(Monday, August 7)

7th level of AOHell

Up until last night AOL at around 7pm (I am not sure when it started) I heard AOL Research had been hosting and making available some "anonymous" user data (about 2 gigs of logs) of searches within AOL (which I believe at this time essentially served up results from Google). I don�t know how long these were actually up but from what I understand, this data that did not correlate with any specific users searches, but was 500k searches of random pulls from their data. This amounted to a wet dream for a lot of online marketers, and will probably be a "search privacy" media fest once it hits the mainstream news. Glossing over a lot of uses for this mass amount of data here is who it could really effect in my opinion: Google (and to a lesser degree Yahoo/MSN) - Basically when you search for something there are organic results (usually on the left hand side of the search engine result page [ aka serp]) and depending on the search engine, paid for results on the top and on the right. These "paid for" results are usually of a PPC (pay per click) business model, which in general means that a company is paying the search engine every time a person clicks on their add. This could be 5 cents a click to $25 a click (and much higher in some cases). The advertiser hopes that this click payment will convert to a sale or something on their end, hence you don't see bob's blog paying for these things, it is usually someone with something to sell. The price for these clicks depends on a large amount of variables, but one is the simple fact that someone else might be wanting to also pay for a specific query. For example, there may be 10 companies that make tents and they all want to be shown on any query for "buy camping equipment", but there are only 3 spaces on the page, so they bid against each other to see who can be placed higher (and /or make it on the page). Well the reason that all 10 companies want to get on the results for that page is because it is really hard to get on that page for organic results (just the websites that naturally make it onto the non paid for portion of that site). So this advertising method is great for new businesses that may not rank well in the search engine results, but that want to be on the serp pages for specific terms. If the costs of the pay-per-click advertising still brings people to their site to order then they can bypass trying to rank in the organic search results. The more people that are competing for a phrase (ie our "buy camping equipment") then the more bidding is done (and some other variables) and the higher the price to pay for that keyword (which I assume means a higher percentage for the search engine doing the selling). So this system works because everyone is happy. The Search Engine is making great profits, the seller is getting to people to their website which converts to sales for them (barring the price they pay for people is worth it of course) and the user *hopefully* finds what they want on the site they go too. The "hopefully" part of that I am going to save for another post sometime because I have many opinions on that that range from 'Does the common person searching understand these are ads and not necessarily what the search engine thinks are the best results?' etc. So this is a very large and generalized view of the how the system works, when this data comes into play everything gets messed up. The one thing that this model gives as an advantage to the search engine is no one besides the search engine knows what the mass amount of people are searching for (what their intentions are). This is almost a guarded secret, because what if instead of "buy camping equipment" more people are putting in a more specific query like "show me a 1 man pup ten that weighs under 2 lbs". Well if the market (people selling the tents and ppc advertising with the search engine) know this then first they are going to try to organically rank well for this specific term, if they cannot organically rank well, then they will purchase this term. The only thing is that this term (which might get a lot more eyes as told from the data) might only be purchased by 2 tent sellers. Because it is so targeted no one ever thought it would get enough eyeballs to warrant paying money for it. So instead of paying money to advertise for the term "buy camping equipment" which was $5 a click (this is just a made up number) they now can pay $.50 per click for "show me a 1 man pup ten that weighs under 2lbs" which is great for the advertiser. They know (from the data) that more people are searching for this, which means they reach their audience more and also no one else bidding on this term (which makes it way cheaper). Now the search engine (and other search engines) want the competitive bidding because (I assume) they make more money from the percentage they charge for the bid. This information out in the open, kind of messes up their system. Now people are more focused and targeted in both search engine optimization and also in the pay per click bidding. It loses the "Fog of War" in the realm which was an advantage to the search engine selling the placement on the terms. The way in which this AOL data was also provided can hurt the consumer (or at least the people who use AOL) also. I ego search all the time, I admit it. I have also searched the last 8 digits of my credit card number before on search engines to see if my credit card number was listed on the net anywhere (I would not recommend doing this now). But I also understand a little about how search engines work, so I of course didn�t search for my credit card number say with my name. Well I am sure someone has before and that search might be in that 500k queries of data. Also if this was the same people doing the searches and they have the same id associated with all the searches then it is easy to aggregate all the searches for one specific person (thus seeing that person a searched for address x and name y in the span of 2 mins (I assume the time data is there). That would be easy to correlate with any large database. I will try to go back over and revise this post and elaborate some more with links to other people�s thought on this matter (which I have read some this morning) and also list some good references for some of the points. It is near the end of my lunch break so I need to get back to work. I have another blog in the works (http://www.cyancode.com) which is going to be specifically my coding blog (with some SEO/Search write-ups also in there). Where I am going to start posting more technical posts like this. Just because I don�t want to post a bunch of technical stuff on this blog and bore everyone that comes here for other subjects. Until then feel the wrath of my bad grama!