Page last modified 14:17, 31 Jul 2009 by Admin

Measures

1. What Is a Measure
1. 1.1. Using Measures for Events that Do Not Occur
  1. 1.1.1. Tracking for Job 1234
  2. 1.1.2. Opens for Job 1234 Data Returned by Measure
2. 1.2. Creating Measures and Conditions
3. 1.3. Adding Data to Return Targeted Results
2. Why Use Measures

This document contains conceptual and procedural information about measures. This document is intended for marketers, managers, and application users responsible for analyzing tracking information. Understanding of relational data structures and SQL is helpful when using measures.

If your account is integrated with a CRM system, measures cannot act on the data in your CRM account. Future enhancements will support this functionality. Until then, you can export behavioral data from your CRM system and import it into a data extension so that measures can act on the data.

NOTE: Contact your ExactTarget representative to request the Measures feature for your ExactTarget application account.

What Is a Measure

A measure is a unit of subscriber behavior that you define in your account. For example, if you use dynamic from addresses on an email job, the tracking for the job is all grouped together. You cannot easily find the statistics by from address. With measures, though, you can define a statistic given the conditions of the job ID and the from address.

After you create a measure, you can include the measure as criteria in a data filter to segment a subscriber list using the measure.

Defining Measures

Your define measures based on the behavioral data tracked by the system. The following categories of behavioral data are available:

Bounce - The bounce data for emails sent from your account.
Click - The link clicks for emails from your account.
FTAF - The information related to email message from your account being forwarded to a friend.
Not Clicked - The information related to email message from your account that were not clicked.
Not Opened - The information related to email messages from your account that were not opened.
Open - The email opens for your account.
Sent - The information about jobs sent from your account.
Survey Response - The information from subscriber survey responses.
Unique Click - The unique link clicks for emails from your account.
Unique Open - The unique opens for emails from your account.
Unsubscribe - The unsubscribes from email lists in your account.

Using Measures for Events that Do Not Occur

Suppose that you send an email to five people using Job 1234. Of those five people, four people open your email at least once, and two people click through your email.

Tracking for Job 1234

Subscriber	Opens in Tracking
A	2
B	1
C	0
D	3
E	1

If you create a measure to count total opens for Job 1234, the measure query behind the scenes will return data for four people, not five. Here is an example:

Opens for Job 1234 Data Returned by Measure

Subscriber	Measure Value
A	2
B	1
D	3
E	1

Measures calculate a total value for subscribers, where the value is greater than zero. If the value is zero, meaning there are no statistics, the value is not calculated for a subscriber. In other words, each open for a subscriber is counted. If a subscriber does not perform any Opens, they cannot be counted.

You cannot segment based on a subscriber not opening an email by including a measure in a data filter where opens equals zero. This does not work because the open file does not contain records where open equals zero.

Instead, use the Not Opened (or Not Clicked) event to segment on unengaged subscriber. These events count the opposite of opens and clicks. In the example of Job 1234 above, the Not Open event would have one record returned in the measure result set.

Note: When using Not Opened or Not Clicked, make sure to specify in your data filter where > 0 means "give me subscribers where the Not Opened count is greater than 0."

Creating Measures and Conditions

You create measures on the Subscribers tab. After you choose the behavioral data file to be the event source for the measure, you create conditions that the system evaluates each tracking event against to determine if it qualifies for the measure. For each condition, you select:

The behavioral data attribute column to evaluate
The operator to use to compare the values (See the list of operators.)
The value to compare against

For example, you might create a condition where:

The behavioral data attribute is EVENT DATE
The operator is IS GREATER THAN
The value is 1/1/2008

Only events that occurred after January 1, 2008 qualify for this measure.

You can put multiple conditions together in a grouping using an AND or OR operator. You can also put multiple groupings together into an outer grouping that uses an AND or OR operator. When you use the OR operator, a row that meets one or more of the conditions satisfies the grouping. When you use the AND operator, a row must meet all of the conditions to satisfy the grouping.

Adding Data to Return Targeted Results

To return the most specific data and to protect performance, you should provide as much information as possible to narrow the scope of your measure:

Give your measures a specific time range. By default, you can select records from the last 180.
Use job-specific criteria for further narrowing. For example, select records where Sender Profile is X or Email Name is Y or Subject Line is Z.
Add job criteria to narrow the scope for the measure, enabling more indexing and more performance.
When qualifying a measure, ask yourself why "Clicks in the last 60 days" is valuable unless it is qualified. Click on what? A link that starts with? A particular email? Qualify to narrow the scope of exactly what it is--an Unsubscribe click is also a Click if unqualified.

Monitoring Note: The best way to monitor is to use a data filter, in a filter activity, tied to a program. This way, you can see status (running or completed) and display notices when running a massive segmentation.