​
Check your GA4 and Google Ads connection

Before you can use Audience Signals in PMax, you need to make sure your data sources are correctly connected. In this setup, we assume both GA4 and Google Ads are already linked and configured to share audiences. That means:
  • Your GA4 audiences are available in the Google Ads Audience Manager
  • User identity mapping is working reliably (audience sizes are > 0)
Once this is in place, we’re ready to activate your signals in PMax.

​
PMax campaign structure

When introducing new audience signals into Performance Max, we recommend starting with a testing phase of 30 to 60 days. This allows the campaigns sufficient time to learn and adapt to the new signals before drawing conclusions about performance.

​
Before launching

Before launching your campaigns, ensure that your Google Ads source is configured to capture campaign IDs for each PMax campaign. In addition, make sure that each campaign is tagged with dynamic campaign IDs (e.g. via UTM parameters or tracking templates). This setup is essential to map traffic and conversions back to the correct Innkeepr audience signal, which in turn supports learning and continuous signal optimization. Without campaign ID tracking, this feedback loop is disrupted and downstream model improvements become unreliable.

​
Split campaigns vs. asset groups

A key decision in testing audience signals is whether to test them via asset groups within a single campaign or through separate campaigns. As of this writing, Google Ads does not yet provide click-based performance reporting at the asset group level via the Ads API. This limits the ability of attribution and measurement providers to reliably associate performance with specific asset groups. Until more granular reporting becomes available, we recommend structuring tests as separate campaigns. This approach ensures clean reporting and better attribution feasibility.

​
Use multiple audience signals

When building your campaign, start with two or three different asset groups for the same listing items. Since Google’s algorithm leverages signals differently across the Google network, each asset group has the potential to surface your ads in different placements depending on the feature vectors of the audience behind the signal. For example, your warm audience (e.g., product viewers) may perform best on Shopping and Search inventory, while colder audiences (e.g., blog traffic) may be better suited for Display or YouTube. Splitting asset groups by signal allows PMax to explore these areas independently - while giving you insight into where your budget is performing best.

​
Advanced Tips

​
tROAS

When testing with a campaign-level split, it’s important to align your tROAS settings across both test and control group campaigns. If multiple Performance Max campaigns promote similar or identical listing items, Google Ads will tend to favor the campaign with the lower tROAS setting, as it allows more flexibility in bidding. As a result, that campaign may receive a larger share of impressions and budget, while the other campaign sees reduced delivery. To ensure a balanced test, we recommend keeping the tROAS setting equal across both campaigns. The actual tROAS value can still be chosen based on your business objectives - it just needs to be the same in both groups to ensure that differences in performance are driven by the tested audience signals, not by bid strategy.

​
Check inventory spend

A key characteristic of Performance Max campaigns is that they dynamically allocate spend across Google’s inventory based on recent performance data. This means your campaigns may shift impressions between Shopping, Search, Display, and Video depending on which channel is currently driving the best results. Because of this dynamic allocation, you may experience short-term performance volatility, especially when new signals are introduced or budgets are scaled. In scenarios powered by Innkeepr audience signals, we’ve observed that Performance Max may start spending on Display or Video inventory earlier than usual - sometimes more aggressively than in traditional campaigns. This is often due to Innkeepr signals meeting conversion goals faster, which leads the PMax algorithm to deprioritize Shopping and Search sooner than usual. If you notice a drop in ROAS or a shift toward less profitable placements, we recommend the following:
  • Increase your target ROAS to push PMax toward higher-quality inventory
  • Lower your daily budget temporarily to slow the pace of optimization
  • Monitor the Performance Max placement report to understand where spend is going
These adjustments can help rebalance your campaign toward more profitable channels while maintaining overall stability.