How to Set Up GA4 Correctly (And What Most Agencies Get Wrong)

Most websites have GA4 installed. Very few have GA4 correctly configured. The difference is significant: a correctly installed GA4 collects data. A correctly configured GA4 collects data that you can trust, segmented by the dimensions that matter, with conversion events defined against the outcomes your business actually cares about.

The problem with a technically-installed-but-incorrectly-configured GA4 is that it produces data that looks credible but isn’t. Traffic numbers. Session counts. A list of top pages. None of this is wrong, exactly — it’s just not connected to anything actionable. The conversion data is missing or wrong. The channel attribution is misconfigured. Internal traffic is inflating the numbers. And the team is making marketing decisions against a dataset they shouldn’t trust.

This post covers what most agencies get wrong in GA4 setup, what a correct configuration actually includes, and how to audit what you have.

What Most Agencies Get Wrong

The most common GA4 failure mode is the default install: drop the tag on the site, configure no additional events, define no conversions, and call it done. This produces a GA4 property that collects page views and sessions but nothing that connects to business outcomes. The agency can report on traffic. They can’t report on what the traffic produced.

The second common failure is no conversion events. GA4 tracks page views, sessions, and a handful of automatically collected events by default. It doesn’t automatically know what a conversion is for your business — that requires explicit configuration. A form submission without a corresponding GA4 event isn’t tracked. A phone number click that isn’t configured as an event is invisible. Without conversion events, GA4 is a traffic counter, not a marketing measurement tool.

The third common failure is no internal traffic filter. When employees visit the company’s own website, those sessions inflate traffic numbers and distort behavioral data. GA4 allows internal IP addresses to be filtered so that internal traffic doesn’t count in the reports. Most default installations don’t include this filter — and some percentage of every metric is the company looking at its own website.

The fourth common failure is incorrect channel grouping. When UTM parameters are inconsistently applied to campaigns, or when medium and source values don’t match GA4’s channel grouping rules, sessions get misattributed. Traffic from a paid campaign shows as direct. Organic search gets mixed with branded paid. The channel-level data becomes unreliable, and the optimization decisions made from it follow.

The fifth common failure is the default data retention setting. GA4’s default is two months of event data. After two months, the data is deleted and historical analysis beyond that window is impossible. Changing retention to the maximum is a two-click configuration change — but it doesn’t backfill data that’s already been deleted. Implementations that don’t address this on day one permanently lose historical data.

What a Correct GA4 Setup Actually Includes

Conversion event mapping

Before any technical implementation, the conversion events need to be defined. What user actions represent a meaningful business outcome? Form submissions, phone clicks, appointment requests, file downloads, chat initiations — each needs to be named, mapped to the point in the user journey where it fires, and confirmed as a conversion in GA4’s configuration. This is a business conversation, not a developer conversation. It requires input from the marketing team about what the site is supposed to produce.

GTM configuration

Google Tag Manager provides the management layer for GA4 events. Each conversion event defined in the mapping stage gets a corresponding GTM trigger and GA4 event tag. GTM’s preview mode allows the configuration to be tested before it goes live, confirming that events fire at the right moment with the right parameters. A properly structured GTM container is version-controlled and documented so that the configuration can be maintained and audited by anyone who inherits it.

Internal traffic filtering

Create a GA4 filter that excludes sessions from internal IP addresses. For organizations with dynamic IP addresses or distributed teams, alternatives include a custom dimension triggered by an internal cookie, or filtering at the GTM level. The goal is the same regardless of approach: internal browsing should not contaminate the conversion and behavioral data that marketing decisions are made from.

Cross-domain tracking

Organizations with multiple domains — a main site and a separate booking system, a subdomain for a customer portal — need cross-domain tracking configured so that a user’s session isn’t broken when they move between domains. Without it, a user who starts on the main site and completes a booking on a separate domain appears as two separate sessions, and the conversion is attributed to “direct” rather than to the channel that drove the original visit.

Consent mode implementation

For organizations subject to GDPR, CCPA, or other privacy regulations, consent mode needs to be implemented so that GA4 tracking adjusts when a user declines consent. This requires a consent management platform integrated with the site and a GTM configuration that reads consent signals and adjusts GA4’s behavior accordingly. Google’s consent mode v2 also affects conversion modeling in Google Ads — implementations that don’t include it may see degraded attribution in Google Ads campaigns.

Data retention settings

Change the event data retention setting from the default two months to the maximum available — 14 months for standard properties. Do this on day one. This setting doesn’t backfill; it only applies to data collected after the change is made. Setting it correctly at implementation ensures the property has the historical depth needed for year-over-year analysis.

How to Audit Your Current Setup

Start with conversions. Open GA4, navigate to Admin → Events, and look at which events are marked as conversions. If the list is empty or only contains automatically collected events like ‘session_start’ and ‘first_visit,’ the implementation is not producing conversion data.

Check internal traffic. Open the Realtime report and visit your own site. If you see yourself in the report, internal traffic is not filtered. Navigate to Admin → Data Streams → Configure Tag Settings → Define internal traffic to confirm whether any IP addresses are excluded.

Check channel grouping. Look at the Traffic Acquisition report. If a significant percentage of sessions are attributed to “unassigned” or if Direct is unusually high (above 20–25% for most sites), channel attribution is likely misconfigured. Pull a sample of direct-attributed sessions and examine their landing pages — URLs that suggest a paid or email origin appearing as direct confirm a UTM or channel grouping problem.

Check data retention. Navigate to Admin → Data Settings → Data Retention. If it’s set to two months, extend it immediately. When to rebuild versus fix: if the implementation has been running for more than six months with the default configuration, a rebuild with the correct configuration — starting from a clean baseline — is often preferable to attempting to fix the existing property. The old property’s data can be kept for reference, but a clean implementation gives the team data they can trust.

If your marketing team is trying to understand what your GA4 data actually tells you — or why it can’t be trusted yet — our marketing teams industry page covers how we work with in-house teams, and our search visibility service includes analytics configuration as part of the engagement.