01
April
2010
|
04:06 AM
America/Los_Angeles

Data Mining: Rethinking Advertising From The Online Publisher's Viewpoint

As the publisher of an online news magazine I know my readership better than anyone. And that's true for any other publisher.

Yet the current way of selling advertising is to try to convince companies to advertise on your site by producing a media kit that shows traffic, the demographic profiles of readers, and all the other stuff. That's the way advertising has always been sold.

But why should I have to convince advertisers that they should advertise when I know from my traffic and readership data that they would do well on my site?

Print publishers are having trouble transitioning to a digital business because they are still trying to do business the old way. Yet things are much different in the online world and there is no reason to do the business of selling advertising the old way.

Print publishers have to conduct focus groups and ask readers what they read, what they like, and who they are. The online publisher knows all of that all of the time, and can apply a lot more creativity to building a thriving advertiser community. There is a tremendous amount of user data that can be mined and converted into valuable information.

Here are some suggestions:

- Since I know my readers better than anyone else I should know what types of services and products they are likely to be interested in. I should identify the best companies providing those products and services and approach them to advertise. This also ensures quality and that readers won't be victims of many scam and spam advertising that is polluting the Internet.

- I could offer my chosen advertisers a performance based plan confident that I can deliver the conversions, clicks, leads or whatever the performance metric suits the advertiser, because I know my readership.

- Instead of an ad server trying to determine where to put advertising I would control on which pages the advertising appears, even what time of the day, because my server data would show when those adverts are likely to be the most effective.

- I could perform A/B testing of advertising messaging to figure which is the most effective messaging for my readers.

- I should refuse to take advertising from companies that I know won't do well with my readership. After all, why subject my readers to promotions that are of no use to them?

This type of approach would build stronger bonds between publishers and advertisers--and between publisher and readers. Both would have a higher quality experience.

Advertisers would find a better fit and performance, and readers would trust that the publication is looking out for them -- rather than trying to monetize the reader relationship in whatever way it can, appropriate or not.