How to build a viewability-aware bid adjustment from log data

How to build a viewability-aware bid adjustment from log data

Paying the same for a guaranteed-viewable slot and a below-the-fold afterthought is a valuation error. Here's how to price viewability into bids using your own data.

1. Join your win logs to post-bid measurement so each won impression carries a measured viewability outcome (viewed / not viewed).

2. Segment by the predictive signals available pre-bid: slot position from the ad request, publisher, device, and ad format. You can only adjust on what you know before bidding.

3. For each segment compute realized viewability rate over a meaningful sample — at least a few thousand impressions to avoid noise.

4. Convert to a value multiplier: a segment viewing at 40% is worth roughly half a segment viewing at 80% for an awareness goal, so scale the bid accordingly.

5. Watch out for the feedback trap — if you bid down low-viewability segments, you'll win the cheapest, lowest-viewability impressions within them. Re-measure after applying multipliers.

6. Refresh segment rates monthly; publisher layouts and your auction mix both drift.

Why it matters: Viewability is usually treated as a post-campaign report card when it should be a pre-bid input. Folding measured viewability back into the bid — segment by segment, on signals knowable before the auction — aligns what you pay with what you actually receive, instead of discovering the gap after the budget is gone.