Google has confirmed that Universal Analytics will stop processing new data on 1 July 2023. That sounds like plenty of time, and that's the trap. The migration itself is a few days of work. The reason to do it now is that GA4 only reports on data it has collected — so every month you delay is a month of year-on-year comparison you will simply not have next summer.
The part people misunderstand
GA4 is not an upgrade to Universal Analytics. It's a different measurement model with a different data structure — events rather than sessions and pageviews. Your historical UA data does not migrate across. There is no import. If you set up GA4 in June 2023, then in September 2023 you will be unable to compare anything to September 2022, because GA4 has never seen September 2022. Run both in parallel starting now and that problem disappears.
What to do this month
Create the GA4 property and let it collect alongside UA. That's genuinely most of the value, and it takes an afternoon. Everything else — rebuilding conversions, remapping goals to events, sorting out consent — can happen over the following months while data quietly accumulates in the background. The mistake is treating this as a single project to be scheduled rather than a clock that's already running.
- Set up the GA4 property today and run it in parallel with UA
- Re-map your goals as events — they don't carry over and the model is different
- Export the UA history you care about before it becomes inaccessible
- Rebuild your core reports early, while you still have UA to check them against
Export your history properly
Google has been vague about how long historical UA data will remain accessible after processing stops. Assume it disappears. Decide now which reports actually inform decisions — probably channel performance, conversion paths and a handful of landing page views — and export them properly rather than as screenshots in a slide. A year from now, when someone asks how this quarter compares to two years ago, you'll want a real dataset.
The honest opportunity in this
Most analytics setups I inherit are a decade of accumulated sediment: goals nobody defined, filters nobody remembers configuring, three conflicting definitions of a conversion. A forced migration is the only time you'll ever get organisational permission to throw that away and rebuild deliberately. Take the opportunity to define, in writing, what actually counts as a conversion and why — that conversation is worth more than the tool.
The migration is easy. The lost year of comparable data is not, and it's the one part you can't fix retrospectively. Set up the parallel property this week.
Want to apply this to your business?
Book a free 30-minute strategy call and we'll diagnose your specific growth blockers together.
Book a Free Call