The GA4 Audit Playbook Part 1

Thorough and robust, Google Analytics 4 (GA4) audits have a long list of checks to perform, for example, my GA4 audit covers up to 80 health & configuration checks as well as structural checks and custom events testing tips and this ultimate GA4 audit playbook will cover all of this. Since there is a lot to cover, we will split this audit playbook into four parts to help us understand when we should perform some of these sub-audits and what to look for. These four parts are as follows; GA account, property, and datastream structure, GA4 configuration checks, GA4 data health review and GA4 custom events testing.

The next part will be released in one week. Give me a follow me on LinkedIn to keep updated as soon as this gets launched! Also, don’t forget to check out our audit products.

Account, Property, and Datastream Structure

First of all, it’s ideal to set up your account and property structure to best practice from the start, however, if you haven’t then there are ways to mitigate data impacts for analysis purposes (which can be seen here). You will likely only need to perform this sub-audit once unless you are a business that has seen large growth and is acquiring other businesses in your business portfolio. In that case, then refresh this sub-audit as and when you need to add more GA accounts and properties.

Account Structure

The best practice is one account per parent business. That’s it. Nice and simple.

Property Structure

The property structure can depend on two factors:

Key Definitions

To help understand why we suggest these structures, it’s important to understand the following GA4 terms:

Therefore, for non-360 customers in particular, property structures are very crucial as GA4 no longer has “views” like Universal Analytics (UA) had.

It is usually one brand per property. However, for global companies that operate in multiple markets it is best to have one GA4 property per market for non-360 customers (note: you could have all markets under one property and filter by country within this property but you could start seeing a lot of sampling and thresholding on your data which isn’t ideal) and for 360 customers there can be two outcomes:

If you are a non-360 customer and operate in multiple markets and decide to have separate properties per market, your only solution to get a “global” property is to export the GA4 data from these properties and merge them. Therefore, linking GA4 to BigQuery will help achieve this efficiently and effectively but there are other options available such as leveraging the GA4 API to a database or using the LookerStudio GA4 connector and merging data there.

Data Streams Structure

This is typically one website, and one app per GA4 property. There are some rare edge cases where there are two websites that cross over a lot during a user journey and therefore you can configure one web data stream for these two websites (with cross-domain tracking enabled!).

We will go through 3 scenarios to cover all the different property structures that will cover 95% of real use cases.

Scenario One

Let’s think of one company, one brand, and one market business such as mine, Know Analytics. Know Analytics has one website and hypothetically, one mobile app. Then the structure should be as follows:

Scenario Two

Let’s think of one company, one brand, but operating in multiple markets. Let’s call this company and its brand, Jon’s Cheese Cakes, and operates in both the UK and Italian markets. For each market, they have one website and one app. Then the structure should be as follows

Scenario Three

Imagine a parent company with two different brands. Both brands cover two markets (UK and Germany). The parent company is called “Sue’s Mega Company” and the two brands are called “Cakes are the Best” and “Where Are the Cutlery” (complimentary I know…). Each brand and market has one website and one app. Then the structure should be as follows:

Tips on Restructuring

The short answer is it depends on the impact on your data. Some scenarios and ways to help maintain historical data views below:

If it's due to unnecessarily separating the properties by, for example, web vs app instead of web & app and only if your web & app are complimentary of each other then move the web or app property that has the least impact on your business within the other property by setting up new data streams within. If you need to analyse the historical data of the property that you have paused (and moved across to the other) then you will need to export and merge both properties' data externally.
If it's due to excessively putting multiple markets and/or brands under one property then you will need to separate these markets and/or brands as suggested in scenarios 2 and 3. Note, if you are not experiencing any sampling and data thresholding whatsoever and can easily filter and analyse these brands and/or markets separately it might not be worth restructuring but it's important to understand that over time you might start seeing these data limits as more data is collected. Make sure to keep the main market and/or brand in the original property and move out the rest. If you need to analyse the historical data of the brands and/or markets you moved to new properties, follow the same suggestions given in the first tip.

Conclusion

Having a good Google Analytics structure is vital in understanding your brands, markets, websites and apps data clearly and concisely whilst ensuring there is no unnecessary “property bloating” which could impact data quality (more sampling and thresholding). If you are setting up GA4 for the first time, this is ideal as you can get this right from the start and mitigate any historical data view loss in the GA4 UI. Conversely, if you are restructuring your setup make sure you do so as soon as possible and follow the tips on this and how to view your historical data still as it is not all doom and gloom.

Follow me next week for part 2 “GA4 Configuration Checks”!