Back to the directoryFind in the timeline
Dark Patterns at Scale: a crawl of 11,000 shopping sites
Mathur and colleagues (2019) crawled about 11,000 shopping sites and found 1,818 dark pattern instances, on about 11.1 per cent of the sites. Luguri and Strahilevitz (2021) later showed in an online experiment that such patterns raised acceptance of a paid offer.
- Where it sits
- Academic psychologyResearch published for other researchers and students.
- Evidence level
- Independent experimentIndependent research, with its design and limits stated.
- Sources inspected
- 3 of 32 with the relevant text, company document or archive item checked, 1 as an abstract or reference article, 0 not verified.
What happened
In 2019 Mathur and colleagues reported a crawl of about 11,000 shopping websites. They analyzed about 53,000 product pages and found 1,818 dark pattern instances of 15 types in 7 categories. According to the project page, these appeared on 1,254 sites, about 11.1 per cent. The authors also report 22 third-party entities offering dark patterns as ready-made tools.
The psychological idea
The authors built a taxonomy of how each pattern influences users and the harm it can cause. Categories include sneaking, urgency, misdirection, social proof, scarcity, obstruction and forced action. A later independent experiment by Luguri and Strahilevitz tested effects on choices. US adults (N = 1,963) were offered a paid identity theft protection plan with no dark patterns, mild ones or aggressive ones.
In practice
Counts from the crawl: low-stock messages 632 instances on 581 sites, countdown timers 393 on 361, and activity messages 313 on 264. Of the 1,818 instances, 234 on 183 sites relied on deception. In the experiment, acceptance among completers was 11.3 per cent in the control condition, 25.8 per cent with mild patterns and 41.9 per cent with aggressive patterns.
Why it mattered
Together the studies give counts of the practice on shopping sites and an estimate of how far such patterns change one kind of choice. They move the term from designers' vocabulary to measured evidence.
Limits and qualifications
The crawl covers shopping sites and the patterns its authors looked for. The Trick Questions definition was narrowed in July 2019, so earlier figures differ, and the link between site popularity and dark pattern use is a correlation. In the experiment nobody was charged and the offer was a survey framing, so the authors say questions of external validity remain. Neither study shows how many real consumers were harmed.
Evidence behind the claims
The abstract reports about 11,000 shopping websites, about 53,000 product pages, 1,818 dark pattern instances in 15 types and 7 categories, 183 sites with deceptive practices and 22 third-party entities offering dark patterns.
The project page reports patterns on 1,254 sites (about 11.1 per cent), 632 low-stock instances on 581 sites, 393 countdown timers on 361 sites, 313 activity messages on 264 sites, and 234 deceptive instances on 183 sites, and says more popular sites by Alexa ranking were more likely to feature dark patterns.
The Trick Questions definition was narrowed in a July 2019 revision, so earlier figures differ.
In Study 1 (N = 1,963) acceptance of a paid identity theft protection plan among completers was 11.3 per cent for control, 25.8 per cent for mild and 41.9 per cent for aggressive dark patterns; counting dropouts as rejections the rates were 11.3, 25.4 and 37.2 per cent.
In Study 2 the authors report that obscuring price information (hidden information) had the greatest effect on overall acceptance and nearly doubled it.
The authors write that a manipulative survey is not identical to an e-commerce interface and that questions of external validity will remain; a footnote says the acceptance figures likely overstate real demand because participants were told they would pay nothing for six months.
People involved
- Arunesh Mathur. Lead author of the 2019 crawl study
- Arvind Narayanan. Co-author of the 2019 crawl study
- Jamie Luguri. Co-author of the 2021 experiments
- Lior Strahilevitz. Co-author of the 2021 experiments
In the Psy101 course
- OpenStax Psychology 2e, section 12.3: Attitudes and PersuasionChapter 12, Social Psychology. Context. The section discusses this topic.The section covers sales tactics such as foot-in-the-door and how communication changes choices.
- OpenStax Psychology 2e, section 2.2: Approaches to ResearchChapter 2, Psychological Research. Method. The section discusses this topic.The section covers experiments as the method that supports cause and effect claims, and correlational research as the weaker design.
This is course relevance only. It is not evidence for the historical or corporate claims on this page.See the full course map.
Sources and inspection
3 of 3 sources were opened and read at the level shown. A source marked as not verified supports no claim on this page by itself.
- [1]Dark Patterns at Scale: Findings from a Crawl of 11K Shopping WebsitesMathur, A; Acar, G; Friedman, M; Lucherini, E; Mayer, J; Chetty, M; Narayanan, A. arXiv 1907.07032 (published in PACM HCI 3, CSCW), 2019. Preprint. Primary source.Abstract or publisher record onlyRead the arXiv abstract page. The full paper was not read.
- [2]Dark Patterns at Scale: Findings from a Crawl of 11K Shopping Websites (project page)Princeton Web Transparency and Accountability Project. Princeton Web Transparency and Accountability Project, 2019. Report. Primary source.Relevant text checkedRead the authors' project page as returned by the fetch tool, including the key figures, category counts and revision notes.
- [3]Shining a Light on Dark PatternsLuguri, J; Strahilevitz, L. Journal of Legal Analysis 13(1): 43-109, Oxford University Press, 2021. Journal article. Primary source.Relevant text checkedRead the Study 1 design and results, parts of Study 2 and the discussion as returned by the fetch tool, covering roughly the first 200,000 of 233,594 characters. The final part was not read.
Access note: Effect sizes are for one task in an online panel. The last part of the Luguri and Strahilevitz page was not read, and the full Mathur paper was not opened.
Independent source review: Claude review agent, independent of the research agent, 2026-10-21. Verdict: pass-with-edits.