Solved: Discrepancy in Unique Visitors data when using different breakdown methods

[AntonKr3]

I notice a peculiar discrepancy in data when I try to break down the same metric - the number of Unique Visitors - via two different methods:

1. Breakdown by User Account Name, an upper table - it gives me somewhat correct data (the numbers match the real number of clients and users that we know from other sources)

2. Breakdown by Month (or any alternative date variable like Day, Month of the Year, etc.), a lower table - the data for the two periods of the last two years have too high numbers (there's 100% confidence that there weren't so many users back then).

Why does the second breakdown show so high numbers and how could it be fixed?

Eventually, I want to have a correct breakdown by Month to make a simple line chart showcasing the monthly number of Unique Visitors for each of the last 3 years.

P.S. I utilized the same breakdown approach for two other products and their AA suites gave me correct numbers, no issues like this one were noticed.

[Jennifer_Dungan · Accepted solution]

Hi, 

If you are comparing the values of the first table (1294 This Year) to the second table (1339 This Year)... my guess is you had some traffic that didn't properly include your Account Name [p29]... you could try to track this down by creating a segment for where p29 doesn't exist...

As for the second table last year data being so high (Jan, Feb and Mar), my suspicion is that you had an uncaught tracking issue, These hits / visitors didn't gave p29 and so they don't show in the first table... but there is data that was captured.

If you want to show the data side by side like this, but want the values to match for people to compare, you can apply a segment at the panel level (assuming there's nothing else in the panel that will be impacted by this) of:

Hit

    Account Name [p29] exists

This should at least hide all the hits (and Visitors) that didn't have an Account Name associated to them.

[AntonKr3]

Hi @Jennifer_Dungan ,

Thanks for the prompt reply!

It turned out the extra traffic for the past two years was coming from another server with lots of test accounts. After filtering it out, the data started to look correct and the issue is solved

[Jennifer_Dungan]

Ouch.. that's a lot of test data. Glad you found the source though!

If this is potentially an ongoing issue, you might need to update your tracking to send that test data to an alternate suite, or create one or more Virtual Report Suites, one specifically for the "Clean" data, and shift all your reports over to using that VRS (to make sure none of that comes through for anyone).