Take almost all the ads out of someone’s Facebook feed and they barely use it more. In a field experiment, hiding 82% of the ads people saw increased their active time on Facebook by only 6%.
The study, by George Beknazar-Yuzbashev of the University of Chicago, Rafael Jiménez-Durán of Bocconi University, Andrey Simonov of Columbia University and Mateusz Stalinski of the University of Warwick, appeared as Warwick Economics Research Paper No. 1602 in February 2026. The authors treat ads as a price users pay in attention instead of money, and ask why platforms almost always charge that price rather than a subscription.
A browser extension that edited people’s feeds
The team recruited 1,810 U.S. Facebook users who installed a custom desktop browser extension. For six weeks it simply recorded their activity. Then users were randomly assigned to one of two interventions for six more weeks, with each arm getting its own control group.
In the first arm, the extension hid most ads. In the second, ads kept appearing at the same rate but were swapped for ads that had originally been shown to other users, which made the ads less well matched to each person. The extension kept running for about 13 more weeks after the main period, so the team could see whether effects lasted. The design was preregistered with the American Economic Association.
Users hardly respond to fewer ads
Hiding ads cut the ad load people experienced by 82%. Content consumption rose by just 9% and active time by 6%. The authors calculate that this implies an elasticity below 0.1, meaning time on Facebook is highly insensitive to how many ads appear. The effect stayed small over 19 weeks of continued hiding.
Removing targeting mattered more. The swapped, less relevant ads drew far fewer clicks, and in the long run active time and content consumption both fell by 13%. Facebook’s own algorithm did not respond to hidden ads, but it reacted mildly to the swaps by serving less-targeted ads.
Neither intervention pushed people to X or Reddit in any meaningful way, and neither changed how much money users said they would need to give up Facebook for four weeks.
Different users pay different attention prices
Even without any intervention, ad loads varied widely between people. On average, about 13% of feed content was ads, roughly one ad in every eight items, but 67% of the daily variation came from differences between individual users. That pattern, which the authors read as a form of price discrimination, appeared on X and Reddit too, with average ad loads of 11% and 9%.
Using these estimates, the researchers simulated alternatives. An ad-funded Facebook earned profits comparable to a subscription model while delivering more engagement and higher consumer surplus. Even a hybrid with a paid ad-free tier, similar to the one Meta introduced in Europe, did little better: around 17% of simulated users chose to pay, and overall outcomes barely changed.
The core reason, the authors write, is that users react far more strongly to monetary prices than to advertising intensity, which makes ads “a relatively efficient revenue source” for platforms. They add that this holds constant the quality of content, and that their earlier work links engagement-driven incentives to the spread of toxic content.
Study Details:
- Title: Social media advertising loads as prices
- Authors: George Beknazar-Yuzbashev, Rafael Jiménez-Durán, Andrey Simonov, Mateusz Stalinski
- Journal: Warwick Economics Research Paper No. 1602 (also CAGE Discussion Paper 792)
- Publication Date: February 2026
- DOI: 10.2139/ssrn.5952337