Why does the metric look clean in the interface?
GA4 gives you new vs returning users by channel with a few clicks. Clean percentages, nice chart, stakeholders happy. I reported this metric for months before I started questioning what it was actually counting.
What is GA4 actually measuring?
GA4 defines a "new user" by device. One person visiting from their phone and their laptop is counted as two new users. In any product with decent cross-device usage, this inflates new user numbers - sometimes significantly. The interface doesn't flag this anywhere.
By channel, this distortion isn't uniform. Mobile-heavy channels inflate more. Desktop channels inflate less. You're comparing apples to slightly different apples.
How do you get a more honest number in BigQuery?
- Use ga_session_number = 1 to identify first sessions per device
- If user_id is available, deduplicate across devices before counting
- Use some tiny tweaks - e.g. user_first_touch_timestamp as the "first seen" anchor
Document the limitation clearly. "Device-new users" and "new users" are not the same thing, and your stakeholders should understand that before they make decisions.
Why does this affect budget decisions?
If new user share is inflated in mobile channels, you might overestimate how well those channels expand your audience. I've seen acquisition budgets built on new user rates that were off by 25%. The channel mix looked fine until someone asked where all those new users went after day one.
Want to get all my top Linkedin content? I regularly upload it to one Notion doc.
Go here to download it for FREE
