93% of people in a peer-reviewed 2020 study agreed to a fictitious social network’s terms of service without reading it closely enough to catch clauses that handed over their first-born child, according to Obar and Oeldorf-Hirsch’s research published in Information, Communication & Society. The finding is not an outlier. Independent research using different methods, from Deloitte’s consumer surveys to MeasuringU’s screen-time logs, puts the accept-without-reading rate for terms of service and license agreements anywhere from 71% to 93%.
What varies by study is not whether people skip the fine print, it is by how much, and what that means for the businesses that write it.
How many people accept terms of service without reading them?
93% of participants in Obar and Oeldorf-Hirsch’s controlled 2020 study clicked agree on a terms of service document they had not read closely enough to notice its contents, and 97% did the same for the accompanying privacy policy. The study set up a fictitious social networking service called NameDrop and gave 543 participants a real sign-up flow, including a “quick join” shortcut that let them skip past the policy links entirely.
Figure 1: Share of participants who skipped straight past NameDrop’s privacy policy versus those who opened it. Source: Obar & Oeldorf-Hirsch, 2020, n=543.
74% of participants chose the quick-join shortcut and never opened the privacy policy at all. The remaining 26% opened the policy screen, yet the terms of service agreement rate stayed at 93% regardless, since most of that group still clicked agree without reading the document in front of them. Skipping the link and skimming past it produced almost the same outcome.
The rate is consistent with older, larger-sample survey data. Deloitte’s 2017 Global Mobile Consumer Survey of 2,000 US consumers found 91% consent to terms of service without reading them, a figure that climbs to 97% among 18 to 34 year olds. Two very different methods, an observed lab experiment and a self-reported national survey, land within two points of each other.
What do people miss when they click agree without reading?
98% of the NameDrop participants missed both of the study’s deliberately absurd “gotcha clauses”: one stating their personal data would be shared with the NSA and their employer, and another stating they agreed to provide their first-born child as payment for the service. Only 2% flagged either clause during the study.
Figure 2: The path from sign-up screen to agreement in Obar and Oeldorf-Hirsch’s study. Source: Obar & Oeldorf-Hirsch, 2020, n=543.
The clauses were written to be impossible to defend as fine print, and they still worked. The researchers designed them specifically so that any participant who read even a few sentences past the opening paragraph would catch them, and the vast majority did not get there. This matters beyond the lab: a terms of service page in production rarely announces its most consequential clause, an arbitration waiver or a broad data-sharing right, with anything like NameDrop’s exaggerated wording, so a real document is arguably easier to skim past than these test cases.
The takeaway holds regardless of how deliberately a clause is written to stand out: if the reading rate stays this low, clause placement and plain language do more work than legal severity ever will.
How fast do people click through instead of reading?
Participants in the NameDrop study spent an average of 51 seconds on the terms of service page, a document that should take 15 to 17 minutes to read at a normal adult reading speed of 250 to 280 words per minute. That is roughly 3% of the time actually required.
Figure 3: How quickly users abandon a license agreement screen. Source: MeasuringU, log analysis of roughly 2,500 users across several consumer software products.
A separate MeasuringU analysis, built from server logs rather than a lab setting, found that more than half of users spend under 8 seconds on a software license agreement screen and at least 70% spend under 12 seconds. Assuming a genuine read takes a minimum of 2 minutes, the same analysis put the ceiling on full readers at 8%, meaning at least 92% leave before they could plausibly have finished. If you want to see just how long a real company’s terms actually run once someone commits to reading them, our breakdown of reading time of famous companies’ terms of service puts some of the longest agreements at nearly two hours.
The pattern across both studies is the same: people leave the page in seconds, not minutes, whether the incentive is a fictitious social network or a real piece of software they already paid for.
How do these acceptance rates compare across different studies?
No large-sample study of the last decade has found that most people read a terms of service or license agreement before accepting it. The specific number moves with the method, self-reported surveys tend to land in the low 90s, and behavioral logging finds figures anywhere from the low 70s to the low 90s, but every methodology points the same direction.
Figure 4: Accept-without-reading and skip rates across four independently conducted studies. Sources: see citations below.
| Study | Year | Sample size | Share who accept/skip without reading |
|---|---|---|---|
| Obar & Oeldorf-Hirsch | 2020 | 543 | 93% agreed to the terms of service |
| MeasuringU | 2016 | ~2,500 | 92%+ spent under 2 minutes on the license screen |
| Deloitte | 2017 | 2,000 | 91% accept without reading |
| Brookings | 2019 | 2,006 | 71% never or only sometimes read the terms |
Warning
These four figures are not measuring the exact same behavior, and treating them as interchangeable would overstate how directly comparable they are. Deloitte and Brookings asked people to self-report their own habits, which tends to understate skipping since people round toward the behavior they think is expected. Obar and Oeldorf-Hirsch and MeasuringU instead observed real actions, a click or the time spent on a screen, which is why their numbers cluster higher. Read the specific methodology before citing any single figure as “the” acceptance rate.
For the fuller breakdown of who reads terms and conditions in the US and why the numbers vary by age, see our terms and conditions statistics roundup, which covers the Deloitte and Pew survey data behind this comparison in more depth.
Whatever the exact number, every study measured in the last ten years lands well above the halfway mark for people who do not read before they agree.
Is the accept-without-reading rate improving over time?
There is early evidence of movement, but only for cookie consent banners, not full terms of service documents. Usercentrics’ State of Digital Trust 2025 report, based on a Sapio Research survey of 10,000 consumers across the US and five European countries fielded in May 2025, found that 42% now read cookie banners always or often, a share the report frames as rising compared to prior years.
Figure 5: Key milestones in accept-without-reading research, 2016 to 2025. Sources: see citations below.
No comparable, recent large-sample study has re-measured the terms of service acceptance rate specifically since Obar and Oeldorf-Hirsch’s 2020 publication, so it is not possible to say with confidence whether the 93% figure has moved. What the 2025 cookie data does suggest is that consumer behavior can shift when the interface itself changes, regulatory pressure in the EU has pushed cookie banners toward clearer reject options over the same period, and that appears to correlate with more people actually engaging with them.
The Bottom Line
Every methodology used over the last decade, lab experiments, server logs, and self-reported surveys, agrees on the same basic fact: somewhere between 71% and 93% of people accept a terms of service document without reading it, and the number moves with how the behavior is measured rather than whether it happens. The 93% figure from Obar and Oeldorf-Hirsch’s controlled study is the most direct behavioral evidence available, and it holds even when the clauses being missed are deliberately absurd.
For a business publishing its own terms of service, the practical implication is not that legal language is pointless because nobody reads it. It is that the document has to work for the small share of people who do read it, and to disclose fairly to everyone else even when they will not. Writing terms that are current, scannable, and specific to what your business actually does is the more defensible position, and a generated document beats a copy-pasted template written for someone else’s business. You can generate a terms and conditions document that matches your actual business model rather than reusing generic language nobody, including you, will ever reread closely.
Frequently Asked Questions
What percentage of people accept terms of service without reading them? 93% of test subjects agreed to a fictitious social network’s terms of service without reading it closely enough to catch built-in ‘gotcha’ clauses, according to Obar and Oeldorf-Hirsch’s peer-reviewed 2020 study of 543 people. That figure lines up with Deloitte’s 2017 finding that 91% of US consumers accept legal terms without reading them.
Do people notice unusual clauses hidden in terms of service? No. 98% of participants in the Obar and Oeldorf-Hirsch study missed clauses stating their data would be shared with the NSA and their employer, and that they were handing over a first-born child as payment for access.
How long do people spend reading a terms of service page before agreeing? About 51 seconds on average, per the same 543-person study, well short of the 15 to 17 minutes a full terms of service page takes at a normal 250-to-280-word-per-minute reading speed. A separate MeasuringU analysis of roughly 2,500 software users found at least 92% spend under 2 minutes on a license agreement screen.
Is the accept-without-reading rate improving over time? Barely, and only for cookie banners so far. Usercentrics’ State of Digital Trust 2025 report, based on a Sapio Research survey of 10,000 consumers, found 42% now read cookie banners always or often. No comparable recent survey shows the same shift for full terms of service documents.
Sources and References
- Obar, J. A., & Oeldorf-Hirsch, A. (2020). “The Biggest Lie on the Internet: Ignoring the Privacy Policies and Terms of Service Policies of Social Networking Services.” Information, Communication & Society, 23(1). Experimental survey, n=543.
- MeasuringU. “Do Users Read License Agreements?” Log analysis of roughly 2,500 users across several consumer software products, 2016.
- Deloitte. (2017). “Global Mobile Consumer Survey, US Edition.” US cut of 2,000 consumers, part of a 51,000-respondent global survey.
- Brookings. (2019). “Brookings Survey Finds Three-Quarters of Online Users Rarely Read Business Terms of Service.” National survey, n=2,006 US adult internet users, fielded May 8-10, 2019.
- Usercentrics. (2025). “State of Digital Trust Report 2025.” Sapio Research survey of 10,000 consumers across the US and five European countries, fielded May 2025.
Note: All figures verified as of August 2026. The Obar & Oeldorf-Hirsch experimental figure remains the most recent controlled study specifically measuring terms of service acceptance behavior, and this page will be updated if a more current replication is published.
