98% Click Agree Without Reading the Fine Print [2026]

In a controlled field experiment, 98% of 543 participants clicked to join a fake social network without noticing hidden gotcha clauses buried in the agreement they had just accepted, including a clause demanding a first-born child as payment, according to Obar and Oeldorf-Hirsch's study published in Information, Communication & Society (2020). That behavioral result lines up with a much larger self-reported figure: Deloitte's 2017 Global Mobile Consumer Survey of 2,000 US adults found 91% say they accept terms and conditions without reading them. Below is what happens when researchers, and companies themselves, actually test the click-to-agree moment rather than just asking about it.

What happened in the real experiment that measured clicking agree?

Researchers Jonathan Obar (York University) and Anne Oeldorf-Hirsch (University of Connecticut) built a working fake social network called NameDrop and recruited 543 participants to sign up for it. The sign-up screen offered a choice between a "quick join" clickwrap that skipped straight to account creation, and a detailed review option that displayed the full privacy policy and terms of service.

74% of participants chose quick join for the privacy policy, skipping the document without opening it at all. Among everyone who reached the terms, the researchers had quietly inserted deliberately absurd gotcha clauses, including a promise to share data with the NSA and with the participant's employer, and a clause requiring a first-born child as payment for the service. 98% of participants missed these clauses entirely and clicked through to join.

Measure	Result	Sample
Chose quick join, skipped privacy policy	74%	543 participants
Missed the hidden gotcha clauses	98%	543 participants
Average time spent on privacy policy	73 seconds	Subset who opened it
Average time spent on terms of service	51 seconds	Subset who opened it

Table 1: Core results from the NameDrop field experiment. Source: Obar and Oeldorf-Hirsch, "The Biggest Lie on the Internet" (2020), Information, Communication & Society.

Secondary reporting of this study shows some variance in the exact figures cited, with a handful of sources putting the privacy-policy skip measure closer to 78% and the average terms-of-service reading time as high as 81 seconds. Treat 74% to 78% and 51 to 81 seconds as the reported range rather than a single fixed figure, since the underlying paper covers more than one study condition. Every version of the result points to the same conclusion: almost no one who reaches a clickwrap screen reads what is actually there.

Figure 2: Four separate studies, four different methods, all landing in the same range. Sources: Obar and Oeldorf-Hirsch (2020), Deloitte (2017), GameStation April Fools promotion (2010), Brookings Institution (2019).

What happened when a real company tested a joke clause?

UK retailer GameStation ran the closest thing to a live version of the NameDrop experiment, without telling anyone it was an experiment. On April Fools' Day 2010, the company added an "immortal soul" clause to its terms and conditions, granting GameStation a non-transferable option to claim the customer's soul, and included an opt-out checkbox that would have exempted anyone who noticed and ticked it.

Only 12% of customers who transacted that day spotted the clause and opted out. The company later said it had legally acquired the souls of roughly 7,500 customers who agreed to the amended terms without ticking the box, and used the results in a press statement about how few shoppers read website terms before checking out.

Figure 3: Real transaction data, not a lab simulation. Source: GameStation April Fools promotion, widely reported including by Pinsent Masons and Fox News (2010).

GameStation's clause was a publicity stunt with no legal weight, but the opt-out mechanic makes it a genuine behavioral measurement: customers had to actively notice a change and click a box to avoid the joke clause, and 88% did not. That is close enough to the NameDrop study's 98% miss rate and Deloitte's 91% self-reported figure that three independent methods, an academic experiment, a real transaction test, and a national survey, now agree on the same rough order of magnitude.

How does the reading time required compare to the time people actually spend?

The gap between required and actual reading time is where the click-without-reading pattern becomes easiest to see. In the NameDrop study, reading the privacy policy at a normal adult pace of 250 to 280 words per minute would take 29 to 32 minutes, and the terms of service would take 15 to 17 minutes. Participants who opened the documents at all spent an average of 73 seconds on the privacy policy and 51 seconds on the terms of service, a small fraction of either estimate.

Figure 4: The drop from required reading time to actual time spent, converted to minutes. Source: Obar and Oeldorf-Hirsch (2020).

At 73 seconds, a participant could read roughly 300 to 340 words of a document that, at typical policy density, runs into the thousands of words. That is consistent with a related finding: if you generate a terms of service document with short, clearly labeled sections instead of one dense wall of text, at least the small share of users who do open it stand a realistic chance of reading the part they care about in the time they are actually willing to spend.

What does the click-to-agree sign-up flow actually look like?

Every experiment above shares the same structural setup: a user reaches a screen with a document link and an "I agree" action, and almost never opens the document before taking that action. The NameDrop study's quick-join option formalized this as an explicit choice, but most real sign-up flows do not offer a visible alternative at all, just a checkbox or button next to a link.

Figure 5: The clickwrap path most sign-up flows follow, whether or not a review option is offered. Source: Obar and Oeldorf-Hirsch (2020).

The practical point for a site owner is that offering a detailed-review option, as NameDrop did, does not meaningfully change the outcome. Even the 26% who opened the document in that study still averaged under a minute and a half on it, well short of the 15-to-32-minute range actually needed. This same pattern shows up across the wider terms of service landscape most sites publish, covered in full in the terms of service reading, length, and trust statistics for 2026.

How has click-without-reading research developed since 2005?

The pattern documented above is not a recent discovery. Independent researchers and at least one retailer have been measuring it for two decades, using different methods that keep landing on the same conclusion.

Figure 6: Five separate research efforts spanning fifteen years, all confirming the same behavior. Sources: PC Pitstop (2005), McDonald and Cranor (2008), GameStation (2010), Deloitte (2017), Brookings Institution (2019), Obar and Oeldorf-Hirsch (2020).

The oldest and most literal test of this behavior may be PC Pitstop's 2005 experiment. The company added a clause to its software license agreement offering a cash reward to whoever emailed the company after reading it. It took more than 3,000 sales and about five months before a single customer, Doug Heckman, noticed the clause and claimed a $1,000 check. That is not a percentage in the strict sense, since PC Pitstop did not publish the exact total number of buyers during that window, but it is a real, independently reported result showing the same behavior other studies later measured with formal sampling.

The Bottom Line

Three independent methods, an academic field experiment, a real retailer's own transaction data, and a national consumer survey, all converge on the same finding: somewhere between 88% and 98% of people click agree without reading what they are agreeing to, whether the research measures self-reported behavior, a lab-style clickwrap choice, or a live opt-out box on a real checkout page. The reading-time data explains why. A privacy policy or terms of service that genuinely needs 15 to 32 minutes to read gets, on average, well under two minutes of attention from the small share of people who open it at all. For a business publishing its own terms, the reading behavior is not going to change. What is within your control is the document itself: shorter, more clearly organized terms give the minority who do click through a real chance of finding the clause that matters to them in the time they are actually willing to spend.

Frequently Asked Questions

How many people click agree without reading terms of service? In a controlled field experiment, 98% of 543 participants agreed to join a fake social network without noticing hidden data-sharing and first-born child clauses buried in the terms, per Obar and Oeldorf-Hirsch's study published in Information, Communication & Society (2020). Separately, Deloitte's 2017 survey of 2,000 US adults found 91% self-report accepting terms without reading.

What was the NameDrop experiment and what did it find? Researchers Jonathan Obar and Anne Oeldorf-Hirsch built a fake social network called NameDrop for 543 participants to join. 74% chose a quick-join option that skipped the privacy policy entirely, and 98% missed deliberately absurd gotcha clauses inserted into the terms, including data sharing with the NSA and employers, and a clause demanding a first-born child as payment.

Has any company actually tested this with a real clause? Yes. UK retailer GameStation added a joke "immortal soul" clause to its terms on April Fools' Day 2010. Only 12% of customers ticked the opt-out box that would have exempted them, meaning roughly 7,500 shoppers agreed to sign over their souls without reading the added clause.

How does required reading time compare to actual time spent? In the NameDrop experiment, reading the privacy policy properly would take 29 to 32 minutes and the terms of service 15 to 17 minutes at a normal adult reading pace, but participants spent an average of roughly 73 seconds on the privacy policy and 51 seconds on the terms of service before clicking through.

Where the Numbers Come From

1. Obar, Jonathan A. and Oeldorf-Hirsch, Anne. (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). Field experiment of 543 participants; 74% skipped the privacy policy via a quick-join clickwrap, 98% missed embedded gotcha clauses, average time spent 73 seconds on the privacy policy and 51 seconds on the terms of service.
2. Deloitte. (2017). "Global Mobile Consumer Survey." Survey of 2,000 US adults; 91% accept terms and conditions without reading, rising to 97% for ages 18 to 34.
3. Pinsent Masons. (2010). "Gamers Agree to Give Up Their Immortal Souls." GameStation's April Fools' Day 2010 "immortal soul" clause; 12% of customers opted out, implying roughly 7,500 accepted the clause unread.
4. Brookings Institution. (2019). "Brookings Survey Finds Three-Quarters of Online Users Rarely Read Business Terms of Service." Survey of 2,006 US internet users conducted May 8 to 10, 2019; 32% never read terms of service.
5. Marginal Revolution, citing PC Pitstop. (2005, as reported). PC Pitstop's license-agreement experiment offering a $1,000 reward for reading the terms; more than 3,000 sales and about five months passed before a customer claimed it.

Note: All figures verified as of July 2026. Secondary reporting of the Obar and Oeldorf-Hirsch study shows minor variation between sources (74% to 78% for the privacy-policy skip rate, 51 to 81 seconds for average terms-of-service reading time), reflecting the paper's multiple study conditions; figures here use the most commonly cited values. Statistics are refreshed at least twice a year as newer research becomes available.