10 Things You Can Do with GA4 Event-Level Data

GA4 is by far the most widely-deployed web analytics tool, yet complaints about its standard reporting are commonplace – hard to customise, inconsistent numbers due to sampling, and a general sense that you can’t quite trust what you’re seeing.

This is where the raw GA4 export comes to the rescue.

Every time your GA4 tag fires, Google captures an event containing a huge amount of useful detail – timestamps, cookie IDs, geo data, browser info, page variables, and more.

That event-level data unlocks a new depth of analysis that restores your faith in web analytics.

Getting access to this raw data involves setting up a feed into Google BigQuery (something we can help with), but once it’s flowing, you can start to get stuck into making the data work a bit harder for you.

Here are ten valuable ways you can leverage the raw data to enhance your analytics.

1. Identify and hide the traffic you should be ignoring

Your data is full of visits you don’t want to include in your analysis – staff working from home, agencies, developer teams, bots, and crawlers. All of it skews your analysis.

The solution is to identify and exclude those cookie IDs from your reporting.

For internal & agency teams, you could create a hidden page (behind a login ideally) that fires a custom event (for example 'is_internal_traffic'). You can now pull a list of cookie IDs that have fire this event. From there, build an exclusions table of all these cookies and exclude them from your analysis.

You can even clean up historic data this way. It’s the simplest way to clean up months of noise in one go.

2. Bring back ‘Visits’

I don’t like Sessions. I don’t think I’m alone. They’re messy – if a visitor arrives from multiple sources in one session, which gets credit? For example someone starts with paid search, goes in search of an Affiliate discount code, and comes back. Or someone clicks an organic result and then returns via a paid click.

With the raw data, you can define your own visit logic using referral URLs, entry pages, and timestamps to create a cleaner, more intuitive way to understand which sources your traffic is actually coming from.

Session attribution is also one of the murkier aspects of the GA4 black box approach, but with the raw data you can see exactly where traffic is coming from.

3. Analyse user behaviour across visits

GA4’s event export contains a cookie ID - labelled as user_pseudo_id. This means you can link sessions together for the same user.

Now you can start to explore behaviour in more detail:

• How many visits occur before a purchase?
• How long is the average time-to-purchase?
• How much time elapses between visits?

When you roll up these behaviours, you start to see real research patterns – how people actually move towards conversion, not just how they behave in one isolated session.

4. Visit point scoring

Not all pages are equal. Watching a demo video for five minutes is worth more than glancing at your homepage. The raw aw data gives you the freedom to assign value to any and all events in a user's visit to your site. You can create a lookup table of pages and their relative important and assign a score according to importance. You could flag which pages are negative ('cancel my account', 'how to complain' etc). Do something similar for events on your site ('watch demo', 'learn more').

Creating a scoring system that reflects engagement (time spent, depth, intent) and roll it up by device or traffic source. You’ll quickly see which marketing activity is driving the most valuable interactions. (I covered this approach in more detail here).

5. Build a custom attribution model

Once you’ve scored your visits, you can start to chain them together into paths using user_pseudo_id as the common key.

Combine your point scoring with visit-based source logic, and you have the foundational data to build a fully custom attribution model – one that reflects how your customers actually convert, not how GA4’s black box thinks they do.

There’s a whole topic here in how to carry out multi-touch attribution - e.g Markov chains, Shapley value model - and specialist tools to tackle the analysis, but having robust, granular and meaningful data is the foundation of truly transparent, data-driven measurement.

6. Link multiple devices to a single customer record

The next step up from visit-based attribution is cross-device linking. Without it, a single person can look like five. Maybe someone researches for weeks on their laptop, then finally converts on mobile from a brand search – and all that upper-funnel work is invisible.

By linking anonymous identifiers (for example CRM IDs) with cookie IDs, you can start uniting journeys across devices into one coherent view of the customer lifecycle.

7. Measure TV impact

Running TV or VOD campaigns? Timestamped sessions let you measure uplifts in traffic to the second.

No models, no assumptions – just clear spikes when your ads air. Finally, a way to measure the online impact of offline media.

8. Link online to offline outcomes

When someone fills out an enquiry form, capture a CRM reference as a custom event alongside their cookie ID.

Later, join that CRM table back to your event data and see which online sources drove actual business results. It’s the missing link between digital campaigns and real-world conversions.

9. Conduct detailed basket analysis

How much of your ‘Add to Basket’ activity represents genuine intent versus comparison behaviour?

• If one user adds ten products, are they serious or just browsing?
• Which products are users adding to the basket together?
• Are their new product combinations that should be promoted or recommended?

The cookie-level data help you find out – mapping which products were added, in what order, and whether they were ever purchased. This can reveal whether 'checkout drop-off' is actually a UX issue or just people window-shopping.

10. Define custom segments with zero sampling

Define your own unsampled segments using any combination of conditions – for example:

• Landed on page X
• Added category Y to basket
• Didn’t buy

Push those cookie IDs into a dataset and track what happens next. Do they return? Convert later? Drop off completely? GA4’s UI won’t show you this. The raw data will.

Taming the data

The first time you open your raw GA4 export in BigQuery, it’s not pretty. It’s dense, verbose, and deeply unintuitive.

But once you’ve got the key fields mapped and the data model under control, the rewards are huge:

• You can clean and clarify your reporting, removing rogue data you don't want
• You gain a much deeper understanding of user and purchase behaviour
• You unlock insights that can drive action and tangible improvements

You stop working within Google’s constraints and start defining your own measurement logic.

There’s no silver bullet – the learning curve is pretty steep – but once you’re over the hump, you’ll uncover insights that drive both performance and confidence.

If you’d like to dig deeper into the possibilities of the raw GA4 and how it could help solve some of your business challenges and questions, drop me a message. I'm always happy to have a chat about anything from understanding the quirks of the data model to helping crafting some custom SQL to answer the questions you need to answer.

Coming Next Week

Next week, we’ll be looking at how to report on domain-level data for your programmatic activity across platforms, including DV360, Amazon Ads - crucial for spotting rogue domains but also valuable for creating your own categorisation of the content your advertising is appearing alongside. 

🔔 Hit subscribe now so you don’t miss it.

Thanks for tuning in,

Ed Campbell

Founder, Bright Analytics

P.S. Want to learn more about Bright Analytics? Book a demo