Building analytics that actually work

Let’s be honest: most analytics stacks don’t fail because teams don’t care. They fail because things get messy somewhere between a cookie banner, a tag manager, and a dashboard everyone’s quietly stopped trusting.

Regulations change. Platform behaviour shifts. The web stack gets more complicated. And before you know it, nobody’s quite sure what’s firing when, or whether the numbers really mean what they think they mean.

This article lays out a practical way to make GA4 and GTM work properly, without over-engineering, panic-reacting to headlines, or crossing your fingers and hoping for the best.

Start with intent, not implementation

Good analytics doesn’t start in GTM. It starts with agreeing what you’re actually trying to do. At a minimum, that means being clear on what’s allowed, what isn’t, and why...and then making sure your setup enforces those decisions automatically.

A few simple principles go a long way

• Consent should be established before anything analytics-related fires
• One GA4 configuration per property (more doesn’t mean better)
• Consent isn’t just a banner click. It’s a signal that needs to flow all the way through your data and activation stack

When teams write this down and bake it into standards and QA, things get calmer very quickly. Fewer debates, fewer escalations, and data people actually trust.

Google Signals, GTM, and the headline panic

Google Signals and that widely shared court ruling around GTM sparked a lot of anxiety. Signals can be useful. They can also blur lines with advertising features. The key isn’t the toggle, it’s the decision behind it. Decide deliberately, market by market, document the rationale, and make sure “allowed for analytics” doesn’t quietly turn into “used for ads”.

As for the ruling: it didn’t say GTM is illegal. It showed what happens when consent is bypassed. If nothing fires before consent and you can explain your setup clearly, GTM isn’t the problem. Most issues here aren’t legal edge cases, they’re implementation shortcuts.‍

Where things usually go wrong

The biggest analytics problems are rarely exciting. They’re usually small, avoidable things that compound over time. Some common examples:

• Events firing before consent or before the GA4 config is ready
• Multiple GA4 configs overwriting each other
• Enhanced measurement creating lots of noise but very little insight
• “Clever” GTM logic no one remembers adding
• Auto-captured data pulling in things it really shouldn’t

The fix isn’t flashy. It’s boring in the best way: clean data layers, clear event names, consent-aware triggers, and screenshots or debug logs saved with each release. When people stop questioning the numbers, progress gets much easier.

Server-side: powerful, but not magic

Server-side GTM can be a great move; but only if it’s done properly.

That means genuinely first-party setup, consent passed through every request, and clear rules about what gets forwarded where. Whether you go SaaS or self-hosted matters less than whether you can see what’s happening and roll back safely when something changes.

Treat it like production software, not a side project, and it earns its keep.

Where AI actually helps

Despite the hype, the most useful AI wins in analytics aren’t about writing copy or building dashboards. They’re about removing operational friction.

Think agents that check containers against standards, flag race conditions, or summarise what changed in a release. Pair that with templates and shared“policy packs”, and suddenly teams move faster without losing control. Humans still decide what’s acceptable. Automation just makes it easier to stick to those decisions.

The final takeaway

Analytics works when it’s intentional, documented, and enforced.

Consent isn’t a banner - it’s a promise. Governance isn’t red tape - it’s how you keep your data usable. And implementation isn’t something you “finish” - it’s something you maintain.

Get those right, and analytics stops being a source of doubt and starts doing what it should have been doing all along: helping you make better decisions and grow with confidence

‍