28 Apr 2026 · 9 min
Reading CAC drift across Meta and TikTok budget cycles
Drift is not a personality trait of an algorithm. It is what happens when spend, auction pressure, and creative age move on different clocks than your reporting window.
A Bangkok beauty brand once celebrated a Thursday CPA that looked 18% cheaper than the prior week. The budget had not been spent evenly. Delivery rushed on Wednesday night after a pacing under-delivery, the auction was softer, and two ads were still new enough that frequency had not caught up. By the following Tuesday the “efficiency” had vanished. Finance, looking at a monthly close, never saw the Thursday miracle.
Three clocks, one tile
Meta and TikTok budget cycles are operational clocks: daily pacing, learning phases, campaign budget optimisation shuffling. Your User Acquisition Cost Analytics clock is usually weekly or monthly and tied to booked orders. When those clocks disagree, CAC drift looks like insight. It is often just misalignment.
We ask Studio Seat students to plot three series on the same chart: cumulative spend, cumulative new-to-file orders (chosen window), and average creative age in days. Drift that appears only when creative age jumps is fatigue. Drift that appears only on under-pacing days is auction weather. Drift that remains after both are stable deserves a harder look — maybe incrementality, maybe a landing-page change, maybe a competitor sale.
TikTok is not Meta with a different logo
Spark Ads, creator usage, and in-feed native formats age differently. A TikTok that looks “cheap” in week one can be harvesting curiosity that does not convert after checkout lag. If your window is seven days, you will over-praise early TikTok and under-praise slower Meta prospecting that converts on day eleven via bank transfer.
A practical reading habit
Do not reallocate budget on a single platform’s day-28 number without checking pacing variance and creative age. In Creative-to-Cost Mapping we keep a simple rule: no scale decision on a set older than a threshold you pre-commit to (often 10–14 days of heavy frequency in a small geo). The number on the tile can still look fine. Contribution margin usually notices first.
For the underlying map, see User Acquisition Cost Analytics. For the classroom version, the fatigue module inside CAC Measurement Studio spends an afternoon on exactly this chart.