Many of you have implemented GA4 and check its reports daily. Yet with the default settings, hidden distortions can remain in GA4 data: approximations caused by sampling, internal visits mixed into the data, and sessions split across domains. Making decisions without addressing these issues risks steering your marketing efforts in the wrong direction.
Source: Google
Based on the latest information as of April 2026, this article explains five important settings that affect GA4 data accuracy, with specific numerical thresholds, setup steps, and verification methods.
5 settings to improve GA4 data accuracy
Improving GA4 data accuracy requires properly implementing five areas: measures to address sampling, internal traffic exclusion, cross-domain measurement, understanding engagement metrics, and using exploration reports.
Setting | Risk if left unaddressed | Impact |
|---|---|---|
Data sampling pitfalls and workarounds | Report figures become estimates on large sites, preventing accurate analysis | High |
Internal IP address exclusion | Internal visits enter the data and distort the picture of user behavior | High |
Cross-domain measurement | Sessions split when users move between domains, obscuring traffic sources and conversion paths | Medium to high |
Understanding engagement rate correctly | Confusing it with UA bounce rate leads to incorrect page evaluations | Medium |
Using exploration reports | You miss insights that standard reports alone cannot reveal | Medium |
Data sampling pitfalls and workarounds
Data sampling in GA4 is a process that selects only part of a dataset when generating a report and uses that sample to estimate the whole. As a result, the numbers shown in reports are "estimates," reducing accuracy.
When sampling occurs
Sampling occurs mainly in exploration reports and is triggered when the number of events in the analysis period exceeds 10 million.
Condition | Details |
|---|---|
Where it occurs | Exploration reports (not standard reports) |
Event count that triggers sampling | When the number of events in the analysis period exceeds 10 million events |
How to check | A green check mark (no sampling) or yellow shield icon (sampling applied) appears at the top of the exploration report |
For GA4 360 | The threshold increases to 1 billion events |
Sampling risk by site size
Site size (monthly event count) | Risk of sampling | Impact |
|---|---|---|
Fewer than 1 million | Almost never occurs | Low |
1–5 million | May occur when analyzing longer periods | Medium |
5–10 million | May occur in monthly reports | High |
10 million or more | Highly likely | Very high |
4 ways to avoid sampling
Method 1: Shorten the analysis period (immediate effect)
Check the date range in your exploration report
If the sampling icon appears, cut the period in half
Adjust the period until sampling no longer occurs
Combine data from multiple periods in a spreadsheet to see the full picture
Method 2: Reduce the number of dimensions
Check the dimensions used in your exploration report
Keep only the dimensions essential to your analysis
Avoid analyses with four or more dimensions, such as "Page path x Source x Device x Country"
Instead, run several analyses using two dimensions at a time
Method 3: Use BigQuery export
BigQuery export is available even in the free version of GA4. It is the most fundamental solution.
Item | Details |
|---|---|
Where to set it up | GA4 Admin > Property settings > BigQuery links |
Export frequency | Daily export (once a day) or streaming export (real time) |
Free version limits | Daily export only, up to 1 million events per day |
GA4 360 limits | Supports streaming and billions of events |
Estimated costs | Usage-based charges for BigQuery storage and queries. Check the official Google Cloud website for current pricing. |
Related reading
How to use Looker Studio: Working with a free data visualization tool
Connecting BigQuery and Looker Studio lets you create reports using unsampled data.
Excluding internal IP addresses
When visits from employees or partners are included in GA4 data, they can inflate session counts, distort engagement rates, and contaminate conversion paths.
Examples of distortions caused by not excluding internal traffic
Type of distortion | Example |
|---|---|
Inflated session counts | Makes the user count appear higher than it actually is |
Distorted engagement rates | Visits from internal testing or development lower or raise the average |
Contaminated conversion paths | Conversions generated by internal testing distort funnel analysis |
Errors in page-level data | Visits to admin screens or internal-only pages are included |
Setup step 1: Define internal traffic
Open GA4 Admin
Go to Data streams > click the relevant web stream
In the "Google tag" section > click "Configure tag settings"
Click "Define internal traffic"
Click "Create"
Enter the following
Field | Example entry | Description |
|---|---|---|
Rule title | Head office | Choose a name that is easy to identify |
traffic_type value | internal (default) | No change needed |
IP address match type | "IP address equals" or "IP address is in range (CIDR notation)" | Choose based on your office's IP configuration |
IP address value | 203.0.113.50 or 203.0.113.0/24 | Enter your company's public IP address |
Tip: If you do not know your company's public IP address, search for "what is my ip" in your browser.
Setup step 2: Enable the data filter
Important: Step 1 alone does not exclude internal traffic. Be sure to complete step 2.
Open GA4 Admin > "Data collection and modification" > "Data filters"
Click the "Internal Traffic" filter
First, verify that it works in the "Testing" state
Check the Realtime report
Once verified, change the filter state to "Enabled"
Related reading
Ad operations strategy and performance improvement: Practical methods by budget size
Accurate GA4 data is an important foundation for measuring advertising performance. Improving data accuracy enables sound advertising investment decisions.
Cross-domain measurement
If your site structure involves users moving across multiple domains, sessions will split without cross-domain measurement, preventing you from accurately tracking traffic sources and conversion paths.
When cross-domain measurement is needed
Scenario | Example domain structure | Problem without setup |
|---|---|---|
Main site + Ecommerce site | example.com to shop.example.net | Moving to the ecommerce site creates a new session and replaces the original traffic source with "referral" |
Corporate site + Application form | example.com to form.example.co.jp | The conversion path to the form is broken |
Blog + Service site | blog.example.com to example.com | All traffic from the blog is measured as "referral" |
Note: GA4 treats subdomains such as www.example.com and blog.example.com as the same property by default, so cross-domain setup is unnecessary between them.
How to set up cross-domain measurement in GA4
Open GA4 Admin
Go to Data streams > click the relevant web stream
In the "Google tag" section > click "Configure tag settings"
Click "Configure your domains"
Click "Add condition" and add the relevant domains
After setup, click a cross-domain link and confirm that the destination URL includes the _gl parameter.
Understanding engagement rate correctly
GA4 fundamentally changed the definition of "bounce rate" used in UA (Universal Analytics). If you do not understand this change correctly, you can seriously misjudge page performance.
Definition of an engaged session
GA4 defines an "engaged session" as a session that meets any one of the following three conditions.
Condition | Details | Example |
|---|---|---|
Lasts 10 seconds or longer | The session duration is at least 10 seconds | A user reads an article for at least 10 seconds |
A conversion event occurs | An event marked as a key event occurs | A form submission, completed purchase, etc. |
Two or more page views | Two or more page_view events occur | A user moves from the home page to a service page |
Engagement rate benchmarks by page type
Page type | Engagement rate benchmark | Notes |
|---|---|---|
Blog posts and reading content | 50–65% | Relatively high because long articles tend to keep users for at least 10 seconds |
Landing pages (ad traffic) | 40–55% | Varies widely depending on how well the page matches visitor intent |
Ecommerce product pages | 55–70% | Users tend to view multiple pages to compare products |
SaaS and service overview pages | 50–65% | CTA clicks contribute to engagement |
Note: The figures above are reference values based on our own analysis data (in-house research).
Using exploration reports
GA4's standard reports are suitable for routine metric checks, but exploration reports are needed for deeper analysis and finding insights. It is important to understand the characteristics of each and use them appropriately.
When to use standard reports and exploration reports
Comparison | Standard reports | Exploration reports |
|---|---|---|
Purpose | Ongoing monitoring (routine metric checks) | In-depth analysis and insight discovery |
Customization | Low (fixed layout) | High (combine dimensions and metrics freely) |
Sampling | Does not occur | May occur above 10 million events |
Suitable use cases | Daily and weekly KPI checks, reporting to management | Root cause analysis, hypothesis testing, and detailed user behavior analysis |
Main exploration report templates
Free-form reports
Combine dimensions and metrics freely for custom cross-tabulation analysis. Use these reports for analyses such as channel x device, performance by landing page, and day of week x time of day.
Funnel Analysis
Visualize drop-off rates at each step leading to conversion. This is effective for analyzing ecommerce purchase funnels, B2B lead generation, and SaaS sign-ups.
Path exploration
Visualize users' page navigation patterns in a treemap to discover the main behavior paths leading to conversions and unexpected routes.
For separate instructions on internal visits and unclassified traffic sources, two common causes of reduced data accuracy, see How to exclude internal IP addresses in GA4 and Why unassigned appears in GA4 and how to fix it.
For how to automatically incorporate more accurate data into advertising reports, see Advertising report automation tools compared: Pricing, supported platforms, and how to choose.
FAQ
Is BigQuery export available in the free version of GA4?
Yes, BigQuery export is available in the free version of GA4. However, it is limited to daily exports and up to 1 million events per day. Streaming exports and higher limits require CPA optimization for GA4 360.
Does internal traffic exclusion apply to historical data?
No. Internal traffic exclusion filters apply only to future data. Historical data from before the filter was enabled does not change. Exclusion starts with new data 24–36 hours after setup.
Do I need to include subdomains in cross-domain settings?
No. GA4 treats subdomains such as www.example.com and blog.example.com as the same property by default, so cross-domain setup is unnecessary. It is needed only for transitions between different root domains.
How can I improve a low engagement rate?
If your engagement rate is low, consider improving page load speed, content quality, navigation, and CTA buttons. Designing content that encourages users to stay for at least 10 seconds is especially important.
How do I change the data retention period for exploration reports?
Go to GA4 Admin > Data collection and modification > Data retention to change the retention period to 14 months. We recommend changing it to 14 months because the default of 2 months makes long-term analysis difficult. The change applies only to future data.
Summary
Improving GA4 data accuracy requires properly implementing five areas: measures to address sampling, internal traffic exclusion, cross-domain measurement, understanding engagement metrics, and using exploration reports.
Setting up BigQuery export and excluding internal IP addresses are particularly important. These measures allow you to make decisions based on accurate data rather than estimates. It is also important to understand how the definitions of UA bounce rate and GA4 engagement rate differ, and to evaluate performance using appropriate metrics.
Improving data accuracy makes it possible to measure the performance of online customer acquisition efforts more accurately and focus your budget on those with high ROI. If you are bringing ad operations in-house, building a foundation for accurate data analysis will be key to success.
If your business is considering automating ad operations and making them more efficient, also consider Cascade's AI agent service for ad operations automation (cascaded.ai). Connecting it with accurate GA4 data enables more precise ad optimization.
