Transferring spend from Google or Meta to a new ad network is not a straight technical switch. You are plugging into different traffic sources, different auction mechanics, and different levels of fraud – and without testing, you don’t really know what you are getting.
Audit the network before the campaign goes live
The initial point of validation is where the network gets its inventory. Does it maintain direct relationships with publishers that serve as a source? If not, the network will likely have brokered traffic, meaning it’s buying inventory from other exchanges. The less directly sourced traffic you have, the more likely fraudulent traffic is to slip through. If impressions are coming from four different exchanges down the line, that’s an easy way for bad players to get access to the economy.
The second point is post-click monitoring. Post-impression fraud detection is the most basic line of defense. Better companies are going to be validating for bots after the click because they’re going to have JavaScript emulators on their servers that actually render the ad and see if it’s in a background tab or preloaded and not visible. That’s an expensive process, but it’s also a very good way of detecting if the person you’re talking to on the other side of the click is a machine.
Use full-screen formats to stress-test traffic quality fast
Banner advertisements are not an effective tool to test a new network. They have extremely low click-through rates and results require a long time before they become significant. In contrast, mobile interstitial ads have much higher performance. In fact, mobile advertising benchmarks generally report that interstitial click-through rates are higher than with standard mobile banners.
This performance gap is natural since interstitial ads are shown at typical points of rest in the app’s flow. Users are more likely to be pausing naturally between actions, and so the ad does not have to compete with what they were trying to accomplish. This gives you quicker feedback on the actual traffic quality of the network and the impressions’ true cost.
If you are in the process of defining your format strategy, working with the best interstitial ad network that can offer you publisher placements and clean delivery mechanisms will ensure that you are not measuring user behavior falsely and getting results based on low-quality clicks.
Cap frequency before you need to
New advertisers on unfamiliar networks often serve their ads more often than they should in the first week. The network doesn’t know your audience overlap. You’re running broad targeting. Users in some geos see the same ad eight times a day.
Frequency capping should be something you decide before you launch the campaign, not add as a fix when performance drops. One to two interstitial impressions per user in a 24-hour window is a good place to start. The UX argument here isn’t just about semantics either. Repeated exposure to an intrusive full-screen ad at high frequency actively damages conversion rates. The users aren’t sick of your product. They’re sick of your ad.
Start wide, then cut hard based on actual data
Segmenting a campaign too much on a new network is a simple error. If you don’t have an idea about how traffic on the network is spread out among devices, operating systems, and publisher IDs, and you’re starting with overly specific targeting, the data is going to be too sparse to be meaningful.
Just launch a broad test with your core geos, then take the performance data from the first week. Use it to determine which publisher IDs, devices, and operating systems are converting. Once you have that info, drop all underperforming sources without remorse. A publisher ID that has spent a budget of 500 clicks but hasn’t seen a single conversion isn’t something you’re about to crack – it’s a failure.
Geo can be isolated early as well. On virtually every network, traffic quality varies by region. What works well in one geo will flop in another, even under the exact same campaign structure.
Treat the first week as a paid research project
Allocate a testing budget of 10% to 20% of your total planned spending and consider the test phase to be an exploration phase, not a performance phase. You do not have to be immediately profitable. You need to be able to establish a benchmark and to become familiar enough with the network’s inventory to be able to scale profitably.
Run at least three or four quite different creative concepts in your A/B test – different headlines, different images, different calls to action. Creative fatigue for interstitials has a quicker onset for burn out than display, since the format is a high-attention format. If you push all your creative through in week 1 without testing multiple approaches, you will not be able to sustain performance on the winners when the others start to fade.
Also measure eCPM, not just CTR and CPA. A campaign that looks cost-effective on a per click basis may be driving your eCPM up and eating your profit as you burn through the most productive segments.
The real goal is a repeatable launch process
It is more important to be able to launch multiple campaigns on a new network than to get one campaign profitable. Focus on taking thorough notes on everything you test, what cut rates or bids you use, which targeting guidelines hold, etc. Then you can replicate the best results quickly.
