Bookshelf Home | Contents | Index | Search | PDF |
Defining Attributes and Buckets
Attributes are discrete variables that may be grouped, or rolled up, into collections of related information. Examples of attributes are geographical values (city, state, region) and time values (month, quarter, year). The data elements that make up attributes are derived from attribute tables in your data warehouse.
Creating a rollup is simply ordering attribute values so that you can segment the data at the highest level, or at any level in between. For example, ZIP Codes roll up into counties, counties roll up into states, and states roll up into regions and markets. You can group attributes into families, then order the values in each attribute family into different hierarchies for convenience and to make the data easier to select for segments, filters, and custom measure aggregations.
You use the attributes you build in Siebel Marketing for many purposes:
- To create groupings for time, demographics, and product data.
- To provide record filtering abilities (for example, only analyze customers who live in Michigan and have bought a specific product).
- To provide customer segmentation (for example, segment customers by account balance, resulting in mailings to customers who have a high or very high balance).
In Siebel Marketing, there are number of ways to group data values:
- Buckets. These are ranges of continuously varying numbers that reference measures, for example, high, medium, and low. For buckets, you create and produce counts based on the value of the measure. For example, a marketer creates a bound measure that is mapped to the ANNUAL_INCOME field in his company's Customers table, and derives numeric information from that field.
The marketer decides to group ranges of income into buckets and then count how many customers fall into each bucket. Table 24 shows how annual income data can be divided into distinct ranges. Siebel Marketing calculates the total balance for each household, determines the qualifying bucket for the household, and adds one to the count for that bucket.
Table 24. Measure: Total Balance by Household Buckets Definition Low <$20,000 Medium Between $20,000 and $50,000 High Between $50,001 and $100,000 Very High >$100,000- Attribute Families. Logical families of related data mapped from fields in a table.
Attribute families are unordered collections of related information that are used as the basis for building attribute hierarchies. Each attribute family can have a number of attributes, and each attribute, a number of values.
Like buckets, attribute families are mapped to a base table and field in your database.
- Buckets reference a measure that is mapped to a numeric database field.
- Attribute families provide a loose structure for individual attributes that reference and derive information from other alphanumeric fields in your database.
For example, a retail marketer wants to group account data to track how many accounts were opened in certain time periods. The marketer creates an attribute family called Date Opened and maps the attribute family to the Accounts table in his database and the DATE_OPENED field in the table. The marketer decides to set up quarterly, yearly, and two-year ranges for the data.
Within the attribute family, the marketer creates the three attributes and defines values for each, as Table 25 illustrates. Attributes are mapped to the accounts table and DATE_OPENED field.
Although these data elements are assigned to an attribute family, the marketer can organize the individual attributes in a number of different ways using attribute hierarchies.
- Attribute Hierarchies. These are used to order the various attribute values found in attribute families.
Attribute Hierarchies allow you to look at your data in different ways by allowing you to create distinctive rankings, or hierarchies, using mapped, unordered attributes of an Attribute Family.
Using the previous example, the attributes in the Date Opened attribute family might be arranged in a hierarchy with two-year periods, yearly, and quarterly as successive levels. The hierarchy levels also might be arranged as quarterly, yearly, and two-year periods. An attribute hierarchy can have up to nine levels.
A retail marketer might design different hierarchies using the attributes in the Products attribute family. Individual attributes such as Product, Category, Brand, and Size might be arranged in hierarchy levels such as Category, Brand, Product, Size, or Brand, Product, Size, Category, and so on. Four data elements in an attribute family can be arranged into 64 different hierarchies.
When attributes are arranged in a hierarchy, they form rollups.
Using a bank as an example, if the hierarchy is State, Region, and ZIP Code, the measures (Number of ATMs and Number of Tellers) that can be calculated for the ZIP Code roll up to provide a total for the Region, and the Regions roll up to provide a total for the State, as shown in Table 26.
ZIP Codes roll up to Region (12+8+20=40 for the South and 7+35+22=64 for the North). Regions roll up to the state level (64+40=104).
You can create as many rollups as needed. For example, even though the data warehouse may not have an attribute for All States, you can create a rollup that combines the information from every state to give you the total you need.
Bookshelf Home | Contents | Index | Search | PDF |
Siebel Marketing Guide, Version 7.5, Rev. A Published: 18 April 2003 |