I had the chance to ask Eric Siegel, the conference chair for Predictive Analytics World, a few questions. The conference is being held in San Francisco in February.

Why do you think there is a need for a conference only dedicated to predictive analytics?

Eric: No matter how you use predictive analytics, the story is the same: Predictive scoring of customers optimizes business performance. Because of this, predictive analytics initiatives across industries leverage the same core predictive modeling technology, share similar project overhead and data requirements, and face common process challenges and analytical hurdles.

There’s plenty of activity out there to cover. The conference program is booked with over 25 speakers presenting case study after case study, from companies such as Amazon, 3M, Charles Schwab, Google, Hewlett-Packard, Netflix and more.

What is your definition of predictive analytics?

Eric: Predictive analytics is business intelligence technology that produces a predictive score for each customer or prospect. Assigning these predictive scores is the job of a predictive model which has, in turn, been trained over your data, learning from the experience of your organization.

For more information, see our Predictive Analytics Guide.

What’s Jim Sterne’s involvement in the conference?

Eric: Jim Sterne is serving as a priceless adviser helping us steer the kickoff of Predictive Analytics World. In fact, PAW is run by the same killer team behind Jim’s eMetrics Marketing Optimization Summit, headed by Show Director Matthew Finlay.

Who’s Prediction Impact?

Eric: Prediction Impact is my consultancy, offering services and training in predictive analytics.

I had the pleasure to ask Sampsa Suoninen, web analytics technology consultant at Trainers’ House Analytics, a few questions despite his tight schedule. Enjoy our blog chat!

Being an independent (working with multiple tools) web analytics implementation specialist, what challenges have you noticed that the various web analytics tools have in common?

Sampsa: Most of the tools have challenges when working with them and depending on the tool this might be data collection, reporting, installation and so on. The level of implementation work needed for tools varies greatly as well, so the only thing that really comes to mind that is true for all is a way to easily gain insight on how to acquire the parameter values automatically for advanced implementations (however, I heard that WebTrends released a tool for this purpose recently, so that kind of ruins my point). However the thing that they all could work on is the licensing and pricing models for different size and type of customers. There is no single vendor that could flex their pricing based on needed functionality and licensing options so that I could’ve included it in every evaluation of WA tools for customers.

What’s the one feature that they have all forgotten to include?

Sampsa: This is a hard one, since if I put together all the properties of all the systems I know, I think a “perfect” combination could be made. But one thing I’ve been struggling with and that is pinpoint accurate measuring of streaming media. I’m not talking about basic level, but to actually gain insight of what was listened, for how long and by how many people. My target is also to find out what was the content of the broadcast in question at a certain time and how many unique listeners/viewers were there at that time.

What do you think is needed to persuade vendors to agree on a common, cross-platform, tag?

Sampsa: I don’t think many of the vendors are really interested in building such options since it might make it easier to compare different tools. The main reason might be that many of the vendors have quite a custom way of setting their special variables, which are set under various conditions and work in various ways, so they would have to include this to the common tag and this would require them to open up the code they are using.

What’s the number one mistake made when implementing a web analytics tool?

Sampsa: I would say wrong account ID errors are the biggest number that comes up. Nothing ruins your report like the data going to the wrong account and therefore showing nothing for the one you need and double for the one it goes (and it cannot be separated from the traffic to the other account to get the reports out).