Expensive : Complicated : Useless

These are words they have used to describe segmentation! Further investigation usually reveals that their segmentation study was conducted so long ago that no one really remembers why. In one case, some muckety-muck told a group of analysts to do it and so they did exactly what he told them to do. No one really knew why it was being done.

Of course then nosy consultants like me come along and ask the question. Why did you do this segmentation? What does it all mean? It used to shock me when they would say that “it’s just the way we have always done it”.

Segmentation is not about grouping data for the sake of grouping data. Segmentation should have purpose. It should be used to solve a particular problem. It is useful for determining sales regions/territories, marketing message and campaign optimization, risk management, web visitor behavior classification etc etc etc.

At the end of the day, your segmentation should tell your group or organization exactly how to focus on a particular audience. Keep in mind that an organizational segmentation is typically a pretty intense exercise, but that you might segment campaign audiences or web traffic on the fly and change it to suit your needs as campaign messaging changes or web traffic ebbs and flows.

Just remember that if your segmentation model does not include factors that drive the problem that you are trying to solve, you should not expect it to be highly actionable. I often am asked “Hey, can you do something with this data? How about a segmentation?” and am given Dun and Bradstreet profiles and customer/prospect information and little else. There is a lot of directional information I can derive about the way that you do business from this data, but I’m sort of left licking my finger and holding it in the air when you ask me how to use the segmentation to drive your media. If the segmentation is actionable, I would say that you should focus on your core, growth and niche segments.

To succeed:
* Start with a problem to solve. It can be as simple as not knowing where the company is the most profitable.

* Insist on historical transaction data that includes both demographic and actionable information. For example, profit margins for sales, conversions for campaigns and goal related metrics for web sites.

* Make sure that you understand the segments that were created and that they make sense to your business. There are n ways to look at data sets. They should be easy to understand as well as to act on. Too many segments can be unwieldy and difficult to act on. Too few can leave you without focus.

The awesomeness of your segmentation depends on following the aforementioned. There are other subtleties to this kind of exercise that are learned over time, but these are basics akin to boil water, put tea bag in cup, pour water. Happy segmenting!

Stepping back: Direct marketing guru and Internet junkie, Gregory Ng, introduced me to twitter. After a couple of days of use, I got the general gist of it, but I really wanted to figure out what I could learn from it. After all, this was really close to my dream of being able to work all day mining AIM or MSN conversations and then return home with my 6 friends to our dark haired princess who washes our clothes and makes us supper. I just want to learn how to serve you better. I promise I am only quasi-evil!

So I built a little MySQL database and PERL script that started storing tweets from people that I felt were influential on twitter. People like Paul Kedrosky and Chris Brogan. I soon discovered that this was the wrong approach. While it’s interesting and helpful to get the opinions and most commonly used word pairs of the experts, it is the twitter population at large that can help you tell a story about what is happening in twitterland. You can use one of the many cool search tools like Summize to get a buzz

pulse, or you can use some of the cool first generation analytics apps, some of which I mention below.

The Twalaxy’s Attempts at Buzz Measurement

• Twitterverse

  is a twitter tag cloud that shows what is happening within the last 1 to 10 hours on twitter. You can set it up to look at one word or two word phrases.

• Intwition

  is a cool tool that tells you the most popular links on twitter for the day. It also tries to suggest what will be popular dude to its hotness within the last 8 hours, and it also gives a 5 day summary.

• TwitterBuzz and even better: Twitturly

  Twitterbuzz is a link aggregator. It shows the most popular links on twitter. It aggregates the primary URLs and then allows you to drill down. The only problem with this is that it aggregates tinyurl, dwarfurl, snurl etc instead of following them to the actual link which would be far more interesting. It also is not setup to capture and display double-byte characters so Chinese tweets are not legible. <a href=”http://twitter.com/waynesutton/statuses/808211269″>@waynesutton just turned me on to Twitturly </a> which I think is even more compelling because it DOES follow the links and has a really slick look and feel.

Twitterverse 2.0

These apps are cute like and they do give us a flavor for what is happening, but what happens if we do more than make a cute app for the sake of saying we did have a cute twitter app? How about some natural language processing to tell us if the message is positive or negative. How about we apply that to a brand? Consider a new competitive product release like the (fictional) Dunkin Donuts Strawberry Cheesecake Latte and (you know you want it) Starshmuck’s Raspberry Ripple Espresso. How about a compete.com-esque tool that allows me to enter the products and gives me a trend line showing the positive or negative buzz about the products?

Even better would be an aggregation of these, similar to what is happening on Trendrr but using natural language processing to care for abbreviations, related works and the like. Trendrr is a great first step and I think could prove to be a bridge to the Twitterverse 2.0.

I will be keeping my bananas peeled on these developments. Tweet me with or comment with others! @schneidermike.

If you enjoyed this article, please tell a friend, subscribe to my feed, or tweet it to many friends. Thank you for reading.