FAQ 4: Real-time Versus Historical Statistics

Question

Why is there a mismatch between 24-hour real-time and equivalent historical reports for the ‘same day’?

Answer

If you are comparing statistics between a real-time module and a historical report on a particular day, you should do this a short time after which there has been no communications to and from your organisation. The chosen time can be anywhere in a time-duration window starting at least 12 minutes from the last interaction and ending at midnight (real-time statistics may be reset at midnight). The 12-minute wait is necessary because it takes this long for communication data to be written to the historical data sources and imported into VIEW. See How VIEW Works.

If a mismatch persists, consider looking at the possible causes described in the sections below.

Historical Metric Design

The design of a historical metric may not warrant comparison with a real-time statistic. If the historical metric in question is a custom metric, you may need to edit it so that its behaviour is close to that of the real-time statistic; if it is a standard metric, you may need to create another metric that is based on the standard one (see Create a Metric Based on an Existing Metric).

A standard or custom historical metric is built from items such as filters on raw historical data, aggregate fields, or other metrics. You should check how the metric is built by looking at these items. Do this by using the metric's or button in the Report Builder's Metric Selector panel.

Whether you choose to design a new metric or modify an existing one, it is recommended that you follow the Metric Design Strategy. This will help you to ensure that your metrics are reporting the correct statistical values.

Statistics Across a Midnight Boundary

If your organisation has handled communications over a midnight boundary, some interactions such as calls and web chats may have straddled this boundary and resulted in a mismatch across real-time statistics and the data seen in historical data export fields and metrics. The mismatch occurs because of the way in which historical records are processed.

In a historical report that is run on a VIEW dashboard, the start date marked against a data record (including any metric that uses the record) is the date on which the record was counted. In the example below, the record started on 24/03/2021 and ended at 25/03/2021. The record's data is therefore included in the statistics for 24/03/2021.

Midnight boundary
24/03/2021--	--25/03/2021
Record	(of start date 24/03/2021)

 

If the report is run as part of a scheduled export, it is the end date marked against the data record that determines the date on which the record is counted. Using the example above, the record's data would be included in the statistics for 25/03/2021.