while its systems determine where and how to serve them based on the campaign objective.

That removes a lot of manual control.

But it doesn’t mean there is nothing left to optimize.

One of the most important inputs you still control is your creative asset mix.

And those assets can influence more than the appearance of your ads.

Your assets can influence your network opportunities

In Google App Campaigns, you don’t manually decide that a certain percentage of your budget should go to Search and another percentage to YouTube.

But you do decide what you give the system to work with.

Different asset types and formats give Google different options for creating and serving ads across its available inventory. Text, image and video assets can be combined in different ways depending on the placement and ad format.

A strong selection of headlines and descriptions gives Google more options for text-based ad combinations, including placements on Search. A diverse video library creates more opportunities across video inventory, including YouTube. Images in different formats and aspect ratios give the system more flexibility across visual placements, including the Display Network.

The relationship isn’t one-to-one - text isn’t limited to Search, video isn’t limited to YouTube, and images aren’t limited to Display. But the composition of your creative library affects the range of ad formats and placements available to the campaign and can therefore indirectly influence network distribution.

I found this particularly useful when I started seeing clear differences in performance across networks.

Instead of thinking only about which individual creative was performing best, I started looking at a broader question:

What types of assets am I giving the algorithm - and how might that asset mix be influencing where the campaign can serve?

That changes the way you think about creative optimization.

Give new assets enough time

One of the easiest mistakes to make with automated campaigns is reacting too quickly.

You upload new assets, wait a few days, see that performance hasn’t improved - or has temporarily become worse - and start changing things again.

The difficulty is that every change gives the system something new to learn.

In my campaigns, I generally tried to give creative tests at least three weeks before drawing conclusions. When budget and campaign conditions allowed, four weeks or longer gave an even better picture.

I also tried to avoid changing too many major variables simultaneously.

If you replace videos, images and copy at the same time and performance changes, you have very little information about what actually caused it.

This is where patience becomes part of the testing methodology.

Change something deliberately. Give it time. Observe the effect. Then decide what to change next.

Creative libraries also need maintenance

The other side of giving creatives enough time is knowing when they have had enough time.

Assets don’t perform indefinitely.

Audiences see the same ads repeatedly, campaigns mature, competitors change their messaging, and creative that once performed well can gradually lose its impact.

For longer-running campaigns, I found it useful to review the creative library roughly every two months.

That doesn’t mean replacing everything.

I would look at which assets were consistently underperforming, introduce new variations, keep enough diversity for the system to continue testing, and avoid changing the entire asset set at once.

Google’s asset performance ratings can be useful signals here, but I wouldn’t use them in isolation.

The more important question is what those assets ultimately contribute to the business.

An asset can perform well according to an intermediate platform metric and still bring users who don’t progress very far after installing the app.

Which leads to the part I found most interesting.

Installs don’t tell you enough

When I looked at network performance, I didn’t want to know only which network generated the most installs.

I wanted to know what happened to those users afterwards.

Did they register?

Did they make a first purchase?

Did they come back?

Did they become valuable customers?

Once I started looking further down the funnel, the differences between networks became much clearer.

In the B2C campaigns I was managing, Search consistently brought some of the highest-value users.

Those users were often more expensive to acquire. But they also tended to have stronger intent and progressed further through the customer journey.

If I had optimized purely around acquisition cost or install volume, I could easily have reached a different conclusion.

Android and iOS didn’t behave the same way

Another pattern became visible when I separated Android and iOS performance.

For the iOS campaigns I managed, Search remained a strong source of valuable users, but YouTube also contributed users who continued further through the funnel.

Android behaved differently.

YouTube could generate plenty of installs and registrations, but many of those users didn’t progress much further.

That distinction mattered.

On the surface, the campaign was acquiring users. But once I looked beyond those initial actions, the quality of that acquisition wasn’t the same.

So I started experimenting with the asset mix.

For Android, I eventually reduced the campaign to one core video rather than maintaining a large video library.

This went against the general idea of giving the algorithm as much creative variety as possible. But my own campaign data gave me a reason to test it.

The intention was to limit the variety of video assets available to the campaign and see whether that change would affect network distribution and, ultimately, user quality.

This is an important distinction: I wasn’t manually moving budget from YouTube to Search.

App Campaigns don’t give you that level of network control.

I was changing one of the inputs available to the algorithm and observing how the campaign responded.

And that is really the point of the experiment.

Test the system, not just the creative

This changed how I thought about creative testing in Google App Campaigns.

Normally, when we talk about creative optimization, we ask questions like:

Which video performs better?

Which headline generates more conversions?

Which image gets more engagement?

Those are useful questions.

But in a highly automated campaign, I think there is another level worth testing:

How does changing the asset mix change the behavior of the campaign itself?

Don’t assume the distribution will stay the same

Another thing I learned is not to treat network distribution as permanent.

The campaign is continuously adapting.

As it collects more data, new creatives are introduced, bidding strategies mature and user behavior changes, the distribution across Google’s available inventory can change as well.

A network that represents a relatively small part of your traffic during one period can become much more significant later.

That’s another reason why I wouldn’t make decisions based on a few days of data.

Look for patterns.

Give the system enough time to respond.

And keep checking whether those patterns still hold.

The part you still control

Google App Campaigns are automated, but automated doesn’t mean passive.

You may not decide exactly where every euro or dollar is spent, but you still make decisions about the environment in which the algorithm operates.

You decide which assets are available.

You decide which conversion events matter.

You decide how you evaluate user quality.

You decide when there is enough evidence to make a change - and when there isn’t.

And sometimes, instead of trying to directly control something the platform doesn’t allow you to control, you can experiment with the inputs that may influence it.

That’s probably my biggest takeaway from working with Google App Campaigns:

Don’t try to outguess the algorithm. Give it better inputs, understand how it responds to them, and use your own data to decide what to change next.

Best practices can tell you where to start. Google can decide where and how to serve your ads. But neither can tell you which users are ultimately most valuable to your business.

That part still comes back to your data.

And that’s where optimization really starts.