About time boundaries for trends

Trend components can filter the data in a Container or a Select component based on time periods. Some of these time periods have a natural definition (such as calendar day) and some do not (such as rolling month).

There are two types of time periods: calendar and rolling.

Rolling periods

Start at the time of day noted in the timestamp of the transaction, and go back to that same time of day on an earlier day. The interval (such as day, week, or month) can be specified.

Calender periods

Calendar days, calendar weeks, and calendar months always begin and end at midnight.

When you define a trend component, you select calendar or rolling for only the last time period, and Opportunity Detect adjusts all previous time periods accordingly. For example, if a spike period is rolling, then the historical period is also rolling. If the spike period is calendar, then the historical period is also calendar.