How to find where your frequency cap is leaking across the supply path

How to find where your frequency cap is leaking across the supply path

A frequency cap (the limit on how many times one user sees your ad) that's set but exceeded usually leaks at a known set of joints. Trace them in order.

1. Pull delivered impressions joined to a stable user identifier and count exposures per user. If the tail exceeds your cap, the cap is leaking — quantify by how much.

2. First suspect: identity fragmentation. If the same person is seen under multiple IDs (different cookies, mobile ad ID vs cookie, post-signal-loss noise), each ID gets its own cap budget. Count distinct IDs that resolve to one household or device.

3. Second suspect: cap scope. A cap set per line item, not per campaign, multiplies across every line a user qualifies for. Confirm where the counter lives.

4. Third suspect: multi-DSP buying. If you buy the same audience through two DSPs, neither sees the other's exposures. The cap can only be global if the counter is.

5. Rank leak sources by impressions wasted and fix the largest first.

Why it matters: Over-frequency is doubly costly — you pay for impressions that annoy rather than persuade. The leak is almost never the cap setting itself but the identity and scope layers beneath it, and only a per-user exposure count from delivery logs reveals which joint is failing.