Why is there discrepancies between GA4 vs UA (GA3)?

If you have encountered this wonderful help center article[UA→GA4] Possible reasons for conversion differences in GA4 vs. UA , chances are you still very confused.

This is very understandable, the help center article aren’t for end user to interpret but more for partner like us to study the platform limit.

How should I expect the discrepancies in GA4?

Forget about it, It’s too troublesome to use

As a new product, it’s quite common to observe discrepancies in the reports for GA4. Many clients are now complaining daily about significant differences between GA4 and the standard GA metrics. I also believe this is an area where the product needs improvement.

However, getting to the point, people often assume that if there are discrepancies in GA4 data, it means the tool itself is flawed and unusable. This cause-and-effect relationship is actually unhealthy. As a website owner, the best approach is to ask yourself the following three questions:

1. Why are these discrepancies occurring?
2. How do these discrepancies affect me?
3. Why should I care about these discrepancies?

The sequence in which these three questions are answered is crucial and will influence our ultimate decisions.

Why are we getting discrepancies in GA4 report？

There’s two important concepts to grasp.

The types of discrepancies

When discussing discrepancies in GA4, they can be categorized into three main types:

1. Differences between GA4 and the Universal Analytics
2. Internal discrepancies within GA4
3. Discrepancies between GA4 and Google Ads

Often, people inadvertently lump these three distinct types of discrepancies together, which is incorrect and doesn’t provide helpful insights for decision-making.

Factors causing discrepancies

When it comes to discrepancies, the first thing that comes to mind is why these discrepancies occur.

I categorize discrepancies into seven main types1:

1. Discrepancies caused by data collection
2. Discrepancies caused by data sources
3. Discrepancies caused by algorithm
4. Discrepancies caused by differences in metric definitions
5. Discrepancies caused by data sampling
6. Discrepancies caused by data thresholds
7. Discrepancies caused by cardinality

These seven reasons are distinct, meaning they can occur independently or simultaneously with other underlying causes.

What should I do when there is discrepancies?

Hold on a sceond, aren’t you going to explain why?

“Why” is indeed important, but I’ll address it later, as we’ve already tackled the first question of why discrepancies occur. The focus now is on the significant impact these discrepancies actually have on us and how to address them.

Drawing from my industry experience, I’ve put together the following table for reference:

[Note: Unfortunately, I can’t display the table visually in this text-based format. However, you can describe the content of the table, and I can help you interpret or rephrase it if needed.]

In the table, items highlighted in red indicate cases of higher impact that should be prioritized, while items in orange are context-dependent.

Feel free to describe the contents of the table, and I can assist you further with understanding or rephrasing the information.

Types of Discrepancies	Factors		Impact	Frequency	Alternatives	Cost of Alternatives	Prioirty
Differences between GA4 and the Universal Analytics	Data collection differences		Big	Low	NA	small	P0, align event trigger logic
	algorithm		small, discrepancies <5%	meidum	Use BigQuery for reporting	medium	Discrepancies is relatively small, can be ignored
					Adjusting reporting dimensions	Low to medium
	Metric definitions		Big	medium	Custom metrics	Low to medium	depending on requirement, can be ignored.
	Data sampling		medium	small	Use BigQuery for reporting	medium	Depending how often the reports are used
	Data thresholds		small	High	Update report identity	Low	Retroactive, can be changed anytime
	Cardinality		Big	medium to High	Use BigQuery for reporting	medium	Depending how often the reports are used
Internal discrepancies within GA4	Data sources difference		medium to high	High	Use BigQuery for reporting	medium	Depending how often the reports are used
					Use the report under right context
	Algorithm		medium	medium	Use BigQuery for reporting	medium	Depending how often the reports are used
	data sampling		medium to High	small to medium	Use BigQuery for reporting	medium	Depending how often the reports are used
	data thresholds		small to medium	small	Update report identity	Low	Retroactive, can be changed anytime
	cardinality		small to medium	small to medium	Use BigQuery for reporting	Low	Recommend to ignore
					Adjusting reporting dimensions		Use BigQuery for one -time report
Discrepancies between GA4 and Google Ads	Data collection differences		Big	Low	NA	small	P0, align event trigger logic
	Algorithm		small to medium	medium	Use GA4 conversion instead of Goolge Ads covnersion	small	If no specific reason (e.g. enhanced conversion), recommend to use GA4 conversion

After reviewing this report, we can understand when we need to care about these discrepancies as follows:

Start by Identifying the Type of Discrepancy

Discrepancies between GA4 and Google Ads

Congratulations! This discrepancy has a significant impact on you, but the solution is also the most direct and has the lowest cost.

Discrepancies between GA4 and Standard GA

Luck is on your side! Begin by comparing the total count of events. Identify which specific event tracking might be causing the issue. If there’s no difference in event counts, then you need to investigate further to uncover the root cause of the discrepancy.

Internal Discrepancies within GA4

Congratulations! You’ve entered the realm of exploring GA4 intricacies. You’ll need to invest time in understanding how different reports should be used and… get comfortable with SQL. For a deeper understanding of report use cases and SQL, keep an eye out for my upcoming articles.

Conclusion

This exploration of GA4 report discrepancies, along with some directions and strategies for addressing them, hasn’t provided an immediate replacement solution. However, in the realm of Analytics, understanding, categorizing, summarizing, and setting priorities are far more crucial than simply executing alternative plans. Often, I observe that many clients allocate their resources to lower-priority items, and the loss of human resources is an easily overlooked consequence.

I hope the above insights are helpful to you!