Help with a Numbers Question

I’m looking for help and opinions around a persistent question having to do with Google advertisers. There are really two questions:

1. What percentage of Google advertisers are SMBs? My sense has always been that it’s something like 70% (or so) who technically qualify as small businesses. The general estimate is that Google may have something like 1.5 million on a global basis but that’s unconfirmed.

2. What percentage of the overall ad spend at Google comes from larger/national vs. SMB advertisers? In other words, what part of Google’s $21.8 billion in 2008 (99% advertising revenues) comes from the different categories (if one had to crudely divide them)? Would it, for example, be 60% of the ad spend coming from national/larger entities? That’s just a hypothetical number.

What do you all think and why?

13 Responses to “Help with a Numbers Question”

  1. Tim Cohn Says:

    You beat me to this!

    Reading Danny’s Saturday post about their revenue per hour got me thinking about the numbers.

    This doesn’t directly address your question but I think the answer lies in the numbers.

    $21.8 Billion / 1.5 million advertisers = $14,333.33 average annual revenue per advertiser.

    $14,333.33 average annual spend / 365 days = $39.27 spent per day per advertiser or the average Google advertiser spends $1.64 per hour 24 hours a day 7 days a week.

    8 Billion searches monthly / 30 = 266,666,667 million searches a day…

    Adding a few more variables should produce a relatively accurate answer.

  2. BTS Says:


    One thing in your question that would need to be a bit clearer is the type of small business. If you are talking about pure web online sites versus what we all think of as a local small business you will get alot different answers.

    On the math, because Google publishes the number of advertisers and the revenue per, etc you can estimate this IF you apply the right distribution. The key decision is if you use a normal distribution or a long tail distribution, each will give you a dramatically different answer on the number of small advertisers.

    From the number of small advertisers, you will have to do another estimate to figure out the number of them who are small local businesses.

    I believe that less tha 500K small local business where buying local internet search advertising a few years ago when we first did an analysis like this. I would assume this number has change alot since 2005.

  3. Greg Sterling Says:

    yes, type of SMB is relevant to a more nuanced analysis. There are lots of ways to segment, but what I’m asking about here is just pure SMB definition (headcount <100) vs. larger entities.

  4. BTS Says:

    You can do that by applying a long tail distribution to the data Google has in its annual. If you use a normal distribution, you will a much lower number than reality.

  5. Greg Sterling Says:

    You’ve got a majority of advertisers who qualify as SMBs; that’s a certainty. But does a majority of the ad revenue come from a much smaller proportion of large advertisers?

  6. earlpearl Says:

    Nice questions, Greg: I’d like to know that too. One way is to scrutinize Google’s financials, of course. They may well obfuscate that information, not report it in that way, segment things differently so it is hard to decipher, etc.

    I wonder how many small businesses advertise in Google through a channel, PPC reseller, etc. I wonder what percentage of small businesses have a website, and of those what percentage advertise, primarily through PPC. Who are the largest Google resellers?

    Wouldn’t we all like to learn, especially MSN and Yahoo. I suspect Google will work hard to keep that information secret.

  7. Richard Foxworthy Says:

    The best way to answer your first question is through primary reasearch – ie by actually examining the ads Google serves in response to popular SMB advertising categories keywords, and determining whether or not those advertisers are in fact SMBs.

    Exane BNP Paribas performed such an exercise as part of a January 2008 equity research paper on PagesJaunes, the French Yellow Pages publisher. For 400 keywords examined, less that 10% of the SMBs advertising in PagesJaunes were also advertising in Google.

    Of course this doesn’t necessarily imply that advertisers are not SMBs (they could have been SMBs that just didn’t advertise in PJ), but it did make me curious enough to conduct a quick study along similar lines in my own market in Australia.

    I took the 20 most searched-for terms in, conducted searches for equivalent keywords, and examined the sponsored listings returned for each search.

    Consistent with the French data, there was very small overlap in advertisers.

    What really surprised me was that over 80% of the Google advertisers on these top SMB keywords were not SMBs – they were actually directories, quote aggregators, verticals, portals and so forth.

    This is only an preliminary result from a small sample, but I think it is instructive that the result was so pronounced. Google has a long way to go in penetrating the SMB market.

  8. Greg Sterling Says:


    Very interesting. If you also look in organic listings — save the maps “one box” — you see the same thing: directories and vertical aggregators. Questions/issues:

    Is this a form of “affiliate spam” and undesirable, or do we simply recognize this as a necessary result of SMBs who don’t have the time or sophistication to do this themselves?

  9. Rob Shields Says:

    With a little time, one could probably create a back of the envelop estimate if one had the following data:

    1. Define SMB. e.g. headcount of <100
    2. Define the market, i.e. are we talking about worldwide advertisers/revenue or just the advertisers/revenue in the US?
    3. Take a random sampling of clicks on Google Ads. This number probably wouldn’t have to be huge as long as it was representative. Maybe 100-200 clicks to get a rough estimate.
    4. Trace each click back to the advertiser and spend some time approximating the size of the company advertising. This would approximate advertiser share X advertiser size.
    5. Be sure to weigh each click by the cost/click in order to estimate revenue share X advertiser size.

    Not sure how available this data is, particularly #3 and #5. Seems like it could be done, though.

    An even rougher approximation would be to poll folks on this blog and use the wisdom of crowds. James Surowiecki would suggest that this approach alone would be fairly accurate.

  10. Greg Sterling Says:

    Good advice Rob . . . one just needs to find the time to do this.

  11. earlpearl Says:

    Greg: I found Richard Foxworthy’s comments interesting…but something bothered me. You can’t use yellow pages topics solely as an indication of where and how primarily local smb’s advertise. Simply excluding the web there are a wide variety of local media in which “local” smb’s might advertise beyond YP. These include print, radio, and tv, among others. There is a substantial amount of non web local advertising that is not YP oriented–a very substantial amount. We have seen from various commentary on YP how certain topics continue to do relatively well in YP while others that had some success no longer do well in that environment, either IYP or print versions.

    I think the idea of the study is appropriate. I would suggest that a “better” more comprehensive way to do a comparison is to find the topics that generate most “local” advertising in total and then compare that to how many of those businesses buy PPC. I believe that would be more revealing.

  12. Greg Sterling Says:

    Agree. Lots of SMB spending is outside YP.

  13. Chris Says:

    Nice one, thanks a lot great piece.

Comments are closed.

%d bloggers like this: