Speed test: does ad blocking make your phone faster?

Where we tested, yes. Across 460 websites from 23 countries, on a Redmi Note 12 with the Ads category list on, pages downloaded a median 27% less data, used 36% less processor time and 60% less JS memory, and loaded in 47% less time. The difference varies by country: outside the European Union data fell by 35%; on EU sites, opened without touching the consent banner, by 13%.

Measured: October 3, 2026Phone: Redmi Note 12

Websites, on a real phone

Redmi Note 12 (Snapdragon 685, 8 GB), Chrome, home Wi-Fi. 20 popular, ad-heavy websites from each of the 23 countries where the app’s languages are spoken; 460 in total. Each site opened once with the list off and once with it on, 45 seconds on the page. Median difference per site.

  • 27%less dataMiddle half of sites: 8% to 50% less
  • 42%fewer requestsMiddle half of sites: 14% to 68% less
  • 36%less processor timeMiddle half of sites: 12% to 59% less
  • 60%less JS memoryMiddle half of sites: 27% to 77% less
  • 47%shorter load timeMiddle half of sites: 28% to 66% less

Why is it faster?

  • An ad that is blocked is not downloaded: ad images, videos and tracking scripts on our lists do not reach your phone.
  • The page opens with fewer pieces. The processor and memory are not busy drawing ads or running the scripts that auction ad slots.
  • Your data goes only to the content you opened. On a slow connection the difference shows most in load time.
  • In apps and games, the ad networks on our lists cannot be reached either, so no ads download from them.

By country

20 websites from each country. The difference varies with how many ads the sites load and with consent banner rules.

United States: across the 20 websites we tested, a median 33% less data, 69% fewer requests, 47% less processor time and 74% less JS memory.

Less data per country (median per site)

  1. India 55%
  2. Turkey 52%
  3. Ukraine 47%
  4. Indonesia 43%
  5. Japan 41%
  6. Brazil 40%
  7. Romania EU 36%
  8. Vietnam 35%
  9. Thailand 34%
  10. United States 33%
  11. France EU 28%
  12. Egypt, Saudi Arabia 28%
  13. Russia 24%
  14. Spain EU 24%
  15. Taiwan, Hong Kong 23%
  16. Malaysia 23%
  17. Czechia EU 16%
  18. Italy EU 16%
  19. South Korea 14%
  20. Germany EU 14%
  21. Netherlands EU 9%
  22. Sweden EU 6%
  23. Poland EU 2%

Sorted by the data difference. On sites in countries marked EU, few ads load before consent is given; see the note below.

Country table

CountrySitesDataRequestsProcessorJS memoryTotal data (MB)
India2055% less57% less54% less75% less309 → 20633% less
Turkey2052% less70% less62% less85% less193 → 10148% less
Ukraine2047% less69% less64% less78% less118 → 6346% less
Indonesia2043% less56% less51% less68% less257 → 16336% less
Japan2041% less62% less49% less63% less100 → 5842% less
Brazil2040% less65% less49% less82% less184 → 14819% less
Romania (EU)2036% less17% less22% less52% less193 → 15420% less
Vietnam2035% less53% less35% less66% less236 → 17327% less
Thailand2034% less65% less57% less73% less262 → 22813% less
United States2033% less69% less47% less74% less270 → 19328% less
France (EU)2028% less15% less20% less31% less91 → 7517% less
Egypt, Saudi Arabia2028% less49% less39% less73% less150 → 9636% less
Russia2024% less39% less35% less51% less126 → 10021% less
Spain (EU)2024% less28% less41% less40% less152 → 11723% less
Taiwan, Hong Kong2023% less45% less33% less65% less288 → 23020% less
Malaysia2023% less62% less48% less68% less207 → 17317% less
Czechia (EU)2016% less24% less19% less61% less148 → 12019% less
Italy (EU)2016% less15% less20% less31% less103 → 9210% less
South Korea2014% less22% less22% less48% less197 → 17412% less
Germany (EU)2014% less12% less24% less15% less126 → 7341% less
Netherlands (EU)209% less11% less7% less14% less83 → 7115% less
Sweden (EU)206% less2% less5% less22% less240 → 27514% more
Poland (EU)202% less7% less7% less10% less93 → 7025% less
Outside the EU28035% less56% less47% less68% less2,903 → 2,11127% less
European Union18013% less15% less15% less27% less1,232 → 1,05115% less
All46027% less42% less36% less60% less4,135 → 3,16324% less

Why did some sites use more data?

  • On 44 of the 460 sites, more data was downloaded with the list on. We did not drop those rows; we measured them again to find out why.
  • We measured the 20 sites with the largest increase 3 more times on the same phone. On 12 sites the increase came back every time; on 8 it varied from run to run.
  • The extra data is not ads; it is the site’s own content. We looked at which addresses these sites requested: the extra went to the site’s own video player, video delivery network and image server. When the pre-roll ad is blocked, the news video starts right away and keeps downloading for 45 seconds; the page also loads sooner, so more images open in the same time.
  • As a control we measured 20 randomly chosen sites 2 more times: on 18 of them data went down with the list on every time, 20% less in total. The overall result held in the repeat.
  • On c*n.com the number of requests went up with the list on (2,571-3,564 instead of 622-664 over 3 repeats). All of the extra requests were retries to a video ad server on our lists (cdn-media.brightline.tv): the page retries the server it cannot reach thousands of times in 45 seconds. Nothing downloads from a blocked address, so these retries carry no data. Requests to addresses not on the list went from 180-183 to 139, data from 27.3-28.9 MB to 24.0 MB.
  • In our first test (12 sites) da******l.co.uk downloaded more data with the list on. We measured it 3 more times; in 3 of them data went down with the list on (median: off 9.0 MB, on 8.3 MB). The first result was a one-off.

All measurements, site by site

Site names are partly masked. Open a country to see its sites: list off → on, and the difference.

India 20 sites · 55% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Turkey 20 sites · 52% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Ukraine 20 sites · 47% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Indonesia 20 sites · 43% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Japan 20 sites · 41% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Brazil 20 sites · 40% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Romania 20 sites · 36% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Vietnam 20 sites · 35% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Thailand 20 sites · 34% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

United States 20 sites · 33% less
SiteData (MB)RequestsProcessor (s)JS memory (MB)Load time (s)
us*****y.com 6.2 → 0.690% less 876 → 22774% less 20.3 → 3.284% less 109 → 893% less 15.1 → 0.994% less
az*****s.com 9.0 → 1.187% less 745 → 5792% less 18.9 → 1.393% less 65 → 790% less not done → 1.1
cb******s.com 18.7 → 7.659% less 1,002 → 31668% less 40.8 → 20.949% less 159 → 1888% less 14.2 → 4.667% less
hu*****t.com 13.4 → 6.254% less 1,815 → 29184% less 36.5 → 15.059% less 182 → 2686% less not done → 3.1
bu************r.com 8.4 → 3.954% less 1,136 → 21481% less 31.0 → 12.560% less 129 → 1985% less 0.8 → 0.535% less
ge***s.com 6.8 → 3.450% less 1,163 → 11890% less 17.9 → 2.486% less 105 → 1387% less 2.7 → 2.120% less
fo********k.com 4.6 → 2.448% less 385 → 12169% less 12.6 → 3.572% less 63 → 1575% less 1.2 → 1.18% less
bu*****d.com 10.2 → 5.447% less 1,248 → 21583% less 30.4 → 9.469% less 106 → 2577% less 5.8 → 4.032% less
fo***s.com 15.8 → 9.341% less 1,080 → 25676% less 38.7 → 24.637% less 162 → 4374% less 14.3 → 6.654% less
al*******s.com 4.6 → 3.133% less 509 → 18264% less 15.8 → 9.043% less 70 → 2170% less 1.7 → 1.230% less
we****r.com 3.2 → 2.133% less 190 → 16812% less 5.2 → 4.416% less 25 → 1827% less 5.5 → 4.225% less
fo****s.com 16.4 → 12.723% less 865 → 26869% less 38.7 → 30.621% less 108 → 2577% less 16.0 → 6.162% less
t*z.com 11.4 → 9.517% less 702 → 27161% less 24.5 → 16.234% less 86 → 4054% less 7.7 → 4.245% less
ny***t.com 48.1 → 40.915% less 725 → 20472% less 22.9 → 8.662% less 85 → 2175% less 9.8 → 6.138% less
ne*****k.com 9.4 → 8.014% less 280 → 19331% less 15.7 → 11.626% less 63 → 3741% less 7.8 → 4.147% less
c*n.com 27.9 → 24.412% less 643 → 2,542295% more 38.0 → 40.98% more 94 → 3860% less not done → 4.6
pe***e.com 12.9 → 11.610% less 858 → 30265% less 29.7 → 11.362% less 99 → 3961% less 2.4 → 2.28% less
e**n.com 6.7 → 6.19% less 253 → 18228% less 7.7 → 5.924% less 39 → 2829% less 6.7 → 1.874% less
c**t.com 13.4 → 12.75% less 323 → 26119% less 7.8 → 6.911% less 28 → 2318% less 2.0 → 1.527% less
eo****e.com 23.0 → 22.81% less 343 → 17150% less 13.5 → 7.346% less 45 → 2056% less 6.9 → 2.760% less
France 20 sites · 28% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Egypt, Saudi Arabia 20 sites · 28% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Russia 20 sites · 24% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Spain 20 sites · 24% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Taiwan, Hong Kong 20 sites · 23% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Malaysia 20 sites · 23% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Czechia 20 sites · 16% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Italy 20 sites · 16% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

South Korea 20 sites · 14% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Germany 20 sites · 14% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Netherlands 20 sites · 9% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Sweden 20 sites · 6% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Poland 20 sites · 2% less

The table loads when you open it. Every site’s measurements in one file: world-real-device.csv

Total data: 4,135.9 MB → 3,163.4 MB, 24% less.

A low-end phone, emulated: slow processor, slow 4G

On a computer we slowed Chrome down 10 times and throttled the network to slow 4G (150 ms, 1.6 Mbit/s). Across the 459 of the same websites we could measure, with the list on pages used a median 37% fewer requests, 36% less processor time and 50% less JS memory.

With the list off, 172 sites did not finish loading within 45 seconds. With the list on, 109 of them finished in 2 to 45 seconds and 63 still did not.

The data difference looks smaller in this run (median 16%): at 1.6 Mbit/s, at most about 9 MB can arrive in 45 seconds. With the list on, the freed-up connection goes to the page’s own content.

Games

Redmi Note 12, 11 games from Google Play, EZOBLOCKER 1.0.0. Each game was opened twice with EZOBLOCKER off and twice with it on, and left untouched for 60 seconds. Median difference per game.

  • 99%less data (downloaded by the game)Range: 95% to 99.7% less
  • 10%less processor timeRange: 1% more to 55% less
  • 10%less memoryRange: 2% to 30% less

With EZOBLOCKER on, the ad and tracker networks on our lists could not be reached. The data each game downloaded fell in all 11 games (95% to 99.7%). The difference in processor and memory varied a lot by game: none in some, processor time 55% shorter in another.

We show the games by genre, not by name; names and package IDs are in the raw data below.

GameDownloaded by the game (MB)Processor (s)Memory (MB)
Puzzle 1 15.02 → 0.0399.7% less 61.5 → 42.331% less 825 → 62824% less
Puzzle 2 24.53 → 0.1399% less 95.5 → 52.245% less 725 → 50830% less
Puzzle 3 9.31 → 0.0699% less 76.1 → 68.410% less 1,310 → 1,17710% less
Puzzle 4 10.20 → 0.1199% less 54.1 → 54.61% more 753 → 7175% less
Puzzle 5 14.07 → 0.7295% less 52.7 → 44.116% less 654 → 6048% less
Hyper-casual 1 10.97 → 0.1399% less 64.4 → 60.46% less 775 → 69311% less
Hyper-casual 2 12.05 → 0.1699% less 70.7 → 68.73% less 782 → 7632% less
Hyper-casual 3 12.61 → 0.1799% less 141.1 → 68.951% less 868 → 8176% less
Hyper-casual 4 12.51 → 0.2198% less 78.6 → 77.32% less 1,024 → 9715% less
Word 3.45 → 0.0299.5% less 44.0 → 39.810% less 193 → 15619% less
Arcade 21.34 → 0.8696% less 125.0 → 56.055% less 639 → 56711% less

All 11 games together, EZOBLOCKER off and on: 146.1 MB → 2.6 MB.

How we measured

Websites

  • Site selection: popular, ad-heavy news, sports, classifieds and recipe sites in each of the 23 countries where the app’s languages are spoken. Candidates were first scanned for 20 seconds; sites that did not open, hit a bot or consent wall, or sent fewer than 5 requests to our list were dropped. 20 sites per country remained.
  • Each site was opened twice: list off and list on. With the list on, requests to domains on the list behind the Ads category were cut off the way they are when DNS gives no answer. That is the in-browser equivalent of what our servers do.
  • We stayed on the page for 45 seconds and scrolled down a little every second, so ads that load on scroll were measured too. No cookie banner was clicked and no account was signed in.
  • Measured: data downloaded, number of requests, time the processor was busy, JavaScript time and memory, and the time until the page finished loading (“not done” if it did not finish in 45 seconds).
  • Real phone: Redmi Note 12, Android 15, Chrome 154; Chrome’s cache was cleared before each measurement. Emulation: Chrome on a Mac mini M4, 360x760 mobile view, processor 10 times slower, network 150 ms and 1.6 Mbit/s.
  • Repeats: sites whose data went up with the list on, randomly chosen control sites and the two outliers of the first test were measured again on the same phone; for some we also recorded how many requests went to which address.
  • Our first test (12 sites, mostly from Turkey) showed a median 68% less data. This broader, more varied test is the source of the numbers on this page; the first test’s files are below too.

Games

  • Redmi Note 12, EZOBLOCKER 1.0.0 (store build), our real servers. All nine categories on, Family lock off.
  • In each run the game was closed, opened again and left untouched for 60 seconds. Off and on runs were interleaved (ABBA) so the order would not skew the result.
  • Measured: data downloaded by the game itself (Android’s per-app network statistics), processor time of the game’s processes, and memory (PSS).
  • A run was invalid if the game was not in the foreground at 60 seconds (for example, a notification permission dialog stayed on top). 8 runs were left out for that reason; they are still in the raw data.

Ad list: HaGeZi Multi PRO, version 2026.1002.0908.48, 227,572 entries, license GPL-3.0.

Raw data

Every number on this page was calculated from these files. After downloading, you can check with SHA256SUMS that the files are unchanged.

  • README.txt What each file contains 2.1 KB · SHA-256 0002b2ccfd5b3804…
  • games-runs.jsonl Games: raw line for every run (invalid ones included) 15.3 KB · SHA-256 70fa6946a2564010…
  • games-summary.csv Games: means per game, with names 0.9 KB · SHA-256 4bdee0674cf3b58c…
  • web-emulated.csv First test, low-end emulation: table per site 1.7 KB · SHA-256 4332d9eff49401bf…
  • web-emulated.json First test, low-end emulation: raw results 10.1 KB · SHA-256 28f419a1c466b6b5…
  • web-real-device.csv First test, real phone: table per site 1.3 KB · SHA-256 332fe1b8999229e1…
  • web-real-device.json First test (12 sites), real phone: raw results 6.7 KB · SHA-256 83b467388cf3befa…
  • world-countries.csv Websites: medians and totals per country 1.5 KB · SHA-256 0ea53f1422f3ab04…
  • world-emulated.csv Websites, low-end emulation: table per site 42.9 KB · SHA-256 4e981523dc4853e8…
  • world-emulated.json Websites, low-end emulation: raw results 376.4 KB · SHA-256 d8bfd3c90ae04989…
  • world-real-device.csv Websites, real phone, 23 countries: table per site 44.5 KB · SHA-256 fe2f4ddfe9d27024…
  • world-real-device.json Websites, real phone, 23 countries: raw results 376.5 KB · SHA-256 ea395d58d68170c7…
  • world-repeat.json Websites: repeat runs and requested addresses 170.5 KB · SHA-256 24a350ca3e380986…
  • SHA256SUMS File checksums 1.1 KB

We keep the measurement scripts, the game screenshots and the original logs unchanged.

Questions people ask

Does ad blocking make my phone faster?

Where we tested, yes. Across 460 websites from 23 countries, on a Redmi Note 12, pages downloaded a median 27% less data, used 36% less processor time and loaded in 47% less time. The result varies by country and site: outside the European Union data fell by 35%, in the EU by 13%.

Why is the difference smaller in Europe?

Most sites in the European Union load only a small part of their ads until the consent banner is accepted, and we did not touch the banner. So across 180 EU sites data fell by a median 13%, across 280 sites outside the EU by 35%.

Will EZOBLOCKER slow down my internet?

Only the small DNS lookup goes to our servers; pages and videos load directly from their own servers. In our measurement pages opened with fewer pieces, so load time was a median 47% shorter.

Does it affect battery life?

We did not measure battery, so we give no battery number. The processor time we measured on the web fell by a median 36%; a processor that works less usually uses less energy.

Does it make a difference in games?

In the 11 games we tested, the data each game downloaded fell by a median 99%, because the ad and tracker networks on our lists could not be reached. The difference in processor and memory varied by game (median 10% and 10%).

Can I check the results myself?

Yes. You can download the raw data from this page; each file has its SHA-256 checksum next to it. You can also compare the same page on your own phone with EZOBLOCKER on and off.

App Store Google Play