Log-level data vs aggregated reporting: the resolution gap that hides your real cost

Log-level data vs aggregated reporting: the resolution gap that hides your real cost

Log-level data is one row per impression (or per bid). Aggregated reporting is the same data pre-summed into daily or campaign rollups. The choice between them determines which problems you can even see.

1. Aggregated reporting gives you averages: blended CPM, overall win-rate, daily spend. Fast, cheap, and it hides every distribution.

2. Log-level data gives you the distribution: the bid-to-clear gap on each auction, the seller ID on each impression, the exact path, the per-impression cost. Heavier to store and query, but it is the only place certain problems live.

3. The problems that are invisible in aggregates: self-competition across SSPs, fat-tail frequency leakage, a single placement dragging average CPM, over-shading on your best segment. An average of $2.40 CPM can hide a chunk bought at $9.

4. The tradeoff is operational: log-level needs pipeline and storage; aggregates need neither. Most teams default to aggregates and stay blind by convenience.

Why it matters: nearly every tactic in programmatic is diagnosed from a distribution, not a mean — yet most decisions get made on means. If you cannot pull per-impression rows, you cannot verify whether bid shading, SPO, or capping is actually working. The log is not a luxury for analysts; it is the instrument that makes every other tool legible.