There was a moment around 2023 when a lot of us looked at GA4 and quietly wondered if Google had lost the plot. The Universal Analytics switchover was forced, abrupt, and the replacement felt like it had been designed by a committee who had never reported website performance to a client on a Monday morning.
The goodwill evaporated fast. I wrote about why people still disliked GA4 in 2025 and I meant it at the time. But opinions, like products, can evolve.
Having spent enough time in the weeds with it, here’s where I’ve landed: GA4 and Google Tag Manager, once you genuinely understand how they work together, are more powerful than what we had before. The AI-assisted features deserve more credit than they get. Anomaly detection, predictive audiences, automated insights. These aren’t gimmicks. The natural language query functionality has made exploratory analysis faster. Not perfect. Faster.
The case for Tag Manager (and why I'm a convert)
I didn’t always think this way. There was a time when dropping Google Analytics directly onto a site felt like the cleaner, simpler option. And in terms of raw data volume, it can still occasionally produce a larger dataset than a Tag Manager implementation. I’ve seen it, and I can’t fully explain it, but it’s worth knowing.
What changed my mind was a site rebrand.
We rebuilt the site. New domain, new structure, new everything. And because everything was already centralised in Tag Manager, GA4, Google Ads, the Meta pixel, consent logic, the tracking migration was a non-event. No hunting through theme files for hardcoded scripts. No hoping a developer had documented what was where. One container. Organised, labelled, version-controlled.
That’s when it clicked. Tag Manager isn’t just a convenience. It’s infrastructure. It satisfies something that I suspect other developers will recognise: the need to have everything in one place, properly categorised, with a clear audit trail. Everything has a trigger. Every trigger has a tag. Every tag has a purpose. That kind of organisational clarity isn’t just aesthetically pleasing. It’s genuinely useful when something breaks at 11pm and you need to diagnose it quickly.
Consent Mode is worth the effort
It looks complicated because it is. But the distinction between basic and advanced consent mode matters. Getting it wrong means either overcollecting, a GDPR problem, or undercollecting, a usefulness problem. Getting it right means your data retains meaningful integrity even when a user declines, because the modelling fills the gaps.
Once built correctly, consent state propagates through everything automatically. That initial investment pays back every time you add a new tag.
The honest take
Is it as quick to set up as the old Analytics snippet? No. Is the learning curve steeper than it should be? Probably. But the tools we have now, properly implemented and properly understood, are better than what we had.
The issue was never that GA4 was bad. It was that Google forced a migration before the product was ready, and the community never quite forgave them for it.
The product’s caught up. The reputation is taking longer.
