• by Matthew Roche , September 20, 2007

Of the different types of optimization technology I often write about, targeting is the simplest way of exploiting differences among your visitors in order to draw in greater revenue.

If you find that different groups of people - say, visitors who arrive from Yahoo vs. visitors who arrive from Google - respond differently to different layout or copy, for example, then you should make sure you can show those groups different content to make the most of their preferences.

Like testing and experimentation, targeting is a "white box" approach that allows you to actually watch what is taking place and to see how different groups of visitors act. Whereas black box approaches use technology that takes place behind the scenes, a white box approach lets marketers set up tests, with a control, learn what visitors are responding to, and make decisions about future interactions based on those actions.

Most marketers are already doing targeting to some extent. The bulk of landing page optimization, for example, is a combination of experimentation and targeting. When a marketer makes a decision about what page he wants his search traffic to land on (interior site page vs. home page), that's targeting. "Rules-based" targeting is simply doing it faster, and under more defined circumstances, to get more yield.

Adwords is a perfect targeting example. When you specify which words you want to buy, along with other rules like type of match and bid, you are defining rules-based targeting for your prospects. 

Targeting and rules-based targeting are probably the most prevalent form of optimization. In its simplest form, targeting can be used for landing page optimization by showing specific content to visitors based on the keyword they typed.

But targeting is by no means trivial. Behavioral targeting can be very sophisticated, from profiles or persona-based targeting to scenario-based approaches that can model a complete customer purchase cycle and target content by stage or maturity.

Targeting's strength is its simplicity and transparency. That is why it is likely the most prevalent form of optimization or personalization in use today.

At its most basic, targeting is about searching for groups of people that respond similarly. First-time buyers may respond in one way, while those who buy several times a year respond differently, and those who have yet to make a purchase respond in a different way still. Successful targeting exploits the differences between those groups by showing different things to the different groups.

In order to perform rules-based targeting successfully, you must be able to do several things:

1. Identify different groups of visitors that will behave differently. Different groups might be those who come via natural search vs. those who come from paid search, email recipients who tend to respond vs. those who don't respond, regular buyers vs. non-buyers, etc.

Note that every company likely has at least four different groups of visitors. These groups may be divided according to your offerings (people who search for home mortgages vs. people who shop for car loans) or according to gender, geographical area, or other attributes that are ripe for exploiting.

2. Identify the elements that "matter." While you may have identified several different groups that have the potential to behave differently, they may only behave differently about certain things. Weekend visitors may act differently than weekday visitors in that weekend visitors respond better to longer-form content while weekday visitors respond better to bullet points. On the other hand, the two groups may respond identically to what image you show - product photograph vs. lifestyle shot.
 
This step is a matter of choosing, among the hundreds of elements within a site or page, what will make a difference.
 
3. Have the ability to run tests, quickly and easily. Of course, you may isolate several elements that you think will matter among different groups, but you won't know if they do until you run tests to see what works. If you believe that people coming from Google will behave differently than people coming from MSN, you need the ability to have a control, to show the control to a certain number of both groups, and to test your variations to both groups in order to learn whether your hypothesis was correct.
 
4. Have the ability to show different things to different groups of people on an ongoing basis. Once you learn that various groups do indeed behave differently, you need the ability to continue to serve those groups different content based on their needs. You must, in other words, be able to set the rules, to say, "In x circumstance, when a visitor behaves like y, I want to serve z content."

The key here is that you must be able to do this without having to go through changing routers, reconfiguring the page, or any other action that requires software development any time it needs to happen. You must be able to do this on the fly, quickly, in order to be able to use it all the time.

If you can do this quickly, it's yours - it's a marketing decision. If it takes too long, or if it requires IT involvement, it's an IT decision and is no longer useful to marketers.

Targeting doesn't happen by magic. You have to set it up and define the rules. You have to be able to test, and then serve different content on an ongoing basis.

But because almost every company has at least four significant customer types that have measurably different buying behaviors, targeting is invaluable in learning how to recognize them and exploit their differences.

************** 

 
Insiders Wanted

Are you an Insider? Do you have inside knowledge about online publishing -- and do you want to share your insights, observations and points of view with the readers of MediaPost? To be considered for the Insider team of online media influencers,please send pertinent info about your credentials, plus one column idea, to Insider@mediapost.com.