Google advertiser legitimacy statistics - July update

[sitetruth]

We collect data on Google AdWords advertisers and advertiser legitimacy. Our AdRater browser plug-in provides advertiser ratings to users, and as users surf the web, we collect information about Google's advertisers.

Our latest statistics run gives us:

Rating A: 1732 11.5% (Site ownership verified)
Rating Q: 5861 39.0% (Site owner identified, but not verified)
Rating U: 1816 12.1% (Appears to be non-commercial site)
Rating X: 5602 37.3% (Commercial, no ownership identified)
Domains reported: 15011

These are unique second-level domains, appearing in AdWords ad links, from "aaacreditguide.com" to "zyrel.com". They're not URLs or subdomains.

We ran this back in March, with only 5000 domains seen, and the statistics were roughly similar; more than a third of advertisers don't have a business address we can find. We call those "bottom-feeders", since in some jurisdictions this
omission is a criminal offense.

[abbottsys]

....We ran this back in March, with only 5000 domains seen, and the statistics were roughly similar; more than a third of advertisers don't have a business address we can find. We call those "bottom-feeders", since in some jurisdictions this
omission is a criminal offense.
Only 5,000. My gut feeling is that there are about 300,000 AdWords advertisers (correct me if I'm wrong.) With such a small sample can you evan claim statistical significance? Is your survey biased?

[sitetruth]

It's a "long tail" thing. By now, with 15,000 domains seen, we've probably seen all the major advertisers. There's an English-language bias, in that the AdRater plug-in is only available in English, and we only have good address recognition for US, CA, UK, and AU.

Since the stats didn't change much as the sample size increased from 5000 to 15000, they're reasonably sound. New advertised domains continue to be reported; we see several hundred new ones a day, but the ratios aren't changing much as more data comes in.

We're interested in this as a metric of the "slime factor" in online advertising.

[Discovery]

I believe this conversation has played out a few times before.

I'm all for weeding out the junk from adwords advertisers, however using a plug-in that uses one persons idea of what is junk over another's is just not going to fly. No address.. so what, no whois info, so what. And bottom feeders???? You've got it back asswards.. the companies who scrape and use whois, DNS and website contact info to generate and sell that info on CD's and websites and so on.. Dude, those are the bottom feeders. Most of us are simply trying to hide from them.

The better approach to weeding out junk sites, junk content partners and junk clicks is simply to create a way to peer review the network of publishers.

If I were to take my monthly placement performance report and appended it by tagging each site listed with a basic review I would have my interpretation of what sites are "bottom feeders" and which are not. Combine this list with 100 others advertising in the same vertical and you can garner some consensus as to which sites are producing junk clicks for our vertical. These sites may work well for mortgage, but terrible for credit counseling or some are plain old garbitrage, domain tasters or worse. The aggregated data based on performance will clearly define the trends. This data can be applied in many ways, the least of which would be to generate a monthly offender report for Google and see if they take action and publically flog the site owners.

At minimum all contributors to the master list can have access to the aggregated data and comprehensive reports and generate a very large, precise and timely site exclusion list according to their vertical.

Again I like all attempts to solve this issue, but I feel you are trying to automate a review with a singular view when the interpretation is often subjective.


Discovery

[sitetruth]

There are accepted criteria for evaluating business legitimacy. Step 1 is finding the real business name and address. Then one can check against various business databases, from Yellow Pages lists to Dun and Bradstreet.

"Businesses" that can't get to Step 1 aren't real businesses, and in some jurisdictions, they're criminal enterprises.

[Discovery]

Sitetruth,
Sometimes I'm not sure if your just jerking our legs to get a laugh or what. Well it worked this time. They're "criminal enterprises" that was very funny.

Leaping to the conclusion that a company that doesnt post or publish their physical address is a criminal enterprise is plainly silly. Sure you may find some vague law that says so in some jurisdiction, like its against grazing your cow in Boston Common, but criminal enterprise?

Next the Governments going to go after all the companies with a .org tld that are not non-profit or not for profit companies. surely this is a sign that they are all criminals!

What do Apple Computer, Hershey Chocolate, Mary Kay Cosmetics, and the Ford Motor Company have in common? They all started out as home based businesses. Over half of all U.S. businesses are based out of an owner's home.

So out of these how many do you think are web based businesses? How many do you think operate out of a PO box? Or hide their physical address so some nut case doesn’t come knocking on their door? We're not talking a few hundred, we're talking about millions of small sites.

Anyhow, again, you can try to use laws and public directories to define what you believe is a legit company, but I think you will only get about a 50% accuracy rate at best.

[abbottsys]

There are accepted criteria for evaluating business legitimacy. Step 1 is finding the real business name and address. Then one can check against various business databases, from Yellow Pages lists to Dun and Bradstreet.

"Businesses" that can't get to Step 1 aren't real businesses, and in some jurisdictions, they're criminal enterprises.
Yes. There's this website I've been using for ages, and now I can't find their business address on the site. Are they a criminal enterprise? Here's the URL..

http://www.google.com

[sitetruth]

Yes. There's this website I've been using for ages, and now I can't find their business address on the site. Are they a criminal enterprise? Here's the URL..

http://www.google.com

No problem there. When we run Google through our mill, we get this: http://www.sitetruth.com/cgi-bin/ratingdetails.cgi?url=www.google.com&ratingid=192134&details=true.
The main page links to the "About" page, which links to the "Contact Us" page, and there's a street address there, which we find.

Google also has an SSL cert, a real one with an address, not one of those meaningless "domain control only validated" certs.

They're in Open Directory, too, which we note.

Most big companies are quite good about putting a valid name and address on their web site. Problems come from sites where Javascript, Flash, or Silverlight has to be executed to get to the contact info. But the big-name sites usually have an SSL cert, so there's a backup source of reliable info. We have some other sources for cross-checking, too.

No, the big players aren't anonymous. Neither are most of the legitimate small players. Incidentallly, both Apple and HP had street addresses from the beginning, even though they started in garages.

Once you have a business name and address, there's a whole infrastructure in place for checking out a business. Without an address, it's difficult to collect from a business that hasn't performed on its obligations. So looking for a business address is the first step in filtering adspam.

[jimbeetle]

Without an address, it's difficult to collect from a business that hasn't performed on its obligations. So looking for a business address is the first step in filtering adspam.
But what that initial check does is ignore that a whole other world exists on the 'Net.

I run mostly informational sites. No transactions. No obligations. Nothing to collect. I do however occasionally run AdWords campaigns. Sometimes for branding and awareness, sometimes to drive a bit of traffic in slow seasons, sometimes when I add new sections to sites.

As I don't want people knocking on my apartment door at night I don't list an address -- it's unnecessary. By your initial ceck my AdWords campaigns are considered spammy.

Who are you to say that my business model is not legitimate?

If this is somehow implemented you're going to be walking a very, very fine line. If sitetruth "technology" reaches the point where it warns people not to visit legitimate sites then your business model had better include a hefty budget for lawyers.

[Discovery]

I just think your coming at the issue from the wrong angle.
Machines can only get you so far in determining whats good and bad, its people that go the final mile.

Computers dont determine what music is is good or bad, people do by rating songs, making purchases, downloading samples and so on. Computers are great at organizing and presenting this data, but they are not the setting the rules as to why a song is good or bad.

So the same applies to website and ad ratings. You need to allow for consumer feedback/input to truly make your ratings more organic and less mechanical.

I think what many are questioning here are the assumptions you are making and that they just don't fit the reality of the internet business owner.


Discovery