Products
Webtrends Analytics 9.x
Webtrends Analytics 8.x
Issue
Why are visitor counts in reports higher when using a custom date range?
Resolution
When using a custom date range, visitor counts cannot be calculated in real time and therefore display higher values in reports.
Example:
From 05/18/2023 to 06/19/2023:
Visitors : 467
Visitors Who Visited Once : 109
Visitors Who Visited More Than Once : 358
Average Visits per Visitor : 5.39
Visits 2,519
From 05/18/2023 to 06/17/2023:
Visitors : 612
Visitors Who Visited Once : 191
Visitors Who Visited More Than Once : 421
Average Visits per Visitor : 4.11
Visits 2,516
Visitor statistics are not always going to be accurate using custom date ranges.
If a visitor comes to your site on Monday, and again on Tuesday, he is going to be counted as a separate visitor if we look at each of those days individually. However, if we look at those two days together, he would be considered a single visitor.
The database tables are compiled into five separate time “buckets” that are used for reporting. They are:
Daily
Weekly
Monthly
Quarterly
Yearly
When looking at a report for a given day, all of the information for that day has been aggregated. This means, if a visitor comes to the site, leaves, and comes back again, we have only counted them as a single visitor (assuming the unique identifier for that visitor remains the same).
The next larger time span is a week. If we look at a weekly report, and if the example visitor has been to the site on Monday and Tuesday, this report will have aggregated that data so that they appear only as a single visitor.
There is not a two, three, four, five, or six day time bucket. If we look at a report using a custom date range so we only get data for Monday and Tuesday, there is no single table covering both of those days. When the request is made for this report, the only option for compiling the report is to sum the numbers. That means in the weekly report, the visitor will only show as a single visitor, but the custom two-day report will show two separate visitors.
The only way to do this differently would be to try and aggregate the numbers on the fly when the custom date range was selected. However, this would literally require a complete reanalysis of the data for that time period.
The result is that the date range chosen requires Webtrends to make a choice on which buckets it is going to use to fulfill the report request. It will take the largest buckets it can through the given time period. In this case, a start date has been chosen that begins in the middle of a week; a Wednesday, to be precise. The fact that this is not the beginning of a month, and in the middle of a week, means only daily buckets can be used up until the end of that week.
While assembling the report, the following days will not be aggregated:
5/18/2023
5/19/2023
5/20/2023
5/21/2023
In the first example, the next four weeks can be taken together, and so we will be able to use the entire weekly buckets through 6/18/2023, plus the final daily bucket of 6/19/2023. However, in the second example, the fourth week (and a day) have been truncated, so now only the next three weekly buckets can be used, and the remaining six days require the use of the daily buckets.
All of the data from each of these buckets is being summed. Returning to the first example, the visitors are added up in the following manner:
5/18 + 5/19 + 5/20 + 5/21 + Week22 + Week23 + Week24 + Week 25 + 6/19
In the second example, they would be added like this:
5/18 + 5/19 + 5/20 + 5/21 + Week22 + Week23 + Week24 + 6/12 + 6/13 + 6/14 + 6/15 + 6/16 + 6/17
Remove a complete week of aggregated visits and change them to be summed, and a higher number is going to be reported nearly every time. The only way it wouldn’t is if all of the visitors only visited the site once and never came back again.
Notice that the “Visits” count does go down? This is because a visit can be considered separate from a visitor. The visit is only the time that the visitor was on the site, so summing this number makes a lot more sense. The exception to this is a visit that spans two given time periods. A visitor whose visit spans two days, or the split between two weeks, will have the visit counted twice. However, this is not common enough to throw the numbers off as much as the visitor counts.
In the mean-time, visitor counts using custom date ranges will often produce unexpected results. For accurate visit counts, the regular date ranges need to be used.