During roll out, the system assigns each of the customers in your audience to one and only one of the segments in your model. In order to build intelligent models, you need to understand how the system prioritizes the criteria it uses for sorting.
To accomplish that, lets take a hypothetical business situation and create a segmentation model.
Example 1:
Let's say, that you're building a direct marketing campaign for an online retail site specializing in electronic and computing equipment. We'll call the business CompWare.com.
It's the end of the year holiday season and you have several items for sale that you want to highlight in your e-mail campaign. These items are: ultralight laptops, inexpensive home computers, big screen televisions, digital cameras, and greeting card software.
Now, let's say you want to target each of these products to a different audience as follows:
- Ultralight Laptops – Everyone who has recently clicked on a laptop link
- Inexpensive Home Computers – Those customers who have children at home
- Big Screen Televisions – Men
- Digital Cameras – Everyone who has recently clicked on a digital camera link
- Greeting Card Software – Everyone else
The Connect system stores segment criteria hierarchically, so when the system reviews an individual customer, it checks the assignment criteria for each segment in the same order that you have them listed in your model. Once a customer qualifies for a segment, that customer is not eligible for any remaining segments. Therefore, if you have a filter criterion that is not mutually exclusive, you need to arrange your segments to accommodate the proper outcome.
Warning: If you create a segment in a customer profile segmentation model without any filters attached to it, the system will associate all available customers with that segment. All following segments will not have customers assigned to them.
Given the audience we want to target, we know a few important things:
- Segments targeting big screen televisions must go after segments targeting laptops, home computers, and digital cameras. If you put big screen televisions first, no men would be targeted for the other segments.
- Segments wherein you want to target the most people, such as segments where the targeting is based on an action, should come before segments with a more general target.
- Segments targeting greeting card software must go last, because it's obviously the leftovers. To better understand leftovers are, see the section Understanding Leftovers.
So, our segmentation model would look like this:
- Segment #1 – Targets everyone who has clicked on a link with the interest of laptop within the last year
- Segment #2 – Targets everyone who has clicked on a link with the interest of digital camera within the last year
- Segment #3 – Customers who have children at home
- Segment #4 – All men
- Segment #5 – Leftovers
Now, let’s take two hypothetical customers:
- Joe Meta, who has clicked on several links this year, some of which had the interest of digital camera.
- Sally Beta, who has two children at home and doesn't shop online much, generally spending less than $100 in any one transaction. She generally only clicks on links for children's goods.
Applying these two customers to the segmentation model above, we can see that:
- Joe will fall into segment #2, even though he also qualifies for segment #4, because Segment #2 comes first. Since Joe can only fall into one segment, you can be assured he'll only receive an e-mail for your digital camera promotion from this campaign.
- Sally will fall into segment #3, because she hasn't clicked on any links with the specified interests, but she does have children at home.
You can see why it is crucial that you build the segments carefully.
Example 2:
Let's assume the same business, targeted audience, segmentation model, and customers from example 1 above.
You are pleased with your segmentation model and you present it to your manager, Rosa Mendez, for a go-ahead. When Rosa looks over your model, she is disappointed to see the big screen TV so low on the list. "The TV is going to be our hottest mover this season!" she exclaims. "Fix it."
There are two approaches you could take to fix this problem:
- Create two separate segmentation models.
- Expand your model, creating more segments with mutually exclusive filters.
Each of these approaches has radically different outcomes. Let's look at each one separately.
Approach 1: Create two separate segmentation models
Segmentation Model 1:
- Segment #1 – All men
- Segment #2 – Leftovers
Segmentation Model 2:
- Segment #1 – Targets everyone who has clicked on a link with the interest of laptop within the last year.
- Segment #2 – Targets everyone who has clicked on a link with the interest of digital camera within the last year.
- Segment #3 – Customers who have children at home.
- Segment #4 – Leftovers. To better understand what leftovers are, see the section Understanding Leftovers.
Applying our customers to these models we would see that:
- Joe would receive e-mail from the first campaign, Segment #1. Joe would also receive e-mail from the second campaign, qualifying for segment #2.
- Sally would receive e-mail from the first campaign, Segment #2. Sally would also receive e-mail from the second campaign, segment #3.
You can see how your customers could easily receive more than one e-mail, using the two-campaign approach, but all men, rather than a subset of men, would be targeted for the big screen TV.
Remember that some customers may view your business as sending out e-mail spam if they receive a large quantity of e-mail from you. Think about your customer potential behavior while constructing your segmentation model.
Approach 2: Expand the model by creating additional segments with mutually exclusive filters
Segmentation Model 1:
- Segment #1 – Targets all men who have clicked on a link with the interest of laptop within the last year.
- Segment #2 – Targets everyone else who has clicked on a link with the interest of laptop within the last year.
- Segment #3 – Targets all men who have clicked on a link with the interest of digital camera within the last year.
- Segment #4 – Targets everyone else who has clicked on a link with the interest of digital camera within the last year.
- Segment #5 – Male customers who have children at home.
- Segment #6 – All other customers who have children at home.
- Segment #7 – Leftovers
In this example, you have created a pair of mutually exclusive segments for each of the original audiences you targeted – one that will go to men and feature the big screen TV along with the other product. Another segment will go to the customers that aren't men and would just target a single product. Then the leftover segment will still target the same people – those that don't qualify for the first 6 segments.
Rosa, your manager, loves this model, because it sends the maximum number of offers out on the big screen television, but still ensures that each of your customers will receive only one e-mail from this campaign.