Conversation Set Filter behaviors

When creating a new conversation set, you must provide filter parameters in order to reduce the resulting volume of listed conversations. The filter parameters you set, and how you set them, allows you to do this with specificity, such that your conversation list reflects only those conversations matching your criteria.

The Create Conversation page allows you to quickly configure a conversation set but relies upon several non-editable filtering behaviors. Before creating a conversation set, it's important to understand the overall effect of having multiple filtering elements at work upon data, in particular the underlying logical operations at play when two or more filters or filter conditions are configured.

The is a collection of configurable filters. They are grouped into areas according to what they generally represent (example: KPI, User Attributes). Some filters ask for more than one filter value either to allow for range specificity or drill-down specificity. Some filters also allow you to input more than one value as a grouping mechanism, to restrict reported data to only those conversations for which ALL filter targets in the filter group are present.

Screen capture of Create Conversation page

Between filters

When you configure a conversation set with two or more distinct filters, the relationship among them is characterized by logical AND operators. That is, data must meet the filtering criteria of all configured filters to be reported.

An example of this is the User ID filter and the Language filter being configured. Data must have the filter-specified User ID value and Language value in order for data to be reported.

Multi-value filters

Several filters allow you to specify more than one value. How these multiple values interact with one another and what their aggregate effect is on the conversation set's filtering depends on which filter it is and what the meaning is of having multiple values in that filter. The reasons for having multiple values in a single filter can be divided into four categories: Drill-down specificity, range specificity, allowing for several true filter states (logical OR), and the grouping of filter values.

Drill-down filtering

There are some filters having more than one value to allow further specificity, for example the Q&A Node and the Data Access filters. Filters of this type typically have two or three columns. The left-most condition for one of these filters is typically the name of the state or task. The conditions to the right of this progressively add specificity to the filter. The filtering effect of the drill-down components in a single row is a restrictive one and the components relate to one another with a logical AND operator.

Screen capture of Q and A node filter as example of drill-down filtering

Range filtering

There are several filters within the Conversation Set Filter page that allow you to reduce the reported data by restricting conversations outside of a specified range. They are all in the User Attributes area of the interface and include Duration in seconds, No Match Count, No Input Count, Collection Count, Confirmation Count, Voice Count, DTMF Count, Select Count, Text Count, and Turn Count. Each of these general filters includes two filters to allow you to specify beginnings and ends of the ranges you would like to see listed. When used in this way, one would typically select a greater than operator to go with the lowest limit of the range and a less than operator to go with the upper limit, thereby excluding all conversations below and above the bracketed range. In this case, the relationship between the filter specifying the lower limit and the filter specifying the upper limit is characterized by an AND operator.

Illustration showing location of both bound filter components of a general range filter

Illustration of range filtering example with bound filters excluding all but bounded range

You may decide, instead, to exclude all conversations between the lower and above the upper limits, but include all conversations below the lower limit and above the upper limit. In this case the relationship between the two filters is characterized by an OR operator.

Illustration of range filtering example with bound filters excluding a bounded range

Comma-separated values

Several filters allow you to configure multiple values as comma-separated entries within a single filter field. Filters that fall into this category are:

  • Any that allow a typed entry (for example, Session ID, User ID).

  • Any with a drop-down list of multiply selectable values (for example, KPI: Exit Reason, dimension filters)

    Note: In the case of the drop-down list, mousing over the field displays a tooltip listing the multiple entries.

In all cases, comma-separated values in the same filter field relate to one another with a logical OR operator. A conversation's data is reported if any of the configured filter values are true.

Screen capture of Session ID filter as example of comma-separated values in filters with typed fields

Screen capture of Standard KPI filter as example of comma-separated values in multi-select filters

Multiple rows

The Task, Q&A Node, Decision, and Data Access filters allow you to configure multiple filter targets by adding rows. The filter targets from multiple rows become an aggregate grouping and are filtered as a single block. That is, a logical AND operator separates filter targets represented as rows such that only conversations for whom all of the filtering targets are true will be reported.

For example, in the Task filter, if there are three filter targets (three rows), only conversations with matching Task parameter values are reported in the conversation set.

Screen capture of Task filter as example of filter with multiple rows