The Internet is full of get-rich-quick schemers, phishing scammers and just straight-up shady dudes — and they all try to get ads for their “businesses” to show up next to Google’s search results. It’s the job of hundreds of Google employees to make sure they don’t.
Virtually every minute, somebody somewhere is trying to dupe Internet users into clicking on an ad for a dubious business. Those ads show up right alongside the results of Google, Bing and other search engines, and they’re sometimes even tagged as a “sponsored ad.” Click on the wrong ad and you could encounter cons including phishing (an attempt to acquire personal information that oftentimes leads to identity theft) and cloaking (a method of deceiving a search engine so it thinks it’s a legitimate website), as well as rings selling counterfeit goods, a wide range of get-rich-quick scams and just about any other disreputable practice you can think up.
Shady ads have become such a problem for search engines that hundreds of employees now work round-the-clock to protect users before they get ripped off. Google has employed “ad cops” charged with sniffing out questionable advertisers basically since the search engine began in the late 1990s. Today, Google earns about 95% of its revenue from advertising, so making sure its ad marketplace is free of illegitimate businesses is a big part of the search giant’s own business.
It should come as no surprise that the number of scammy ads is on the rise. As search traffic grew over the years, so did search-engine advertising — as well as the number of people trying to take advantage of that advertising. One big reason why hundreds of millions of sketchy ads go live each year is because it’s so easy for anyone to advertise a business online. All a scammer needs to advertise on Google is an account and a form of payment.
Since 2008, Google has been tracking the number of bad ads that appear and are taken down. Perhaps unsurprisingly, the number has grown each year. In 2012, Google “disapproved” of a record-high 224 million ads and suspended 889,000 accounts for violating its advertising policies, which include everything from writing an ad in ALL CAPS to selling stolen goods. Compare that with 2008, when Google disapproved of 25 million ads and suspended only 18,000 accounts.
Google’s main search competitor, Bing, also reports a rise in sketchy ads. While the company wouldn’t release any numbers, David Gottlieb, Bing’s group product manager, says he believes Bing is dealing with more scammers and illegitimate businesses not necessarily because there are more shady actors but because the search engine is growing and in turn taking on more advertisers. In the third quarter last year, Bing’s traffic growth increased by 9.6%, and it now has about 16% of the search market, behind Google, which still has two-thirds of the market cornered.
Overall, the numbers suggest that year in, year out, more attempts are made to scam people on the Internet. But it may also be that after years of determining how best to sniff out schemers, the ad cops employed by search engines are better at their jobs.
Google employs a mix of machines and humans to catch bad ads. Its first line of defense is software that searches for ads with specific attributes Google finds unsavory. For example, ads related to online gambling automatically get snagged because the practice is illegal in the U.S. Because the automated system isn’t perfect, Google also relies on humans to determine some ads’ legitimacy. And even then, about 2% to 3% of accounts that are shut down are overturned after being mistakenly suspended.
A few years ago, it appeared that the automated system was defective. What appeared to be perfectly legitimate sites advertising used cars in China were being flagged as suspicious by Google’s system.
“The engineers thought that the model that they were building had some fundamental flaw, and it was going to catch a lot of really good advertisers and frustrate them by either shutting them down automatically or delaying their ads,” says David Baker, Google’s director of engineering and advertising, who is essentially acting commissioner of the Google police department.
But Google’s model was smarter than they realized — smarter than even the engineers themselves. As it turned out, the people behind the questionable ads were taking photographs of cars parked on the street and uploading them to sites to be purchased. Then, once a buyer emerged, they stole the cars.
“It was this rather nasty scam,” says Baker, “and this new model caught it without the engineer having any knowledge of this space. It was purely based on looking at attributes that related this scam in some way to past scams.”
Baker says that while he believes the problem of bad ads may be growing, he also knows that Google’s detection systems are getting more sophisticated, and therefore they’re catching more ads. One way Google’s ad cops have figured out they’re improving is through a weekly evaluation of a sample of the search engine’s top advertised sites to determine if any illegitimate ads have seeped through. Baker says Google reduced the percentage of bad ads found in those weekly evaluations by 50% in 2011 and by another 50% last year.
A few specific types of sketchy ads decreased in 2012, including advertisements for counterfeit goods as well as cloaking scams. Overall, however, the numbers keep going up. Still, Google’s Baker is hopeful that bad ads are close to peaking.
“I have this aspiration that we will get so good that there is going to be an inflection point, where the bad actors realize it’s not worth their time and energy,” he says. “It’s possible that the 2012 numbers are an indication that we’re approaching that point. But this is going to be a constant battle for as long as I want to be employed at Google. In a somewhat sadistic way, I have job security.”