I've written before that while most people realize how dominant Google is in search, they don't realize how dominant Google is in web analytics. and imho, even fewer people understand why Google is placing such a huge bet on analytics (& checkout, & a/b testing), which i believe is required in order for them to pave the way for future growth in CPA-based advertising.
yep, they're still The Juggernaut, bee-atch!
side note: many folks may be aware that one of the more notable transitions that occurred during the end of the last dot com bubble / recession of 2000-2004 was the dropoff in CPM-based advertising (Yahoo & AOL) and the simultaneous birth / growth of CPC-based advertising (Google & Overture->Yahoo). this isn't really that surprising, since during tough times there will always be a flight to quality / lower risk, and qualified clicks (CPC) are certainly lower-risk / higher-quality than impressions (CPM).
looking forward, while i don't quite agree with folks who think another internet recession is looming, i would observe that any such pullback will likely create a new & further flight to quality, which i believe could spur further growth in CPA-based advertising (which typically results in a specific conversion and/or transaction). in other words, CPC (clicks) is better than CPM (impressions), and CPA (actions or transactions) is better than CPC.
with impression-based advertising, i hope some % of users who see an ad click on it, go to a website, and convert to an action or sale. with click-based advertising, i know how many people click thru to the site, but i still don't know how many convert to an action or sale. but with CPA, i only pay for the specific conversions, actions, or transactions that actually happen. wotta deal!
This inexorable, incremental advertising [r]evolution is illustrated below:
now if you believe this analysis, you might suggest companies that control point of sale / transaction & conversion data are well-positioned to take advantage of this potential next wave of advertising... and you would be half-right, in that companies like eBay & Amazon (& actually also Yahoo Stores, Craigslist, & several other e-commerce sites) know exactly when sales occur, and could possibly setup a way to allocate & charge for advertising based on the potential for leads that actually convert.
however, most e-commerce companies aren't adept at offering "turn-on-the-faucet" advertising solutions that drive more users to their listings or eBay storefronts at the drop of the hat. furthermore, i don't think they're well-prepared to gather data via web analytics to see which merchants they should be sending the best traffic to (just give me the GOOD leads, murray!). this is kind of curious, since theoretically they have all the data they need to make it happen. but so far, Google has been eating their lunch WITHOUT having hardly any point-of-sale transaction data (albeit with the modest success of Google Checkout they now have a little, and with Google Analytics they are gathering quite a bit).
still, all of this conjecture is moot without a strong ability to understand user behavior on websites prior to actions or transactions occurring. and to date, Google Analytics -- like most analytics tools -- has not been very decision-oriented. that is, the output of most web analytics tools do not drive towards making a decision... they simply sit there & look pretty. great if you're sending a report to your boss, but kinda crappy if you're trying to figure out what features & marketing actually drive better performance.
Google Website Optimizer (multivariate testing / landing page optimization) is an absolutely critical piece of the web analytics / CPA puzzle that *does* provide exactly that type of feedback. it allows you to conduct a series of experiments that combine various copy & graphics, and select the combination(s) that work best.
wake up folks: time to smell the coffee.
and start COOKING... with GAS.