Incentive Misalignments in Programmatic Advertising: Evidence from a Randomized Field Experiment

For your marketing

Possible founder application

Highlights the difference between predicting who will buy and identifying who changes behavior because of an ad. Ask for incrementality tests and examine whether optimization targets persuadable users.

Reported result

What the study found

Ads increased purchase and return probabilities, but consumers targeted more aggressively were not more responsive to ads. The authors estimate the platform's bidding optimization reduced advertiser profit and overall welfare by about 6% relative to a firm-optimal regime.

Research question

What the authors investigated

Do ad platforms target users who are persuadable, or users who were already likely to purchase?

Design

Randomized retargeting field experiment; consumers were assigned to retargeting ads or orthogonal public-service ads, with a separate welfare analysis.

Sample and setting

208,538 consumers eligible for retargeting at a European consumer-electronics e-retailer over 49 days.

Scope and transfer

Limitations to keep in view

One retailer and one retargeting campaign. The design does not capture defensive advertising against competitors, and the analysis observes only winning bids.

Research sources

INFORMS publisher abstract retrieved with Firecrawl; sample, randomization, findings and stated limitations were available on the page.