Stress-Testing AI Max: How Forthea Turned Google's Newest Search Feature Into Real Growth

Google’s default AI Max settings focus on reach, not control. Before activating anything, we created brand-exclusion lists, set URL exclusions, and wrote text guidelines for each test campaign. We also added broad-match versions of each account’s main keywords to make sure future reports didn’t mistakenly credit AI Max traffic to existing keywords. This preparation gave us the confidence to test aggressively without risking brand safety or data clarity.

We picked non-brand prospecting campaigns with reliable conversion tracking and real budgets. We never used a client’s brand or ran nationwide campaigns. AI Max was tested as a true experiment, compared to a control group, instead of turning it on for every account. This approach not only limited potential downside but also revealed how performance changes could be attributed to AI Max itself, rather than external variables.

Search term matching, text customization, and final URL expansion can each be turned on or off separately. This let us see which part of the suite drove results and which needed more review before scaling up. By isolating variables in this way, we reduced the risk of misattributing success or failure and ensured that our recommendations were based on evidence rather than assumption.

For the first six weeks after launch, we checked the search terms report to spot competitor leakage, irrelevant queries, and unexpected landing pages. We quickly added negatives and exclusions as needed. This test required ongoing attention, not a set-and-forget approach. Without this level of vigilance, subtle inefficiencies and wasted spend could have easily compounded, undermining the value of the entire experiment.