Google Analytics：带GUID过滤时Unique Page Views高于Page Views的原因咨询

Let’s break down this counterintuitive result—it all boils down to how your filter is applied across the two metrics, plus the core difference in how Page Views (PV) and Unique Page Views (UPV) are calculated.

First, let’s ground this in basic definitions to set context:

• Page Views: Counts every individual page load/refresh that passes your GUID filter. Each valid request gets counted separately.
• Unique Page Views: Counts each unique user (in your 7-day window) who had at least one page load that passed your GUID filter. Even if a user has 10 total requests but only 1 is valid, they’re still counted once in UPV.

Your filter regex ^[a-zA-Z0-9_=-]+$ tells a clear story in the numbers:

• 93.88% of all unique users had at least one valid GUID request (hence the near-complete retention of UPV)
• Only 60.27% of all page requests had a valid GUID (so PV is heavily truncated by the filter)

If you do the quick math (2,355,275 / 2,936,422 ≈ 0.8), the average valid request count per qualifying user is less than 1. That means most users only had one valid GUID request, while their other page loads had invalid or missing GUIDs that got filtered out of PV.

Common Scenarios Causing This Discrepancy #

Here are the most likely reasons users have partial valid GUID requests:

• Cookie Loss/Blockage: A user gets a valid GUID set in a cookie on their first visit, but later requests lose the cookie (expired, deleted, or blocked by the browser). Those subsequent requests fail the GUID filter and don’t count toward PV, but the user is still counted in UPV because their first request was valid.
• Inconsistent GUID Generation: Your system might generate invalid GUIDs in edge cases (e.g., cross-domain requests, privacy mode, or server-side glitches). For example, a user might have one valid GUID request, but all other requests get a malformed GUID that doesn’t match your regex.
• Filter Application Granularity: Many analytics tools apply filters to PV at the individual request level, but apply them to UPV at the user level. So even if a user has 5 total requests but only 1 passes the filter, PV counts 1 while UPV counts the user once.

How to Confirm This #

To verify this is the root cause, you can:

• Check user-level data to see how many valid GUID requests each qualifying user has. You’ll almost certainly find a large chunk of users with only 1 valid request.
• Audit invalid GUID requests—are they missing GUIDs entirely, or do they contain characters outside your regex (e.g., periods, spaces, or special symbols not allowed in [a-zA-Z0-9_=-])?

In short, the mismatch happens because your filter cuts far more individual requests than it cuts unique users.

内容的提问来源于stack exchange，提问作者Moogle