400 million Google Earth users. Really?

In his Geoweb 2008 keynote speech, Michael Jones mentioned that among the 1 billion online, there are “400 million Google Earth users”, and that this “constituency” is bigger than the number of Americans.

Time for a reality check. That number for “Google Earth users” can’t be unique users. Downloads, sure, but not users. Just think about it: Let’s even allow 1.4 billion online people in 2008; less than a quarter of them have access to broadband, according to Gartner. And among broadband users, businesses are over-represented — businesses where a program like Google Earth is less likely to be downloaded. To get to the total of 400 million users, every computer on the planet connected to broadband would have to have Google Earth installed, plus a whole lot of people sucking Earth through a 56kbps straw.

Whence the discrepancy? I’ve done my part, downloading Google Earth at least a few dozen times between different versions and successive machines from all sorts of different IP addresses. And likely so have you. This is not to take anything away from the fantastic uptake of Google Earth, not least in the zeitgeist of the world’s technology elites; but just as websites no longer advertise the number of “hits” on their site, isn’t it time for a more conservative number on Google Earth users? Surely Google Earth can concoct a unique hash number for each install so that Google can acquire data for unique visitors per month, just like top websites do? Why not then release that if advertising the popularity of Google Earth is important?

14 thoughts on “400 million Google Earth users. Really?”

  1. Because 400 million users sounds better than “400 million downloads and only +/- 2 million active users”

    It is all marketing, you should know that by now Stefan.. the $$ bottom line is what is important.

  2. Chad, what are your claims of fact based on? Also, what would be negative two million users? I don’t know you but you sound like a petulant child. Are you bitter about something?

  3. 400 million is a silly number to put forward. The audience for that comment would be clueless journalists (is that redundant?) so they can regurgitate it. I doubt that they even track unique users or even IP addresses since it is a free service. If that is the case, the only number they have is downloads, which probably includes all downloads for all versions since the Keyhole days to get that impressive number. However there is no profit motive for the Google Earth team, they are a loss leader with unlimited funding. Nice to be the king.

  4. @Doug: Boy.. You don’t know what +/- means when in front of a number? It means could be more, could be less. Bitter? Naa, I am just not a “Google is master of all” fanboy. Google has spewed the “billions of users” lines before and counted all downloads as users.

    @Matt: Not as much king, but “Nice to have the best PR department around” ;)

  5. I’ve questioned Google about this number before. I was told this number does NOT represent downloads – it represents the unique activations of the client. I have 5 or so computers at my house, each one has an activated Google Earth client. Most have GE Free, and two have GE Pro also. I’ve downloaded dozens of versions of GE during the last 3 years (with different beta versions). If my assumptions are right, then only 7 activations of GE on the machines in my house would count towards the number. There are four people in my house who use these computers and GE, so I don’t think the number is all that inflated.

  6. Yes, I can see how that number might mean unique activations — and that number might be a little lower than the number of downloads, in that some completed downloads will not be installed and activated.

    But activations is not users. And I suspect (but can’t know) that every time a new version is downloaded and used once, it is counted. That’s a number that is different from the number of people who have used Google Earth, though this latter definition is implied by Google when that number is compared to the number of people online or the number of Americans.

  7. The number I cited in Vancouver is my attempt to share the number of times that the following happens: a newly-installed Google Earth (“first execution”), running on a machine with no previous install of Google Earth (“fresh machine”), connects for the first time to a Google Earth server, and does so via a new IP address (“fresh address.”)


    if (firstSuccessfulConnectionSinceInstall) then;

    { ;

    if noOldGEFlagsInRegistry then SayNEW ;

    else if oldGEFlagIsSameVersion then SayREINSTALL;

    else if oldGEFlagIsOlderVersion then SayUPGRADE;

    else if oldGEFlagIsNewerVersion then SayDOWNGRADE;

    else SayCONFUSED;



    if SaidNew then;

    { if newIPAddress then ++activationCount else ++reinstallCount };

    else if SaidUPGRADE then ++upgradeCount;

    else if SaidDOWNGRADE then ++downgradeCount;

    else if SaidREINSTALL then ++reinstallCount;

    else ++confusedCount

    This is my best understanding of exactly what happens (though it is implemented differently since the server side is log based rather than synchronous.) We generally do not quote activity levels per unit time (this week, hours/day, etc.) for Google products as they might be misconstrued as revenue predictors. We do not quote downloads for many reasons, not the least of which being that some people download but do not install and many get their installers from download.com, customer IT department distribution servers, and other sites that we have no way to count.

    For example, download.com claims 3,397,523 downloads of GE version 4.3 — these are only part of the 400,000,000 number in the instance where users did successfully download, successfully install, successfully activate (first run of the program that successfully connects), and the machine history and IP address obey the rules above (no prior GE evidence, new IP address, etc.)

    Chad is right that some people and some companies lie. There were times, as a child, that I lied; that was long ago and not any part of my adult life. Nor is it typical of any public company. My belief is that calling someone respected a liar without evidence, proof, or even the slightest understanding is in poor taste, to say the least. The numbers I cite are the absolute engineering truth as best I know it, and are understated considerably to avoid the risk of overstatement. (Yes, there may be more activations than users as some may have multiple machines, but then again our actual counts for activations is notably larger than the count we quote–we are intentionally conservative rather than intentionally disingenuous as Chad asserts without evidence.) As stated above, none of these installs includes upgrades, downgrades, or reinstalls.

    Perhaps I was unclear (in the whole 60 minute talk not the 9 minute excerpt) but I know no other way to convey the amazing pace of uptake globally for GeoWeb / consumer GIS / GeoBrowsing than to cite our carefully-measured activation counts. If you were there (Chad?) then you may remember that I constantly included ESRI, Yahoo!, Mapquest, Microsoft, and others in the description of the amazing progress to worldwide embrace of geobrowsing. My numbers were used to share the power of ten involved, not to invoke any echoes of bolshevik/menshevik posturing.

  8. Thanks Michael, that method is indeed quite a conservative way of counting activations. But that still leaves the question as to how that number is significantly higher than the number of global consumer broadband connections (323 million in 2007 according to Gartner) It could be of course that Gartner is wrong. Perhaps business activations lead to double counting. It could be that each internet connection feeds multiple computers. It could be that over a three year period, most anyone who has downloaded GE will do so again on a newly purchased computer via an IP address that is randomly assigned by the ISP. This last factor alone might overestimate the number of actual users by up to a factor of 2.

    Perhaps unique activations is the best proxy there is for actual users, but I still don’t think it’s a number that can be used to compare against the number of Americans or the number of web users.

  9. Better watch out Stefan. You may be accused of calling Michael a liar with that questioning. :)


  10. I agree that it is difficult to count noses by looking at IP traffic. Our technique is the best we’ve devised so far. (Remember that we estimate the “double count” rate (f) and cite nice round figures well below the (1/(1+f)) compensation multiplier.)

    One possible idea, which you’re discounting completely yet is certainly true at least partially, is that some people don’t have broadband connections and use Google Earth (et. al.) anyway. I was for more than a year one of those people (using Keyhole from my Sony VAIO via modem from the hotels of the world.) In fact, I did many of my 3000+ Google Earth Community posts in just that way. The secret was to uncheck “Imagery” and enable only roads. Then, driving around the world to find just the spot desired and only then re-enabling the imagery. While you may consider this unacceptable it was in fact just “highly unpleasant.” Up until recently more than half of Google’s network traffic was from non-broadband lines, though I doubt this would be the case with Google Earth.

    Remember that it is a big world out there with *wildly* divergent infrastructure, expectations, and experience-driven tolerances.

    One of the things we looked at for comparables was to watch FireFox, Skype, and other big societal wave product counts. (downloads of the Vista beta, downloads of XP service packs, new versions of Photoshop, etc.) The problem there is that they tend to count downloads which says something, but not as much as would be desired.

    P.S. KoS — presumably you’re joking and can see the difference between civilized curiosity, question, and debate on the one hand and the verbal equivalent of throwing tomatoes at the Pope. I respect Stefan irrespective of our position on any topic. This is helpful to each participant and entirely different in character than the rude post.

  11. Michael, you presume correctly…rather I would call it a smart ass comment!

    I will say, in all seriouness, I didn’t find Chad’s post rude or the equivalent of throwing tomaotes at the Pope. I hope you didn’t try and place yourself in the same category as the Pope. :)

    Btw…not sure if you have seen a picture of NRCS’s current Chief? Y’all sure look like brothers. :) At least IMHO.



  12. One more thing: From personal experience, Michael is a remarkably generous person whose integrity is beyond reproach (and not just because he’s so smart). I’m obviously not questioning his motives here, just the methodology.

    Meanwhile, I’ve long given up moderating for civility on this blog — it has to get pretty egregious before I censor:-p

  13. hi,

    i came up with an idea google earth, and i need a person i can talk to from inside, or email so please give me a contact adress.


    s. picard

Comments are closed.