Most companies hope that one day, they will be able to use website traffic data to predict how their online visitors will behave. In his blog, Avinash Kaushik, author of Web Analytics: An Hour a Day, goes over the six barriers that are preventing this:
MESSY DATA
It’s hard to find complex trends in Web data when so much of it is anonymous or incomplete, Mr. Kaushik says. For example, many people fill out forms with false or incomplete information, or don’t give their full name and address. “When you want to do traditional data mining (and not just analysis) and predictive analytics, all of these things are poison.”
IT'S A 'NONLINE' WORLD
A consumer might go online to gather information about a product and then visit a physical store to make a purchase. Mr. Kaushik uses the term “nonline” to describe the blending of online/offline customer experiences. “People flow between channels and touch points and there could be an outcome … at a completely different channel than where most of the interaction was,” he says.
THE TOO-WIDE WEB
Online users visit a company’s site for various reasons, such as to research something, to find contact information or to make a purchase. Businesses can’t use Web data to predict customer behaviour until they know what their site’s primary purpose is for those who visit. “It is incredibly difficult to take your massive collection of clicks … and then assign them into each primary purpose bucket and then predict on top of that,” he warns.
YOU'RE BACK!
If your business can’t isolate multiple site visits by a single visitor, then making predictions is even riskier. Consumers rarely travel to a store – or even phone them – to find answers to every little question they have. But they do go online, sometimes on several different computers. “You can have the same person come to your website … many times to solve a different issue,” Mr. Kaushik says.
DATA, DATA EVERYWHERE
When reaching customers, companies often have many “touch points,” such as telemarketing, direct mail or in the store. It’s hard to analyze Web traffic if you don’t factor in all of these other business-customer interactions. “If you knew all the costumer touch points and had merged the data, then it gets much, much easier to understand current behaviour and predict future behaviour and outcomes,” he says.
THE WEB IS-A-CHANGIN'
Companies might be able to predict what habits customers are going to form if they also had a sense of how the Internet is evolving. “Doing mining and predictive analytics on past behaviour requires a certain amount of stability about your future (customers, business, outcomes etc.),” Mr. Kaushik writes. “But if the ‘environment’ changes too much, or even enough, then your predictions on past behaviour will have only tiny chances of success.”