Dark Patterns at Scale: Findings from a Crawl of 11K Shopping Websites

Accepted at ACM CSCW 2019

Dark patterns are user interface design choices that benefit an online service by coercing, steering, or deceiving users into making unintended and potentially harmful decisions. We conducted a large-scale study, analyzing ~53K product pages from ~11K shopping websites to characterize and quantify the prevalence of dark patterns.

Read the paper (final version) »

Please see the list of revisions here.

Findings

We discovered 1,818 instances of dark patterns on shopping websites, which together represent 15 types of dark patterns.
These 1,818 dark patterns were present on 1,254 of the ∼11K shopping websites (∼11.1%) in our data set. Shopping websites that were more popular, according to Alexa rankings, were more likely to feature dark patterns.
We demonstrate which of the dark patterns that we discovered rely on consumer deception. In total, we uncovered 234 instances of deceptive dark patterns across 183 websites.
We identify 22 third-party entities that provide shopping websites with the ability to create dark patterns on their sites. Two of these entities openly advertise practices that enable deceptive messages.

Dark pattern categories

Sneaking

Attempting to misrepresent user actions, or delay information that if made available to users, they would likely object to.

1. Sneak into Basket

Adding additional products to users' shopping carts without their consent.

Prevalence: 7 instances across 7 websites.

Sneak into Basket on avasflowers.net. Despite requesting no greeting cards, one worth $3.99 is included.

Sneak into Basket on cellularoutfitter.com. The pre-checked box means the screen protector worth $4.99 ends up in cart.

2. Hidden Costs

Revealing previously undisclosed charges to users right before they make a purchase.

Prevalence: 5 instances across 5 websites.

Hidden Costs on proflowers.com. The Care & Handling charge ($2.99) is disclosed on the last step.

3. Hidden Subscription

Charging users a recurring fee under the pretense of a one-time fee or a free trial.

Prevalence: 14 instances across 13 websites.

Hidden Subscription on wsjwine.com. Selecting the WSJwine Advantage option does not reveal the recurring subscription of $89 unless "Learn More" is clicked on.

Hidden Subscription on ross-simons.com. Joining the VIP Rewards Club does not reveal the recurring subscription of $95 unless "Terms and conditions" is clicked on.

Urgency

Imposing a deadline on a sale or deal, thereby accelerating user decision-making and purchases.

1. Countdown Timer

Indicating to users that a deal or discount will expire using a counting-down timer.

Prevalence: 393 instances across 361 websites.

Countdown Timer on justfab.com. In this instance, the stated offer is valid even after the hour long timer expires.

Countdown Timer in a popup displayed on leesa.com. In this instance, the stated offer is valid even after the half hour long timer expires.

2. Limited-time Message

Indicating to users that a deal or sale will expire will expire soon without specifying a deadline, thus creating uncertainty.

Prevalence: 88 instances across 84 websites.

Limited Time on samsung.com. The website states that the deal is "Limited Time Only" without disclosing the deal's deadline.

Misdirection

Using visuals, language, or emotion to steer users toward or away from making a particular choice.

1. Confirmshaming

Using language and emotion (shame) to steer users away from making a certain choice.

Prevalence: 169 instances across 164 websites.

Confirmshaming on radioshack.com. The option to dismiss the popup is framed to shame the user into avoiding it.

2. Visual Interference

Using style and visual presentation to steer users to or away from certain choices.

Prevalence: 25 instances across 24 websites.

Visual Interference on greenfingers.com. The opt-out option is grayed out to indicate it is disabled or cannot be clicked, when it can.

3. Trick Questions

Using confusing language to steer users into making certain choices.

Prevalence: 9 instances across 9 websites.

Trick Questions on newbalance.co.uk. Normally, checkboxes are designed to be ticked to opt in. In this case however, the user is required to tick to opt out.

4. Pressured Selling

Pre-selecting more expensive variations of a product, or pressuring the user to accept the more expensive variations of a product and related products.

Prevalence: 67 instances across 62 websites.

Pressured Selling on topcoat.store. On adding a product to cart, a popup appears asking the user to upgrade their purchase.

Social proof

Influencing users' behavior by describing the experiences and behavior of other users.

1. Activity Messages

Informing the user about the activity on the website (e.g., purchases, views, visits).

Prevalence: 313 instances across 264 websites.

Activity Notification on thredup.com highlighting the names and locations of those who purchased the product. The message always signals sold products as "just saved" by customers even though the products have been sold for a long time.

Activity Notification on jcpenney.com highlighting the number of people who viewed the product in the last day.

2. Testimonials of Uncertain Origin

Testimonials on a product page whose origin is unclear.

Prevalence: 12 instances across 12 websites.

Testimonials on spanx.com. The website does not disclose how these were sourced, or whether they were submitted by actual customers.

Scarcity

Signaling that a product is likely to become unavailable, thereby increasing its desirability to users.

1. Low-stock Message

Indicating to users that limited quantities of a product are available, increasing its desirability.

Prevalence: 632 instances across 581 websites.

Low-stock Message on 6pm.com. Choosing product options shows *Only 3 left in stock*. The out-of-stock message makes it always seem that the item just sold out.

Low-stock Message on orthofeet.com. The message does not disclose the quantity in stock to users and appears for all products on the website.

2. High-demand Message

Indicating to users that a product is in high-demand and likely to sell out soon, increasing its desirability

Prevalence: 47 instances across 43 websites.

High Demand on fashionnova.com. The message is shown for all products on the website.

High Demand on pacificcoast.com. The message is shown for all products on the website.

Obstruction

Making it easy for the user to get into one situation but hard to get out of it.

1. Hard to Cancel

Making it easy for the user to sign up for a recurring subscription but cancellation requires emailing or calling customer care.

Prevalence: 31 instances across 31 websites.

Hard to Cancel on savagex.com. The only way to cancel the $49.95 auto-renewing membership is to call the customer support. Signing up, on the other hand, can be completed online.

Hard to Cancel on 1800flowers.com. The only way to cancel the $19.99 auto-renewing membership is to email customer care unlike signing up, which can be completed online.

Forced Action

Forcing the user to do something tangential in order to complete their task.

1. Forced Enrollment

Coercing users to create accounts or share their information to complete their tasks.

Prevalence: 6 instances across 6 websites.

Forced Action on therealreal.com. Even viewing products requires signing up and creating and account.

Forced Action on musiciansfriend.com. Agreeing to the terms of use also requires agreeing to receive emails and promotions.

Third parties that enable Social Proof Activity Notifications

Prevalence of third parties that enable Social Proof Activity Notifications are given below (based on our data set of 11K shopping websites and Princeton Web Census crawls of home pages of Alexa Top million websites). Where available, we list additional dark patterns the third parties claim to offer.

Third-party Entity	# Shopping Websites	# Alexa Top Million	Additional Dark Patterns
Beeketing	406	4,151	Pressured Selling, Urgency, Scarcity
Dynamic Yield	114	416	Urgency
Yieldify	111	323	Urgency, Scarcity
Fomo	91	663	--
Fresh Relevance	86	208	Urgency
Insider	52	484	Scarcity, Urgency
Bizzy	33	213	--
ConvertCart	31	62	--
Taggstar	27	4	Scarcity, Urgency
Qubit	25	73	Pressured Selling, Scarcity, Urgency
Exponea	18	180	Urgency, Scarcity
Recently	14	66	--
Proof	11	508	--
Fera	11	132	Pressured Selling, Scarcity, Urgency
Nice	10	80	--
Woocommerce Notification	10	61	--
Bunting	5	17	Urgency, Scarcity
Credibly	4	67	--
Convertize	3	58	Scarcity, Urgency
LeanConvert	2	0	--
Boost	1	3	--
Amasty	1	0	Pressured Selling, Scarcity, Urgency

Revisions

Update (July 17, 2019): We have released our paper on arXiv. The previous version can be found here.

Update (July 1, 2019): In our previous release, we had marked all opt-out choices accompanied by checkboxes as Trick Questions dark patterns. Our classification was guided by the definition of Trick Questions provided by Gray et al.. Since then we have narrowed our definition to only include opt-out choices that are misleading, such as when the user has to check a box and the text began with an affirmative statement about the undesirable practice. We have updated the paper and the CSV file (previous version of the paper and data can be found here and here). We have also made minor proofreading edits in the current version of the paper.

Citation

   @article{Mathur2019DarkPatterns,
	title        = {Dark Patterns at Scale: Findings from a Crawl of 11K Shopping Websites},
	author       = {Mathur, Arunesh and Acar, Gunes and Friedman, Michael and Lucherini, Elena and Mayer, Jonathan and Chetty, Marshini and Narayanan, Arvind},
	year         = 2019,
	journal      = {Proc. ACM Hum.-Comput. Interact.},
	publisher    = {ACM},
	volume       = 1,
	number       = {CSCW},
	issue_date   = {November 2019}
   }

Arunesh Mathur	`[email protected]`
Gunes Acar	`[email protected]`
Michael Friedman	`[email protected]`
Elena Lucherini	`[email protected]`
Jonathan Mayer	`[email protected]`
Marshini Chetty	`[email protected]`
Arvind Narayanan	`[email protected]`