This week, I was able to enjoy all the articles I intended to look at outlined in my previous post. First, I obtained a headache from simply reading about all the terabytes of data points that are collected by web analytics and research tools. I always overly simplified how easy it was to extract user’s movements on the web and convert that data into useable marketing information. However, this perception of ignorance was validated by the 10/90 rule presented in the “Critical Components of a Successful Web Analytics Strategy” article. I don’t think it is unreasonable to pay 100,000 thousand dollars for a web analytics vendor, but holding to the 10/90 rule (which is just an estimation), you would need to pay 1 million dollars to have a team to extract value from that information. For large companies like Amazon or Ebay, they have the capital to make that investment. Smaller companies or startups may not, and therefore might be losing valuable pieces of information since they aren’t able to convert user web use to profitable results.
Also, I always thought of Google as a power house search engine with most of their revenue generated from advertisements. Well now I can understand why their advertising space is just so valuable. As noted in “the Numerati”, Google is the Marquee company, it is built almost entirely upon math, and it’s very purpose is to help us hunt down data.
The challenge will be to build intelligent software that can analyze user data inputs and target value back to the users. I think there are two approaches to this, passive and active. I see much of the passive approach today. This would be advertisements targeted to you concerning items/services you may have already searched or purchased. If you recently searched for or bought a TV online, and then get advertisements online about Samsung TVs, then this is the passive approach. You may have already purchased a TV, and the ad has not reached you in time. Next, is the active approach. These are ads that try to predict your next step, before you have even thought of it. An example of this is the movie to car rental relationship discussed in the Numerati (I am having a hard time believing there is a connection between these tw)o. More often, there are logical connections that can be made. The best one I have thought of is the relationship status on social media sites (facebook, myspace, ect). When someone changes this status from relationship to engaged, this info should be an instant red flag to analyzers. If it were me, I would display copious wedding advertisements to this user. Before he/she even knew that one needed save the date invitations for a wedding, an ad would appear on their screen for this, and out of ignorance, they would click on this ad, and subsequently purchasing these invitations. This is one example that I would use data and web metrics from the internet, and be able to convert this data into value for the user and business.
Subscribe to:
Post Comments (Atom)
We have found first hand how difficult it is to evaluate web sites here at Kelley. We subscribe to a service that is quite costly which generates reports about the traffic on our site. The reports are generated from the information on our servers. It is easy conceptually to think about tracking visitors to a web site, but I quickly found that it is quite complicated. People used to talk about "hits" to a web site. This metric no longer has much meaning with modern sites that include videos, images, documents to download, etc. Each time a server displays one of these items, it is counted as a hit, so looking at hits doesn't give you the information you really need. As another example, about 10-20% of the visits to the Kelley sites are from bots that are crawling the web for various reasons. We have to subtract out this "non human" traffic. The list of issues goes on.
ReplyDeleteF.