Dynamic filters

Dynamic filters are based on test criteria (e.g., volume greater than zero). When a dynamic filter is on, the widget only includes data for the members that meet the test criteria. The filter updates automatically based on the data context rather than storing a static list of members. Dynamic filters may be saved as dynamic collections so they can be reused and shared.

To apply a dynamic filter

1. Go to the filters area for the widget or column:
   
   To apply a dynamic filter to a widget - Select the widget and click the filter  button in its upper-right corner (visible on mouseover).
   
   To apply a dynamic filter to an individual column (in Multi Comparative only) - Click the column heading and select Filter Date Range.

Click Build a Dynamic Filter.

3. In the Dynamic Filter area, choose the key to test from the Accumulate By menu (e.g., filter out customers that do not pass the test).

4. Set up one or more rules:

• Select the measure to test.

• Select the date range (This or Last) or the variance (Diff or % Change) you wish to test. By default, the date ranges correspond to the widget date ranges. For example, if the widget compares the month-to-date with the prior month, then a rule for "This" would look at the month-to-date and a rule for "Last" would look at the prior month. If the widget does not have a Last date range, then a rule for "Last" will default to one year earlier (YAG) than the "This" date range. If you prefer, you can test a different date range (i.e., something other than the widget date range) by changing the Advanced settings as explained below.

• Select an operator (e.g., greater than).

• Enter a value to compare against.

• To use multiple rules, click Add Rule, repeat the above steps, and choose or to control which rules must be met.

Tip: If the desired key or measure is not available, you may need to change the cube under Advanced Options.

5. Optionally, expand the Advanced Options area and choose settings to refine the test. 

Advanced option	Explanation
Cubes	To test based on data from a different cube, select from the Cubes menu. This controls which keys and measures are available for selection.
Equivalent	To test data after it has been converted to equivalent units (if applicable), select from the Equivalent menu. An equivalent table selected here overrides an equivalent table applied to the widget.
Passed In	Choose the part of the date range to consider for the test. These options affect the outcome when the date range includes multiple units of time (e.g., 12 weeks).  All Periods (In Total) - looks at the entire date range (e.g., total of all 12 weeks). First and Last Period - looks at the first and last units of time (e.g., 1st week and 12th week); both units of time are needed to pass the test. First Time Period - only considers first unit of time (e.g., 1st week). Last Time Period - only considers last unit of time (e.g., 12th week). Any Period (Individually) - considers all units of time individually with only one unit of time needed to pass the test (any week). All Periods (Individually) - considers all units of time individually with all units of time needed to pass the test (e.g., every week).
Date Range	Choose the date range to test. Dynamic Date Range (default) - When this option is on, the test will use the date range of the widget. For example, if the widget shows the month-to-date, the test will look at that month to determine which members pass. If the month updates over time (e.g., November to December) or the widget timeframe changes (e.g., one month to 6 months), then the test will automatically update to look at the new dates. Specific Date Range - You can choose this option to test data for a specific date range, which can be different than the widget's date range. For example, you might want to find new products based on three months of data but look at data for a longer timeframe in the widget. After choosing this option, the date setup dialog will appear. You can then select the date range(s) for the test. Any "This" rules in the test will look at the "This" date range you select here; any Last" rules will look at the "Last" date range. The date range(s) can be fixed or update over time based on the current date (e.g. month-to-date). However, the resolution and number of dates used in the test will not change.
Filters	To filter the data prior to testing it, click the arrow beside the Filters menu. This option will filter the data used in the test but will not directly filter the data shown in the widget. See the following new product example.You can load a collection of saved members or create your own filter by selecting members. As in other areas of dashboards, you can combine multiple filters.

6. Do one of the following:
   
   • To immediately apply the filter without saving it, click Apply.
   
   • To save the filter as a dynamic collection, click Save. Note that this will save the test criteria rather than a list of members.

Example - New product filter

You could use a dynamic filter to find and analyze products that were recently introduced. In this example, a product would be included if it had volume of greater than zero during This date range (based on the widget's This timeframe) and volume of zero or less during Last date range (based on the widget's Last timeframe). 

Now suppose you are interested in products that are new to a particular channel. You could use a filter in the advanced area of the test to find those products.

When this setting is on, the test runs the test on the filtered data (e.g., sales for convenience stores only) but does not directly filter the data in the widget. In this example, the widget includes data for other channels, but the results are filtered on products that are new to the convenience channel.

Example - Active customer filter

You could use a dynamic test to filter on active customers. In this example, a customer would be included if it had volume of greater than zero during This date range (based on the widget's This timeframe). 

If your definition of an active customer is based on a more precise timeframe (e.g., sales during last 12 months), you could assign a specific date range that is not based on the widget date range. This would allow you to change the date range of the widget without affecting the test.