🌐 한국어 버전: Read the Korean version at 애널리틱스 고민 노트 (한국어).
📚 Series — Part 1: The Problem · Part 2: Reasoning · Part 3: The Decision · Part 4 Current State & Open Questions (this post)
To be clear upfront: this is not a validation
The merge is too recent, and there’s no long enough before/after window or control group to compare against. So this post is not proof that “consolidating GA4 improved traffic or rankings.” It’s a snapshot of what’s visible right now, plus a set of criteria I can use to check my own judgment later.
The current snapshot
Splitting sessions by hostName in the merged GA4 property over the trailing 90 days looks roughly like this:
| Host | Sessions | Active users |
|---|---|---|
| saramjh.github.io | 672 | 651 |
| 127.0.0.1 (local testing) | 16 | 16 |
| localhost (local testing) | 1 | 1 |
This confirms exactly one thing: the hostname filter works as intended. Local development traffic is cleanly separated from real user traffic, so the “narrow down only when needed” design from Part 2 at least holds up functionally. These numbers don’t represent the merge’s “effect” — there’s no comparable pre-merge period measured the same way to compare them against.
What I still don’t know
- Whether this actually affected SEO. As noted in Part 2, shared root-domain trust has never had a proven causal chain behind it. I’d need to watch Search Console’s indexing and impression trends over months before I could say anything — and right now there’s no basis to judge.
- When to consolidate Clarity. I deferred it because traffic is low right now, but I haven’t set a concrete threshold for when to revisit that call.
- Whether this structure holds up if the number of tools grows a lot more. If data streams go from around ten to several dozen, whether managing them all under one property is still the right call is worth re-examining — along with the free-tier data thresholds at that point.
- Whether this problem is even specific to “solo developer, many small projects, one root domain.” A team running multiple products is a different situation, so I’d be cautious about generalizing this series’ conclusions beyond that specific setup.
For anyone in the same situation
If you’re running several GitHub Pages project pages without custom domains, at minimum it’s worth checking these three things on your own account:
- Whether your custom-domain-free projects really do all sit under the same root domain (
https://<account>.github.io/<repo>/) - Whether you’re actually distinguishing “property” from “data stream” in GA4, or repeating my old habit of spinning up a new property every time
- Whether Search Console is verified as a domain property, or per-project via URL-prefix
This series doesn’t hand you a definitive answer, but I think those three questions are worth checking for yourself regardless. If traffic grows or I learn something that changes the picture, I’ll keep adding to this record.