How Many Websites Use Google AdSense? 1.28M Sites and What adsbygoogle.js Proves
As of November 16, 2026, webatla counts 1,282,226 websites running Google AdSense, 0.39% of the 325.8 million websites it tracks. That works out to roughly one site in 254. AdSense is easy to spot from outside, because
As of November 16, 2026, webatla counts 1,282,226 websites running Google AdSense, 0.39% of the 325.8 million websites it tracks. That works out to roughly one site in 254.
AdSense is easy to spot from outside, because its loader and markup are documented and the publisher ID is printed right in the page. It is also easy to misread. A page can carry the full integration and earn nothing, and a page that earns every day can keep the code out of a scanner's view.
The live count is on webatla's Google AdSense technology page. The longer write-up, with country and domain extension tables, is the Google AdSense statistics article on the webatla blog.
Three ways to divide 1,282,226
A percentage is only as useful as its denominator, and the AdSense count changes meaning depending on what you divide it by.
The same 1,282,226 AdSense sites are 0.39% of all websites, 74.6% of publisher ad tool installs and 6.4 for every 100 Google Analytics sites.Every website. The 325,803,430 base includes parked domains, redirects, placeholders, online shops and company pages, most of which have no reason to show third-party ads. Against that base, 0.39% is small by construction.
Publisher ad tools. This is the base that answers questions about Google AdSense market share. The group here is the 15 biggest tools in webatla's Advertising category that work for the publisher, not the advertiser. Four are Google's (AdSense, DoubleClick Ad Exchange or AdX, Google Publisher Tag, and DoubleClick for Publishers or DFP). Two are recommendation widgets (Taboola and Outbrain). One is WordAds. The other eight are AppNexus, PubMatic, Ezoic, Mediavine, MGID, JuicyAds, TrafficStars and Media.net. Their counts add up to 1,719,786 installs, and AdSense accounts for 74.6%. A site running two of these tools counts twice. Advertiser tags such as Google Ads stay out, since they measure campaigns instead of selling space.
There is a trap in that list that applies to any technology dataset. webatla's DFP page describes DFP (14,565 sites) and Google Publisher Tag (14,566 sites) as one deployment seen through two fingerprints, and says the counts should not be added. Treat DFP as an alias, take it out of the base, and the AdSense share becomes 75.2%. The difference is small here. Still, before summing rows from any technology list, check whether two of them describe the same thing.
Google's measurement tags. Google Analytics is on 20,171,372 sites, so there are about 6.4 AdSense sites for every 100 Analytics sites. Google Tag Manager is on 7,226,750, which outnumbers AdSense by about 5.6 to 1. Far more sites measure traffic with Google than sell ad space through it.
Markers of an AdSense integration you can look for
The integration is small. A site loads Google's script adsbygoogle.js from pagead2.googlesyndication.com, usually with the publisher ID in a client query parameter. Publisher IDs in the page start with ca-pub-. Ad units placed by hand are ins elements with the class adsbygoogle plus data-ad-client and data-ad-slot attributes, each followed by a short inline script that pushes onto the window.adsbygoogle array. With Auto ads the loader alone is enough, and Google picks the positions.
Two more markers often travel with it. One is ads.txt at the root of the domain, the IAB Tech Lab file in which a publisher lists who may sell its inventory. The AdSense entry there names google.com, the publisher ID without the ca- prefix, and the relationship DIRECT. The other is a google-adsense-account meta tag, one of the ways AdSense lets an owner verify a site.
Treat these as checks you can run yourself. They describe a typical AdSense setup, and they are not a description of how webatla's scanner works.
What a detection proves is narrow, and the AdSense technology page is frank about it. The tag shows installation. It does not show an active, approved or paying account. A snippet can outlive the account in an old theme, and a new site carries the code while it waits for review. From outside, both look the same as a site that earns every day.
Each row also has its own date. webatla's scan records, per domain, the day the technologies were last checked. On the five preview rows of the AdSense page those dates run from October 18 to November 15, 2026, while the count itself is dated November 16. A row describes the page on its check date.
Checking one domain by hand
For a single site you need neither a crawler nor an account. Start with the HTML the server sends and with ads.txt. The first line counts publisher IDs and loader references in the raw response, without running any JavaScript. The second prints the google.com lines marked DIRECT in ads.txt, if the site publishes the file.
curl -sL https://example.com/ | grep -oE 'ca-pub-[0-9]+|pagead2\.googlesyndication\.com' | sort | uniq -c
curl -sL https://example.com/ads.txt | grep -iE '^google\.com,.*DIRECT'
A hit in the first command means the code is in the server's response. An empty result proves less. Some servers return a different page to unknown clients, and sites that inject the loader through a tag manager hide it from curl, which never executes JavaScript.
The second check does not depend on the page, and it needs a careful reading. A google.com line by itself does not mean AdSense. Ad Manager uses the same domain, and ad networks that resell a site's space through Google add google.com lines marked RESELLER. The line for the site's own AdSense account is marked DIRECT and carries the same publisher ID as the page. If that line is there and the HTML shows nothing, the tag may arrive another way, or the file may have outlived the integration.
For the rendered page, open DevTools, answer the consent prompt if there is one, scroll a bit, and paste this into the console:
// AdSense units in the rendered DOM, with publisher ID, slot and fill state
const units = [...document.querySelectorAll('ins.adsbygoogle')]
console.table(units.map(u => ({ client: u.dataset.adClient, slot: u.dataset.adSlot, status: u.dataset.adStatus })))
Watch the status column. Once AdSense has processed a unit it sets data-ad-status to filled or unfilled, the same attribute Google's help pages use to hide empty slots with CSS. That separates two things a scan cannot tell apart, a page that holds AdSense units and a page that is showing you ads right now. For a second view, filter the Network panel by googlesyndication and see whether the loader was requested.
Why a single fetch misses AdSense sites
A crawler's request is one visitor, from one location, in one consent state, and it rarely scrolls. The AdSense page names three setups that keep the call out of that visit: tag managers, consent platforms and lazily loaded scripts. It also notes that delivery has been moving that way.
- Tag containers. If the loader lives inside a Google Tag Manager container, it only appears once GTM runs in a real browser. GTM is on 7,226,750 sites, and its count changed by +9.7% from October 2023 to October 2026.
- Consent first. Where consent is needed before ads run, the loader often waits for the visitor's answer. Funding Choices, Google's tool for consent and ad-block messages, is on 50,030 sites, +63% over the same three years. Its technology page points out that these messages fire by visitor geography and stored consent, so a crawl sees only one branch.
- Deferred loading. Some sites hold the script back until the reader scrolls or interacts, which a fetch never does.
The 1,282,226 figure is therefore a floor for sites with AdSense code. Part of any drop can be a measurement effect: when a publisher moves an existing setup behind a container or a consent prompt, an outside scan can log a removal that never happened in the account. Nobody can size that effect from outside, so read the trends below as trends in visible AdSense.
Six Google technologies on one page of history
The history model behind these lines is built from archived page code, one month at a time, and pinned to webatla's October 2026 census. Early years are approximate, so the chart starts at the end of 2020 and stops at October 2026, the model's last month.
In the history model, Analytics, Tag Manager, Site Kit and AMP grew from the end of 2020 to October 2026, while AdSense and AdX shrank. Each panel has its own scale.The Google technologies around measurement and page delivery grew while the two ad products shrank. For anyone counting ad code from outside, the Tag Manager panel matters most, because every new container, like every new consent prompt, is one more place where an ad call can hide from a scan.
A note on the last point of each line. The model puts AdSense at 1.27 million in October 2026, and today's count is 1,282,226. Those figures come from different methods, and webatla changed its counting model from November 2026, so the gap between them should not be read as a rise.
Turnover: how fast a tag list goes stale
Most Google AdSense usage statistics are snapshots, and snapshots hide turnover. In the past year 478,000 sites joined the AdSense list and 602,000 dropped off it. Averaged over the year, that is about 1,310 arrivals and 1,650 departures a day, a net loss of around 340 sites each day.
Departures beat arrivals in nine of the ten months from January to October 2026. March was the exception, with 44.5K in and 42.4K out.In 2026 the gap shows up almost every month. October had 26,700 arrivals against 51,900 departures, and its arrivals were the lowest of any month since 2019 in the model. October is also the newest month, and months at the edge of a series swing, so treat it as something to watch rather than a new baseline.
In each month of 2026, departures equaled 3.1% to 4.6% of the count at the start of that month. At that rate a list of sites by technology, kept for monitoring, research or a side project, loses accuracy within weeks. Store the check date with every observation and re-scan before you act on an old result.
A missing tag can mean a missing site
A site leaves the AdSense count in one of two ways. It removes the tag and stays online, or the site itself shuts down because the domain expired or stopped answering. The mix between those two has moved a lot.
Closures were 25.1% of all AdSense departures in 2019 and 47.8% from January to October 2026. The rest are sites that removed the tag and stayed online.From 2019 to 2023 closures stayed between 25.1% and 28.2% of exits. Then they climbed to 32.1% in 2024 and 40.8% in 2025. In the last 12 months, 291,000 of the 602,000 departures were sites that shut down.
This matters if you diff two scans to find technology changes. A vanished tag is ambiguous until you check HTTP status and DNS. Lately about half of the "AdSense removed" events in such a diff would be dead sites, not publishers that switched networks or went ad-free. Month by month in 2026 the split stays close to even, with one exception. October shows 43,700 removals against 8,220 closures. As the newest month that may be a measurement effect, which is one more reason to confirm before you label anything.
Server country is a hosting signal
In webatla's data, country means where the server is, by IP address. It describes hosting and says nothing about where readers or owners are. With that caveat, AdSense does not follow the general hosting map.
Japan hosts 1.7% of all websites but 8.4% of AdSense sites. South Korea and Finland lean the same way, and Germany is level at 10.6%.US servers hold 57.6% of all websites with a known country and 50.0% of AdSense sites. Germany is level. Japan, South Korea and Finland hold far more AdSense sites than their share of the web would suggest.
If you segment sites by server country, you are segmenting by hosting provider as much as by market. A country that looks oversized deserves a hosting question first.
Notes for people who maintain AdSense sites
-
Remove the code with the account. A dead snippet keeps a site in outside AdSense counts and still loads a third-party script. When an account closes, delete the loader, the units and the google.com
DIRECTline with your own publisher ID inads.txttogether. Leave the other google.com lines alone if another network or Ad Manager still sells your space. -
Keep the IDs in step. The publisher ID in the page and the one in
ads.txtshould match. Check both after a theme change or a domain move. - Reserve space for each unit. Slots that fill late and push content down are a classic source of layout shift. A fixed minimum height on the container helps keep Cumulative Layout Shift down.
-
Know what outside tools see. If the loader sits behind GTM or a consent prompt, third-party scanners and your own
curlchecks will often miss it. That is fine, as long as nobody reads their silence as proof the tag is gone.
FAQ
How many websites use Google AdSense right now?
1,282,226 websites as of November 16, 2026, or 0.39% of all 325.8 million sites in webatla's index. The number on the technology page is rebuilt daily.
Does a detected tag mean ads are showing?
Not necessarily. It means the code is on the page. The account may be closed or still under review, and units can come back unfilled. The console snippet above shows the fill state for your own visit.
Why can a site with AdSense be missing from the count?
Common reasons are a loader that waits for a tag manager, a consent choice or a scroll, and a server that returns a different page to a crawler. A single fetch can miss the tag in each of those cases, so the count is a floor.
Is AdSense usage growing or shrinking?
Shrinking. The technology page gives a 12-month change of -8.9% and a three-year change of -28% (October 2023 to October 2026). Nearly half of recent departures were sites that shut down.
Sources
Originally published by Dev.to WebDev. Aggregated on AIWithGhost for educational purposes — full credit and traffic to the original publisher.



