Direct buy advertisers procure advertising inventory at fixed rates from publishers and ad networks. Such advertisers face the complex task of choosing ads amongst myriad new publisher sites. We offer evidence that advertisers do not excel at making these choices. Instead, they try many sites before settling on a favored set, consistent with advertiser learning. We subsequently model advertiser demand for publisher inventory wherein advertisers learn about advertising efficacy across publishers' sites. Results suggest that advertisers spend considerable resources advertising on sites they eventually abandon—in part because their prior beliefs about advertising efficacy on those sites are too optimistic. The median advertiser's expected CTR at a new site is 0.23%, five times higher than the true median CTR of 0.045%.

We consider how pooling advertiser information remediates this problem. Specifically, we show that ads with similar visual elements garner similar CTRS, enabling advertisers to better predict ad performance at new sites. Counterfactual analyses indicate that gains from pooling advertiser information are substantial: over six months, we estimate a median advertiser welfare gain of $2,756 (a 15.5% increase) and a median publisher revenue gain of $9,618 (a 63.9% increase). 

This paper empirically studies a central question in market design: how information provision shapes trade surplus and competition. I study online display advertising, one of the largest auction markets in the world. Using a dataset of 1.4 billion auctions, I estimate a model of advertiser entry and bidding with endogenous market thickness. The model reveals a fundamental tradeoff: more granular information raises bids through a matching effect, as advertisers better identify high-value users. But it also reduces market thickness, since precise targeting allows advertisers to enter fewer auctions. Counterfactual simulations show that eliminating access to consumer-level information would reduce total ad spending by 65%, driven by an 89% decline in match values, partially offset by a 24% increase from stronger competition.

This paper develops an experimental test of private values (PV) against common values (CV) for repeated auction settings in which bidder counts and losing bids are unobserved, but the researcher observes a bidder's own bids, wins, payments, and ex post outcomes. The test uses exogenous variation in a bidder's submitted bid. A bid increase expands the set of auctions won toward auctions with stronger rival signals. Under PV, rivals' signals do not affect the bidder's value conditional on its own information, so the ex post value of won auctions should not increase. Under CV, the marginal auctions won after a bid increase should have higher realized value. I implement the test in real-time bidding auctions for online advertising using a field experiment that varied campaign bids. Bid increases raise payments, confirming that the experiment moves campaigns to more competitive auctions, but do not raise click-through rates. The results provide no evidence of CV in this setting, favoring the PV assumption commonly used in the literature.