For an online education provider, traditional monthly reporting was masking a much longer customer journey. By analyzing cohorts and the time between key actions, we found that students could first visit the website nearly a year before starting a course and register as much as four months before the course began. These insights changed how the organization could approach marketing timing, course planning, and demand forecasting.

Digital advertising platforms are incredibly powerful optimization engines, but they were originally built for a very specific type of buying behavior. For institutions offering certificates, professional training, or degree programs, the enrollment journey is usually much longer, which creates a challenge for modern ad platforms. Predictive modeling offers a solution.

We rolled out AI Max for Search across campaigns for clients in multiple verticals, including eCommerce, online education, and nonprofit. In most cases, it captured 5–10% of Search spend, produced ~30% stronger ROAS than non‑AI Max traffic, and uncovered incremental volume. However, it also proved to “cherry-pick” volume from existing keywords, which masked some inefficient spend, emphasizing the need for strong conversion signals and constant guardrails.

Predictive models are powerful tools for making smarter marketing decisions, but as with any model, the details matter. One common consideration is whether to include source as a variable. At first glance, it makes sense since different audiences generate different types of leads. But when predictive values are used not just for reporting, but for network optimization, including source as a variable can introduce risks that undermine campaign performance.