Dateline: Feb 28, 2011
Google officially filed for Chapter 11 protection today marking the end of an era. The once iconic giant is expected to sell off its assets and the still popular google.com domain to rival MSN. While the announcement was not unexpected given the rapid decline in Google's revenue over the past 4 quarters it still left many industry analysts wondering how it could happen.
"It all started with Quality Score" explains Nicolai Took, former VP of product development for Google. "Originally ads were served based on the advertiser's keyword and match type and ranked based on bid times the click-through-rates. That was great for us in that we served incredibly targeted ads, and we maximized revenue from those ads that were most responsive to the user's search."
"However, we weren't satisfied with the results, and neither were our users" continued Took, "We noticed that if the ads landed on cheesy AdSense Spam pages, that was a lousy user experience, so we wanted to clean that up. To do that we needed some way to arbitrarily penalize the sponsored listings tied to those pages. The notion was that we'd just punish the 'bad' landing pages and that would be it. Later, we realized that if landing pages were really slow loading, that was a bad user experience, too, so we folded that into QS as well.
"All that was good! We were rightly focused on making the ads as useful as possible to our users, thus encouraging them to use the sponsored links. Quality Score was the right idea, but it opened up Pandora's Box in a very real sense."
"QS gave us a black box to do essentially whatever we wanted to do in terms of ranking the sponsored links, and we took that freedom and ran with it. Our VP of Accounting, Ewell Cheatham, first came up with the concept of Expanded Broad Match. He said 'If a company is willing to spend $4 for clicks on keyword "A", and only $0.25 for clicks on Keyword "B", why would we ever serve Keyword "B"?' He said: "As long as the CTR of "A" is more than 1/16th of "B", we win!' Some people at the meeting objected that the keywords might not have anything to do with each other, but Cheatham persuaded the group that the market would sort things out. If it worked, we'd make more money, and if not we'd simply dial back up the relevance and go our merry way."
"The problem was, it worked great!" said Took with a grimace.
"It turned out that people didn't read the ad copy very carefully, so the fact that we were serving the 'wrong' ad didn't depress the CTR that much. The more emphasis we placed on the bid and the less we placed on the relevance the more money we made."
"Some of the savvy PPC agencies caught onto this quickly and started being very careful with match types and developed clever ways to get around this process, but most didn't. And as our addiction to pleasing Wall Street grew we also realized in late 2008 that Quality Score gave us the ability to penalize the folks who didn't play the game we wanted to play..."
"That's when our lead council, Agnes Fairchild, resigned. She saw the FTC, FCC and SEC troubles brewing and wanted no part of it."
Asked whether the government regulatory actions played a major role in Google's eventual collapse, Took responded "Yes and no. It certainly didn't help, but the real problem came from ignoring our users."
"We thought the fact that CTR didn't drop much initially meant no harm was done. We figured as long as revenue grew we must be heading in the right direction, but we lost sight of two important trends: First, we didn't realize that much of the revenue growth in 2007 through 2009 was really the last phases of rapid growth in the channel. When more and more people were poring into search and using our engine it hid a lot of tactical mistakes. Second, we thought that CTR was the best measure of satisfaction with results, but it turned out to be a trailing indicator, by the time CTR tanked we were already sunk."
"The real measure of success of course was whether people found what they were looking for after clicking on a link. Because we chose high bid ads rather than the ads that exactly matched the users search the landing pages were wrong, either too deep or too shallow, and users left the retailer's sites frustrated."
"We hated the long tail. If anyone in the office mentioned the long tail he or she had to put a buck in a penalty jar. Huge keyword lists were great for retailers, but they ate our bandwidth and depressed our revenue, or so we thought. As we saw it, they just allowed thorough agencies to get better conversion rates at lower CPCs, and we didn't see how that benefited us."
"For a while, everything was great. People didn't land on the right page, but they would come back to the original SRP and click on another link hoping to find what they wanted -- we made even more money getting multiple clicks from the same users. We were really the only game in town at this point, but that's when MSN came out of nowhere."
"We had always planned that the sponsored links would be for retailers and the organic links for information. By penalizing retailers in organic rankings in favor of "information sites" we would encourage folks to use the ads when they were looking to buy something and the organic links for information. Unfortunately somebody didn't get the message. As our sponsored links became less and less relevant people started doing their shopping on MSN's platform."
"Advertisers jumped on it and revenues flowed to MSN. By the time we realized what was happening the word on the street was out: use MSN for shopping, use Google for information. It was too late for us to reverse the trend..."
Asked for final reflections, Nicolai Took responded with a forced smile: "Remember when 'Google' was a verb?"