Each cluster's monthly demand curve is the average of constituent states' 5-year Google Trends data.
For each client: the most recent completed year (model validation) and the current year's YTD pacing + annual forecast.
Growing tier ($1-2M) blended 3.3x = Meta 2.6x + Google 4.1x. Two of three clients are losing money on Meta specifically while Google works for them. Fix the channel mix and blended ROAS jumps without new investment.
Year-to-date actuals vs the seasonality model's prediction. Projects an annual run-rate based on current pace and predicted distribution.
Between Mar and Apr 2026, the cohort saw clicks drop ~54% and CPC nearly double — but conversions HELD and conversion rate roughly doubled. Google's auction is funneling spend toward fewer, higher-intent users. The headline numbers (clicks, CPC) look bad. The bottom line (conversions, CPA, ROAS) is intact or better.
All client Google spend aggregated by month. Notice the divergence starting Apr 2026: clicks line declines while conversion-rate line rises.