GA4 Setup Guide: The Complete 2026 Configuration Playbook

GA4 setup is where most analytics implementations either pay back or quietly fail. Universal Analytics taught a generation of marketers to expect data that "just works". GA4 does not. It requires deliberate setup at every layer — properties, streams, events, conversions, audiences — and most teams skip the configuration steps that matter most.

This guide is the GA4 setup playbook we apply on every client account. It covers the property structure decisions, data stream configuration, event setup, conversion definitions, audience building, and the technical choices that determine whether your reports tell the truth or mislead the team.

The work is technical but not deep. A correct GA4 setup takes 4 to 8 hours for a typical SME. Done right, it produces data you can trust for years.

Why GA4 setup matters more than Universal Analytics did

Universal Analytics shipped with reasonable defaults. Pageviews tracked automatically, events tracked with minimal configuration, ecommerce worked out of the box for most stacks.

GA4 ships with a different philosophy. Many things that "just worked" in UA require explicit setup in GA4:

• Custom events need explicit configuration
• Conversions must be defined manually
• Audiences require deliberate construction
• Cross-domain tracking is not automatic
• Enhanced measurement settings need review
• Data retention defaults are short

Most teams installing GA4 stop at "the snippet is on the site" and miss 70 percent of the configuration. The result is reports that look fine but cannot answer real business questions.

We covered the broader conversion tracking foundation in our Google Ads conversion tracking setup guide. GA4 is the analytics layer underneath that foundation.

The GA4 setup hierarchy

Setup happens at four layers. Each one builds on the previous.

Layer 1 — Property structure

How many properties you create and how they relate. Decided once, hard to change later.

Layer 2 — Data streams

The actual connection between your site (or app) and the GA4 property.

Layer 3 — Events and conversions

What user actions you track and which ones count as conversions.

Layer 4 — Audiences and reporting

Segments, audiences, custom reports that turn raw data into insights.

Work through each layer in order. Out-of-order setup creates rework.

Section 1 — Property structure

The first decision and the hardest to change.

One property per business, not per region

Common mistake: creating separate GA4 properties for example.com, example.fr, example.de. This fragments data and prevents cross-region analysis.

The right pattern: one GA4 property with multiple data streams (one per region/domain). Data is unified, you can still filter by hostname.

Multiple properties only when

• The brands are truly separate (different companies, different products)
• Legal or compliance requirements demand isolation
• Different teams need exclusive access

For most SMEs, a single property is correct.

Internal vs external properties

Some agencies set up a separate GA4 property for internal staging or development. This works but is rarely necessary. Use a single property with filtered "internal traffic" exclusions instead.

Property settings to set immediately

After creating the property:

• Set the time zone (matters for daily aggregation)
• Set the currency (for revenue reporting)
• Set the industry category (informs benchmarks)
• Set the business size (informs feature rollouts)

These take 60 seconds. Skipping them causes incorrect reporting later.

Section 2 — Data streams

The bridge between your site and GA4.

One stream per domain

For example.com and example.fr, create one stream each. They report into the same property.

Enable Enhanced Measurement

Enhanced Measurement automatically tracks:

• Page views
• Scrolls (90 percent depth)
• Outbound clicks
• Site search
• Video engagement
• File downloads

Enable all of these unless you have specific reason not to. They cost nothing and provide useful baseline data.

Configure Cross-Domain Tracking

If your site spans multiple domains (e.g., example.com and checkout.example.com), configure cross-domain tracking in stream settings.

Without this, a user who moves between domains looks like two separate users. Cross-domain tracking unifies them.

Set up Referral Exclusions

Exclude domains that should not count as referral traffic — typically your own payment processor (stripe.com), your own subdomains, and partner domains where the user has already started a journey on your site.

Common offenders: paypal.com, stripe.com, klarna.com. Without exclusion, returning users from these payment partners look like new referral traffic.

Enable Google Signals (carefully)

Google Signals enables cross-device tracking and demographic reports. Useful but:

• Requires user consent in EU/UK contexts
• Triggers data thresholding (small audiences are hidden for privacy)
• Has implications for some reporting

For most SMEs, enable Google Signals with appropriate consent management. The benefits outweigh the costs.

Section 3 — Event tracking

GA4's event-based model is more flexible than UA's session/pageview model — but requires more setup.

Understand the four event types

Automatically collected: page_view, session_start, first_visit. Always collected.

Enhanced Measurement: scroll, click, video_start, etc. Enabled in stream settings.

Recommended events: standardised events with reserved names — purchase, sign_up, add_to_cart, generate_lead. Use these names for compatibility.

Custom events: anything specific to your site that doesn't fit the recommended list.

Use recommended event names

For common actions, use Google's recommended event names instead of inventing your own:

• purchase not order_complete
• sign_up not account_create
• generate_lead not form_submit
• view_item not product_view
• add_to_cart not cart_add

Recommended events:

• Appear in standard reports without configuration
• Map cleanly to Google Ads conversion imports
• Make AI insights smarter
• Make the data understandable to teams without context

Required vs custom parameters

Each event has required and recommended parameters. For purchase:

• transaction_id (required for deduplication)
• value (required for revenue reporting)
• currency (required for multi-currency)
• items (recommended array of products)

Without these parameters, the event fires but doesn't populate revenue reports correctly.

Send events through GTM, not directly

For most setups, fire events through Google Tag Manager rather than calling gtag() directly in code.

Why:

• Marketing team can adjust events without developer
• Easier to debug
• Easier to add or remove events
• Cleaner separation of concerns

We covered the GTM setup in our Google Tag Manager best practices guide.

Don't create too many custom events

A typical SME site needs 8 to 15 events total. Most of those are recommended events plus a few custom ones for site-specific actions.

Sites with 30+ custom events usually have tracking creep — events added over time that nobody reviews. The data becomes unmanageable.

Section 4 — Conversions

Conversions are the events that count as business outcomes.

Define which events are conversions

In the Conversions section, mark events as conversions:

• purchase (for e-commerce)
• generate_lead (for lead-gen)
• sign_up (for SaaS)
• book_appointment (for service businesses)

Plus a couple of micro-conversions:

• add_to_cart (e-commerce funnel)
• start_checkout (e-commerce funnel)
• view_pricing (SaaS funnel)

Mark only events that genuinely represent business outcomes. Marking everything as a conversion makes the data unhelpful.

Conversion deduplication

For events like purchase, deduplicate using transaction_id. Without this, refresh-clicks on the thank-you page cause inflated conversion counts.

Linking conversions to Google Ads

In Admin → Google Ads Links, import GA4 conversions into Google Ads. This is the bridge that lets Smart Bidding optimise on your actual business outcomes.

We documented the full Google Ads integration in our Google Ads conversion tracking setup guide.

Section 5 — Audiences

Audiences segment users for analysis and remarketing.

Build the standard audience set

Every account needs these audiences:

• All users: default, captures everything
• Purchasers: completed a purchase
• Leads: completed a lead-gen conversion
• High-value users: top 20 percent by revenue or engagement
• Cart abandoners: started checkout but did not purchase
• Site visitors (no conversion): for retargeting cold traffic
• Returning customers: purchased more than once

Each audience powers a use case — reporting, retargeting, or analysis.

Build audiences with intent

Most audiences should have a clear purpose:

• Reporting comparison ("how do purchasers behave vs non-purchasers")
• Retargeting in Google Ads or Meta
• Filtering reports for specific segments

Audiences without a use case are clutter. Build the minimum set, expand as needs arise.

Set audience triggers carefully

Audiences support triggers that fire conversions when a user enters or exits the audience. Use sparingly — they can create double-counting in conversion data.

Section 6 — Custom dimensions and metrics

For data that does not fit standard reports, custom dimensions extend GA4's flexibility.

Common useful custom dimensions

• user_role (logged-in user type)
• content_category (article topic, product category)
• lead_quality (cold, warm, hot)
• plan_tier (free, pro, enterprise)
• traffic_source_detail (more granular than default source/medium)

Setting them up

1. Send the dimension as an event parameter
2. Register the parameter as a custom dimension in Admin
3. Wait 24 to 48 hours for data to populate

Custom dimensions cannot be backfilled. Set them up before you need the data, not after.

Watch dimension limits

GA4 has limits:

• 50 custom dimensions per property
• 50 custom metrics per property

For most SMEs, you'll use 5 to 15. Plenty of headroom.

Section 7 — Data retention and BigQuery

GA4's default data retention is short. Configure for your needs.

Default 2-month retention

In Admin → Data Settings, change retention from 2 months to 14 months for user-level data.

Without this change, year-over-year comparisons break after 60 days. The 14-month setting is GA4's maximum for the free tier.

BigQuery export for long-term storage

For unlimited retention and custom analysis, enable BigQuery export (free for most SMEs).

Benefits:

• Permanent data retention
• Custom SQL analysis
• Integration with other data sources
• Faster reports than GA4 UI

Setup: Admin → BigQuery Links → Link. Choose a Google Cloud project. Done.

The first 10 GB of storage and 1 TB of queries per month are free. Most SMEs stay within free tier.

Section 8 — Filters and internal traffic

Exclude internal traffic so it doesn't contaminate reports.

Internal traffic filter

In Admin → Data Settings → Data Filters → Define Internal Traffic. Set IP addresses for your office and team's home IPs.

Then enable the "Internal Traffic" filter to exclude this traffic from reporting.

Developer traffic filter

For dev and staging environments, set a filter that excludes traffic from those hostnames. Avoids polluting production data.

Bot traffic

GA4 has built-in bot filtering that catches most known bots. For sophisticated bot traffic, additional filters can be configured at the GTM layer.

Section 9 — Consent mode and privacy

For EU/UK and increasingly other regions, consent is required before tracking.

Implement Consent Mode v2

Consent Mode v2 lets GA4 collect modelled data even when users decline consent. The data is not user-level but provides aggregate trends.

Setup involves:

• Cookie consent banner that signals user choice to GTM
• GTM consent settings configured
• GA4 tags inheriting consent state

The full setup is platform-dependent. For most SMEs, plugins like Cookiebot, OneTrust, or Iubenda handle the integration.

Default consent state

Set default consent to "denied" for EU traffic. Users must explicitly accept before tracking fires.

Default "granted" is acceptable for non-EU traffic depending on your local privacy law.

A 14-day GA4 setup plan

If you are setting up GA4 for the first time or auditing an existing setup, follow this sequence.

Days 1 to 2 — Property and streams. Create property, configure streams, enable enhanced measurement, set referral exclusions.

Days 3 to 5 — Events. Implement recommended events via GTM. Configure event parameters. Verify in DebugView.

Days 6 to 7 — Conversions. Mark business-outcome events as conversions. Set up Google Ads conversion import.

Days 8 to 9 — Audiences. Build the standard audience set. Configure remarketing audiences.

Days 10 to 11 — Custom dimensions. Identify dimensions needed for reporting. Set them up.

Days 12 to 13 — Data retention, BigQuery, internal filters. Configure for long-term data hygiene.

Day 14 — Validation. Walk through every report. Verify data matches expectations.

Most accounts complete this setup in 14 days. The result is a foundation that supports the next 2 to 3 years of analytics work.

A real example — Marseille cosmetics GA4 audit

We took over a Marseille cosmetics e-commerce GA4 account with these issues: default 2-month retention, no enhanced ecommerce events, no audiences, no Google Ads link, no custom dimensions, internal traffic from the office polluting reports.

After 8 days of setup work — 14-month retention enabled, BigQuery export configured, ecommerce events properly implemented, 8 audiences built, Google Ads linked, internal filter applied — reports became usable. Smart Bidding on Google Ads improved 31 percent over the following 60 days using the cleaner data. The full story is in our Marseille cosmetics case study.

Common GA4 setup mistakes

These are the patterns we see most often.

Multiple properties when one would do. Fragments data, prevents cross-region analysis.

Default 2-month retention. Breaks year-over-year reports.

Custom event names instead of recommended. Misses standard reports and AI insights.

No referral exclusions for payment processors. Inflates referral traffic numbers.

No Google Ads link. Smart Bidding cannot use GA4 conversions.

No internal traffic filter. Office and dev traffic pollutes reports.

Marking everything as a conversion. Makes conversion data unhelpful.

No BigQuery export. Loses long-term data when 14-month retention expires.

Frequently asked questions

Do I need to use Universal Analytics if I still have data there?

UA stopped processing data in July 2024. Existing UA data is read-only until eventual deletion. Migrate fully to GA4 and archive UA data via BigQuery export if you need it long-term.

Can I track multiple websites in one GA4 property?

Yes, via multiple data streams. This is the recommended pattern for sites with multiple regional domains or subdomains.

Should I use GA4 if I have a small site?

Yes. Even with low traffic, GA4 provides useful traffic source data, content performance metrics, and conversion tracking. The setup investment pays back at any scale.

Is GA4 free?

The standard property is free. Free tier covers most SMEs. Enterprise sites with very high event volumes hit free-tier limits and may need GA4 360 ($150K/year+).

Why do my GA4 reports differ from my old UA reports?

Different measurement models. GA4 uses event-based tracking with different default settings (engagement-based sessions, no bounce rate, different attribution). Some differences are expected.

Should I use BigQuery for analysis?

For SMEs with active analytics work, yes. The BigQuery free tier covers most analysis needs. The flexibility of SQL beats the GA4 UI for complex questions.

Get a GA4 audit

We audit GA4 setups free of charge. Within 48 hours we deliver a per-section breakdown of configuration gaps and a prioritised setup roadmap.

Book a free 30-minute audit. We screen-share, walk through your GA4 property, and you leave with a clear action plan.

Or explore our Google Ads service for the full system we run on accounts that need integrated paid media and analytics.