This document presents measurements of browser metrics based on a crawl of 100 websites. I made three types of measurements on each site once in an unmodified web browser, and once in a web browser using an ad-blocking extension (UBlockOrigin). I found that this ad-blocking extension reduces the number of third party http(s) requests, javascript API calls, and cookie accesses.
From the above plots I can observe that the number of third-party HTTP requests made in the unmodified browser is significantly higher than that of the browser with the ad-blocking extension. The Kernel Density Estimate (KDE) also shows that the browser with ad-blocking is much more likely to make fewer third party HTTP requests.
Without ad-blocking:
Domain | # Third-Party HTTP Requests |
---|---|
ssl-images-amazon.com | 1008 |
pinimg.com | 786 |
msocdn.com | 498 |
alicdn.com | 470 |
redd.it | 416 |
pstatic.net | 411 |
redditstatic.com | 321 |
awsstatic.com | 314 |
qhimg.com | 300 |
doubleclick.net | 297 |
With ad-blocking:
Domain | # Third-Party HTTP Requests |
---|---|
ssl-images-amazon.com | 502 |
pinimg.com | 390 |
redd.it | 280 |
msocdn.com | 249 |
alicdn.com | 197 |
pstatic.net | 170 |
redditstatic.com | 169 |
sinaimg.cn | 151 |
qhimg.com | 150 |
awsstatic.com | 149 |
The most common third party site HTTP requests were sent in both browser instrumentations was ssl-images-amazon.com
which seems to be an image hosting service by amazon. One domain that appears in the "Without ad-blocking" list but does not appear in the other list is doubleclick.net
. Double Click is a company that provides ad serving services.
In the above figure I observe that the ad-blocking browser almost entirely eliminates third party cookie accesses. From an observation of this data.
Without ad-blocking:
Domain | # Third-Party cookie accesses |
---|---|
pubmatic.com | 110 |
doubleclick.net | 73 |
yahoo.com | 65 |
linkedin.com | 48 |
casalemedia.com | 48 |
google.com | 44 |
bing.com | 44 |
microsoft.com | 40 |
youtube.com | 38 |
rubiconproject.com | 34 |
With ad-blocking:
Domain | # Third-Party Cookie Accesses |
---|---|
youtube.com | 19 |
google.com | 17 |
amazon.com | 16 |
live.com | 12 |
bing.com | 11 |
microsoft.com | 11 |
bbc.com | 9 |
aliexpress.ru | 8 |
linkedin.com | 6 |
mmstat.com | 5 |
Without ad-blocking many cookies are accessed by third-parties. Some of these domains such as pubmatic.com, doubleclick.net, and rubiconproject.com are associated with advertising and marketing. With ad-blocking, none of the top third party domains seem to be associated with advertising and marketing.
In the above figures, I observe a similar trend as I did in the other metrics. Javascript API calls by third party scripts are significantly lower.
Without ad-blocking:
Domain | # Third Party JS API Calls |
---|---|
forbesimg.com | 5675 |
wsimg.com | 2696 |
youtube.com | 2607 |
media.net | 2603 |
alicdn.com | 2164 |
itc.cn | 1618 |
segment.com | 1481 |
google-analytics.com | 1112 |
adobedtm.com | 995 |
redditstatic.com | 947 |
With ad-blocking:
Domain | # Third Party JS API Calls |
---|---|
forbesimg.com | 5675 |
wsimg.com | 2092 |
youtube.com | 1271 |
alicdn.com | 893 |
itc.cn | 774 |
redditstatic.com | 497 |
guim.co.uk | 388 |
segment.com | 365 |
microsoft.com | 336 |
bbci.co.uk | 309 |
Unlike the other metrics, the top third-party domains to make Javascript API calls do not vary greatly between the with and without ad-blocking samples. This may suggest that the number of Javascript calls is only a crude indicator of advertising/tracking behavior.