Population of the US: 349 M, of which 250-300 M use Google services, multiplied by 1605 USD per user = from 401 B USD to 481 B USD, but in 2025 Alphabet did 403 B in total, from every service, in the whole world.
Correct. If 250 M people use it in the US out of a population of 349 M, Google would make 401 B USD out of them, vs 403 B USD in worldwide revenues. These numbers do not look right to me.
median is the sample in the middle of the distribution (when it is treated as a sequence of samples ordered by their value), e.g. if you have sort(seq(dist))=[100$, 5$, 5$, 3$, 1$] the median is 5$
average is sum(dist)/size(dist), so avg * size(dist) => sum(dist)
in the example above that would be median 5, avg. 22.8, total 114, size 5
if you where to multiple the median by size you would have 25$ for the total value, which is very much very wrong
There are some people who don’t use Google. I use Duck Duck Go for search. Additionally, with the rise of LLMs I have been using search much less in general.
and ad blockers also can kill ad revenue in various ways (like by not displaying them, or even if displayed by causing them to not be counted due to not realizing they where displayed or finding irregularities due to which they are classified as bot views).
similar anti-fingerprinting tech can kill ad revenue as it makes users non distinguishable from bots (but likely doesn't matter here)
proton did 54,000 samples of US users and made an average of what advertisers are willing to pay to target, not what they actually did across the whole population
and plus this isn’t to inform you, it’s to sell you on another proton honeypot
Caveat I'm no expert on Google ads. Never bought one, never plan to, never advertised anything at all on any service. But since I'm capable of doing a basic web search, I found:
This is the process for determining which ads get run. The bid is only one of many factors, so as their support document indicates, the price you pay is often quite lower than the bid, which reflects a ceiling rather than a real sale price.
This is their guidance on demographic targeting. Note there is no category allowing you intentionally target children. This doesn't mean advertisers can't figure out some way to do it anyway, but it means Proton can only sample from adults. Presumably, some probably very large number of the people who "use Google services" in your estimate are children, which childstats.gov indicates represent about 22% of all Americans. That makes it more like 195-235M adult users of Google services.
As indicated here, you don't pay to place an ad. You pay for clicks, so regardless of what you bid and who you target, Google isn't getting revenue for the number of placements you bid on, which is what Proton is sampling here. Presumably of the 250 x 0.78 to 300 x 0.78 million adult users advertisers are placing those $1605 average bids on, quite a bit fewer than 100% actually click on at least one ad.
1st: no need to personalise, I agree. Sorry about that.
2nd: where did you get the 200 bn USD figure for the US ?
3rd: if we multiply 250 M people (likely Google users in the US out of a population of 349 M) and multiply it by the 1,605 USD that Google is said to make out of an average user in the linked blog post by Proton, we get 401 bn USD, not 200 bn.
This could be explained by the 250-300 M you refer to not matching the same distribution due to
1. this seems to be google ad network specific, not google services per-see
2. the analysis seem to only include users which do in general generate ad revenue, e.g. all AD Block everywhere users are not included in the distribution
3. given the lower bound I assume ad views which have no clear attributable user, and/or users with a very low and irregular amount of views, are not included (e.g. some mostly "offline" people, people mostly using an ad-block but sometimes somewhere still seeing an add, also it's G-Ads, so anyone using only FB, TickTock etc. would not show up I think)
The population of the US is not necessarily the same as the population of Google's users with personalized ad profiles in the US, even though it sometimes feels that way :). E.g. you go from 349,000,000 to 267,000,000 just by removing the under 18s.
I think you mistook the example of removing a single group as a tally of what number to use instead. The final number should be significantly less than the 250,000,000 who aren't children.
FTA: "One caveat: this analysis estimates advertiser demand for access to a given profile. It does not reflect the exact revenue Google receives from any individual user. What the model reveals is the ceiling, the maximum price the market places on your attention."
From this I believe that your problem can be solved. In accounting terms, the 1605USD isn't a flow (e.g., revenue) but rather a stock (e.g., receivable). They've estimated value about how much a profile is worth, which you shouldn't use to draw those conclusions about revenue.
Add Meta and the rest of the Attention Economy and for a family of 4 they extract 10K a year. The rest of the world its like $700-1K. The US Attention pool gets overfished because thats where most of the world cash sits. Over optimized Cream Skimming.
This is a good example of why averages (by themself) can be very misleading:
- avg. $1_605
- but mean is $760, i.e. half the users generate $760 or less
I also wouldn't be surprised if the sampling distribution has two maxima even if smoothed (on around the mean and another at the lower end). Would be nice to have that plotted out properly.
It's also why Google doesn't want to just charge for services. I wouldn't pay more than about $60 a year or so, because I am very much on the low end for value I bring to Google, but they would never accept that because it would be a huge haircut on the high value users, and if they set it high enough to be profitable for high value users, I literally can't afford it.
The ad ecosystem allows them the equivalent of ideal price discrimination. You bring in exactly how much revenue they can squeeze out of the economy for access to your eyeballs.
This is why I'm not concerned about robotic labor surpluses. A human can generate tons of invaluable data. Right now they aren't capturing that much value, but eventually it will be worth it. That's why the EU is moving fast on data rights. The US will probably wait until ~$100K is being extracted from every man woman and child. In addition to data centers I'd like to see last mile fiber to every home. Whichever company does that will be rolling in money in 2040.
It's interesting/non-intuitive to me that parents would be worth less than non-parents. I wonder: is that because they actually tend to spend less, or is it because more of the clicks are accidental ones, from when the kids get handed the phone?
But I guess that would apply more to display ads than to search ones, so I'm not sure. Confusing.
My best guess is that for certain goods, people make purchasing decisions around major life events like getting married or becoming parents. If they have already crossed those thresholds those purchase patterns may be harder to unseat and replace, they may already be solving those new needs and have habits around existing brands.
My second guess is that DINKs have more disposable income.
Disposable income is Gross Income minus Taxes. A 2 parent family with 2 kids making the same as a DINK couple would have a lower tax burden and more disposable income, but the DINK couple almost certainly has more discretionary income, which money left after paying for necessities like housing, food, and medical care.
I have never seen a single ad online, except may be for one moment the ad blocker malfunctioned. I am certainly an around 0$ ad value user for Google. So some must be much more than 1605.
I forget how bad the internet is without an adblocker until I use someone else's device at which point they get inducted into what an adblocker is via "You want your web experience to be more pleasant?"
It's gotten to a point where if I can't run an adblocker like Ublock Origin I won't use the browser/device (I still need to look at DNS level blocking but everything I use runs FF so not super pressing).
Chrome (and derivatives) are only installed for testing stuff, FF for all it's sometimes questionable UX and AI crap (which can at least now be disabled with a single toggle) has been my default for a long time and is still for that reason.
In fairness though, I don't use many google products any more either (YT (with an adblock and sponsorblock) mostly).
If search was frictionless then it would be far cheaper. People wonder why google search has been enshittified vs 15 years ago. If people could find exactly what they needed they would need far less ads.
Population of the US: 349 M, of which 250-300 M use Google services, multiplied by 1605 USD per user = from 401 B USD to 481 B USD, but in 2025 Alphabet did 403 B in total, from every service, in the whole world.
median is the sample in the middle of the distribution (when it is treated as a sequence of samples ordered by their value), e.g. if you have sort(seq(dist))=[100$, 5$, 5$, 3$, 1$] the median is 5$
average is sum(dist)/size(dist), so avg * size(dist) => sum(dist)
in the example above that would be median 5, avg. 22.8, total 114, size 5
if you where to multiple the median by size you would have 25$ for the total value, which is very much very wrong
similar anti-fingerprinting tech can kill ad revenue as it makes users non distinguishable from bots (but likely doesn't matter here)
proton did 54,000 samples of US users and made an average of what advertisers are willing to pay to target, not what they actually did across the whole population
and plus this isn’t to inform you, it’s to sell you on another proton honeypot
https://support.google.com/google-ads/answer/6366577?hl=en
This is the process for determining which ads get run. The bid is only one of many factors, so as their support document indicates, the price you pay is often quite lower than the bid, which reflects a ceiling rather than a real sale price.
https://support.google.com/google-ads/answer/2580383?sjid=17...
This is their guidance on demographic targeting. Note there is no category allowing you intentionally target children. This doesn't mean advertisers can't figure out some way to do it anyway, but it means Proton can only sample from adults. Presumably, some probably very large number of the people who "use Google services" in your estimate are children, which childstats.gov indicates represent about 22% of all Americans. That makes it more like 195-235M adult users of Google services.
https://support.google.com/google-ads/answer/2464960
As indicated here, you don't pay to place an ad. You pay for clicks, so regardless of what you bid and who you target, Google isn't getting revenue for the number of placements you bid on, which is what Proton is sampling here. Presumably of the 250 x 0.78 to 300 x 0.78 million adult users advertisers are placing those $1605 average bids on, quite a bit fewer than 100% actually click on at least one ad.
remember the average 'world' user is about 100x to 500x less valuable than a US user.
Do you wear a red baseball cap?
google made about 400bn last year. 200bn of that was from the US alone.
now you can estimate how much the UK brought in.
& for PPC rates - you can see conversion rates as well.
I run an alternative to Google analytics for a niche market so yeah.
2nd: where did you get the 200 bn USD figure for the US ?
3rd: if we multiply 250 M people (likely Google users in the US out of a population of 349 M) and multiply it by the 1,605 USD that Google is said to make out of an average user in the linked blog post by Proton, we get 401 bn USD, not 200 bn.
1. this seems to be google ad network specific, not google services per-see
2. the analysis seem to only include users which do in general generate ad revenue, e.g. all AD Block everywhere users are not included in the distribution
3. given the lower bound I assume ad views which have no clear attributable user, and/or users with a very low and irregular amount of views, are not included (e.g. some mostly "offline" people, people mostly using an ad-block but sometimes somewhere still seeing an add, also it's G-Ads, so anyone using only FB, TickTock etc. would not show up I think)
From this I believe that your problem can be solved. In accounting terms, the 1605USD isn't a flow (e.g., revenue) but rather a stock (e.g., receivable). They've estimated value about how much a profile is worth, which you shouldn't use to draw those conclusions about revenue.
- avg. $1_605
- but mean is $760, i.e. half the users generate $760 or less
I also wouldn't be surprised if the sampling distribution has two maxima even if smoothed (on around the mean and another at the lower end). Would be nice to have that plotted out properly.
Median*. Mean and median are both measures of averages, though colloquially average is taken to exclusively mean the mean.
The ad ecosystem allows them the equivalent of ideal price discrimination. You bring in exactly how much revenue they can squeeze out of the economy for access to your eyeballs.
But I guess that would apply more to display ads than to search ones, so I'm not sure. Confusing.
My second guess is that DINKs have more disposable income.
It's gotten to a point where if I can't run an adblocker like Ublock Origin I won't use the browser/device (I still need to look at DNS level blocking but everything I use runs FF so not super pressing).
Chrome (and derivatives) are only installed for testing stuff, FF for all it's sometimes questionable UX and AI crap (which can at least now be disabled with a single toggle) has been my default for a long time and is still for that reason.
In fairness though, I don't use many google products any more either (YT (with an adblock and sponsorblock) mostly).