Most businesses are flying blind. They have Google Analytics installed, a green tick somewhere in their dashboard, and absolutely no idea if the data they're collecting is accurate, complete, or even legally compliant. GA4 isn't something you simply "install and forget", it's a measurement infrastructure that, when configured correctly, becomes the single most valuable asset in your entire marketing operation.
What You'll Learn in This Lesson
By the end of this 55-minute lesson, you'll understand how to set up Google Analytics 4 from scratch, configure it correctly for reliable data collection, and avoid the critical errors that corrupt reporting before a single campaign even launches. If you're configuring GA4 for a new property or auditing an existing one, this lesson gives you the professional framework Byter uses with every client. You'll also walk away with a repeatable configuration checklist you can apply to any property, new or inherited, and the language to explain configuration decisions clearly to stakeholders and clients who assume analytics just "works."
AN902-01: GA4 Setup and Configuration, Key Concepts
Here is the number that should end every argument about analytics being a "technical" concern rather than a strategic one.
According to Econsultancy (2024), only 34% of marketing professionals report high confidence in the accuracy of their analytics data. Two thirds of all marketing decisions, budget allocations, channel investments, creative pivots, are being made on data that the people collecting it don't even trust. And in the UK specifically, where ICO enforcement of GDPR has resulted in fines exceeding £1 million for household-name brands, the consequences of a misconfigured GA4 property extend well beyond inaccurate dashboards.
GA4 became Google's default analytics platform in July 2023, replacing Universal Analytics permanently. Yet according to Google's own Looker Studio usage data (2024), a significant proportion of GA4 properties are still running with default configurations, missing critical event tracking, and collecting data in ways that breach UK GDPR without the teams responsible even knowing. The ICO's guidance on analytics cookies is explicit: consent is required before any non-essential tracking fires. Default GA4 configurations do not meet this standard out of the box.
This matters commercially too. When a senior stakeholder asks why paid search revenue is down 18% quarter-on-quarter and your honest answer is "we're not sure the data is reliable enough to say," you've lost credibility that takes months to rebuild. Conversely, when your GA4 implementation is airtight, you become the person in the room who can actually answer questions with confidence. That is an enormous professional advantage. The Byter Revenue Attribution Matrix, which maps every marketing pound to revenue using first-touch, last-touch, and multi-touch models, is only as useful as the GA4 data feeding into it. Garbage in, garbage out. This lesson exists so you never have to say the data might be wrong.
Understanding GA4's Architecture Before You Touch a Setting
Before configuring anything, you need to understand how GA4 thinks about data. This isn't abstract theory, it directly affects every decision you'll make during setup.
GA4 is built on an event-based data model. Unlike Universal Analytics, which categorised data into hits, sessions, and pageviews, GA4 treats every single interaction as an event. A pageview is an event. A scroll is an event. A button click is an event. A purchase is an event. Everything is an event, and every event can carry up to 25 parameters that describe it in detail.
This is powerful, but it also means that what you measure is entirely intentional. GA4 won't magically know what matters to your business. You have to tell it.
The GA4 Property Hierarchy
Understanding the account structure prevents costly mistakes:
Account, The top-level container, typically one per organisation
Property, Represents a single app or website (or both together)
Data Stream, The individual source of data flowing into a property (web, iOS, Android)
A common mistake is creating multiple properties when one would do, or vice versa. For most businesses with a single website, you need one account, one property, one web data stream. E-commerce businesses with separate mobile apps may need additional streams, but the principle is always to consolidate unless there's a specific reporting reason to separate.
Consider a mid-sized UK retailer that runs a transactional website, a blog subdomain, and a wholesale trade portal. A poorly thought-through setup might create three separate GA4 properties, one for each. The result is fragmented audiences, no cross-domain journey visibility, and three times the configuration effort. The correct approach for most such scenarios is a single property with careful URL filtering in reports and appropriately scoped custom dimensions to distinguish user types. Only create separate properties when the business genuinely cannot share data between those environments, such as two entirely distinct brands with separate P&Ls.
GA4 vs Universal Analytics: The Key Differences
If you or your clients have experience with Universal Analytics, recognise upfront that GA4 isn't simply a rebrand. It's a fundamentally different measurement philosophy. In UA, the session was the primary unit of measurement. In GA4, the event is. In UA, goals were configured within the interface. In GA4, any event can be promoted to a conversion. In UA, you needed separate properties for apps and websites. In GA4, a single property can unify both through a cross-platform data model.
This shift means UA muscle memory can actively mislead you. Practitioners who assume GA4 behaves like UA, particularly around session counting, bounce rate (now replaced by engagement rate), and attribution windows, frequently build dashboards that surface the wrong metrics entirely.
Step-by-Step: Creating a GA4 Property Correctly
1. Account and Property Creation
Navigate to analytics.google.com and create a new account. Use your organisation's legal name, this matters for data processing agreements. When creating the property, select your reporting time zone carefully. This should reflect where your primary business operations are, not necessarily where your audience is. Changing this later creates historical data discrepancies that are genuinely painful to explain to clients.
Set your currency correctly from day one. If you're running an e-commerce property and your transactions are in GBP but your property is set to USD, your revenue reporting will be wrong in ways that aren't immediately obvious. We've seen this exact issue at Byter during client onboarding audits, a fashion retailer had been reporting revenue in dollars for eleven months before anyone noticed. The underlying transaction data was fine, but every board-level revenue report had been inflated by the prevailing exchange rate. Fixing the property currency setting going forward doesn't retroactively correct historical data in Explorations.
2. Configuring Your Web Data Stream
When adding your web data stream, you'll be asked for your website URL. Use the canonical version, either with or without www, but be consistent. Mismatched URL configurations are one of the most common sources of duplicate session data.
Enhanced Measurement will be toggled on by default. This automatically tracks scrolls, outbound clicks, site search, video engagement, and file downloads. Leave most of these on, they provide valuable data without additional tagging. However, turn off "Form interactions" unless you've specifically configured your forms. The default form tracking in Enhanced Measurement is notoriously unreliable and frequently fires on non-meaningful form interactions, polluting your conversion data.
Also review the site search settings within Enhanced Measurement. By default, GA4 looks for the query parameter q in your URLs to identify search terms. If your site's internal search uses a different parameter, search, s, query, or something custom, update this setting or you'll see no site search data at all. Check your URL bar after running a search on your own site to identify the correct parameter before finalising this.
3. Installing Your GA4 Tag
You have three main options:
Google Tag Manager (GTM), Recommended for almost every situation. Provides flexibility, version control, and doesn't require developer access for most changes.
Direct gtag.js installation, Acceptable for simple sites, but creates dependency on developer deployments for every change.
CMS integrations (WordPress plugins, Shopify apps), Convenient but often create duplicate tags or miss key configurations. Use with caution.
Byter's recommendation: Always use Google Tag Manager. The marginal setup time pays dividends within the first month.
When setting up GTM, create a GA4 Configuration tag (now called the "Google Tag" in newer GTM versions) with your Measurement ID, it begins with G- and is found in your data stream settings. Set this tag to fire on All Pages. Then, for any additional event tags you build later, reference this same Measurement ID rather than creating separate configuration tags. A clean GTM container has one GA4 configuration tag and as many event tags as your measurement plan requires, not multiple configuration tags that create conflicting data.
Byter Tip
Byter Insider: We audited a hospitality group in Shoreditch, East London, running three restaurant sites and an events booking platform. Every site had GA4 installed via a WordPress plugin, and two of them also had a legacy Universal Analytics snippet still firing from a previous agency's work. The result: triple-counted sessions on peak evenings, a reported bounce rate of 8% (which is physically impossible for that type of site), and conversion data so corrupted that six months of paid social optimisation had been based on entirely fictional numbers. We stripped everything back, deployed a single GTM container across all four properties, configured Consent Mode v2 to satisfy UK GDPR requirements, and rebuilt the event tracking from a clean measurement plan. Within eight weeks, their actual conversion rate was visible for the first time. It was 2.3%, not the 14% the broken setup had been showing. Uncomfortable to discover, but essential. You cannot optimise what you cannot accurately measure.
Critical Configuration Steps That Most People Skip
Configure Your Data Retention Settings
By default, GA4 retains event data for two months. If you're building year-over-year reports, comparing seasonal campaigns, or tracking long conversion cycles, which is almost every business, this is completely inadequate.
Go to Admin → Data Settings → Data Retention and change this to 14 months. This is the maximum available on the free tier and should be the first thing you change after creating a property. Note: this doesn't affect aggregated reports in the standard reporting interface, only the raw event data you can query in Explorations. If you need longer retention or more granular data access, GA4 360 (the paid tier) or a BigQuery export connection are your options.
Set Up Internal Traffic Filters
Your own team visiting your website contaminates your data. If you're a 15-person agency checking the client's site constantly, you could be inflating session counts by 20–30% or more for smaller properties. For a business with 500 monthly sessions and a 10-person team checking the site daily, internal traffic could represent the majority of reported activity.
Go to Admin → Data Streams → Configure Tag Settings → Define Internal Traffic, add your office IP address(es), then go to Admin → Data Filters → Create Filter and set the filter to exclude internal traffic. Set it to Active, not Testing, once you've verified it's working correctly.
One practical note: if your team works remotely or uses dynamic IP addresses, consider also excluding internal traffic by user property. You can fire a GTM trigger that sets a custom user property (e.g. internal_user: true) when someone visits a specific URL or enters a developer mode. This isn't a perfect solution but provides a reasonable supplement to IP-based filtering for distributed teams.
Set Up Cross-Domain Tracking (If Applicable)
If your business uses multiple domains, for example, a main website at yourbrand.co.uk and a checkout hosted at checkout.paymentprovider.com, you need to configure cross-domain tracking or GA4 will create a new session every time a user crosses between domains. This makes your checkout funnel completely unreadable.
In Admin → Data Streams → Configure Tag Settings → Configure Your Domains, add every domain that should be treated as part of the same user journey. GA4 will then maintain session continuity and user identity across those domains. This is particularly critical for businesses using third-party booking tools, Shopify storefronts, or payment gateways hosted on separate domains.
Link Google Search Console
This is free, takes four minutes, and gives you keyword-level organic search data directly inside GA4. Go to Admin → Search Console Links and connect your verified Search Console property. Your organic landing page reports will transform from "not set" nightmares into genuinely useful channel intelligence. Once linked, the Search Console data appears in GA4 under Reports → Acquisition → Search Console, giving you queries, clicks, impressions, and click-through rates in context with your on-site behaviour data, a combination that no other free tool provides.
Connect Google Ads
If you're running paid search, this link is non-negotiable. Without it, your Google Ads traffic appears as generic CPC with no campaign-level attribution. Go to Admin → Google Ads Links and connect the relevant Ads account. Once linked, enable Auto-tagging within Google Ads to ensure GCLID parameters populate correctly. Also confirm that your GA4 audiences are being imported back into Google Ads, this enables remarketing and smart bidding signals to use your GA4 conversion data, which is substantially more powerful than pixel-only conversion tracking.
Common Mistakes Practitioners Make
Warning
These are the five configuration errors Byter sees most frequently when auditing client GA4 accounts. Each one compromises your data in ways that can take months to diagnose.
Not verifying tag implementation, Adding GA4 to GTM and assuming it's working. Always confirm with the GA4 DebugView (found under Admin → DebugView) before considering setup complete. You should see events firing in real time when you browse the site. Specifically, look for page_view firing on every page load and confirm the page_location parameter is populating correctly.
Ignoring consent mode, In the UK and EU, deploying GA4 without a properly configured consent management platform (CMP) and Google Consent Mode v2 is a GDPR violation. This isn't optional. The ICO has been explicit that analytics cookies require prior consent under UK GDPR, and enforcement activity against British businesses has increased materially since 2023. Platforms like Cookiebot, OneTrust, or CookieYes all offer GTM integrations. Consent Mode v2 allows GA4 to collect cookieless, modelled data from users who decline tracking, meaning you lose less measurement signal while remaining legally compliant. Skipping this doesn't just create legal risk; it also degrades your data quality by excluding a growing proportion of privacy-conscious users from your reports entirely.
Marking everything as a conversion, GA4 allows you to mark any event as a conversion. New users mark page_view, scroll, and session_start as conversions, then wonder why their conversion rate is 95%. Mark only meaningful business outcomes: purchases, lead form completions, phone call clicks, and specific high-intent content engagements. As a rule of thumb, if you wouldn't report it to your MD as evidence of commercial progress, it shouldn't be a conversion.
Leaving the default channel grouping, GA4's default channel grouping misattributes significant traffic. Email campaigns without UTM parameters land in Direct. Organic social from newer platforms gets dumped into Unassigned. Spend an hour configuring custom channel groups under Admin → Channel Groups. At minimum, create rules for your email provider's known sending domains, your key social platforms, and any affiliate or partner traffic sources you use regularly.
Not documenting the configuration, You'll onboard new team members, switch agencies, or return to this property after six months. With no documentation, you'll have no idea why certain events exist, what triggers them, or which configurations are intentional. Maintain a living measurement plan document from day one. We've taken over client accounts where a bespoke event called form_success_v3 was firing on the most important conversion action, with no record of what v1 and v2 were, or why they were replaced. It took three hours of archaeology to understand a five-minute configuration decision made eighteen months earlier.
The Measurement Plan: Your Configuration Blueprint
Before touching a single setting in GA4, professional analytics practitioners start with a Measurement Plan, a structured document that defines:
Business objectives, What does the business need to achieve?
Key questions, What does marketing leadership need to answer?
User actions, What interactions indicate progress toward objectives?
Events and parameters, How will those interactions be captured in GA4?
Success metrics, How will you know if performance is good?
This framework, popularised by Simo Ahava and widely adopted in the analytics community, prevents the most common GA4 failure mode: collecting enormous amounts of data that answers none of your actual business questions.
A practical example: a B2B SaaS business might define their primary business objective as "increase qualified pipeline from inbound digital channels." Their key questions become "which channels drive the most demo requests?" and "which content types convert researchers into form submitters?" Their measurement plan then flows directly from those questions, events are defined for demo_request_submitted, content_download, pricing_page_view, and video_complete, each with parameters that capture enough context (content title, traffic source, user type) to actually answer the question. Without the measurement plan as a starting point, the same business would likely end up tracking scroll and outbound_click religiously while having no idea how many demo requests came from paid versus organic.
The GA4 Measurement Plan Framework, strategy before configuration ensures every event answers a real business question
Understanding GA4 Event Types
GA4 organises events into four categories, and understanding the distinction shapes how you plan your implementation:
Automatically collected events fire without any additional configuration once the GA4 tag is installed. These include first_visit, session_start, and user_engagement. You cannot disable these and they don't require GTM setup beyond the base configuration tag.
Enhanced Measurement events are collected automatically when Enhanced Measurement is enabled on your data stream. These include scroll, click (for outbound links), view_search_results, video_start, video_complete, and file_download. As noted earlier, configure these deliberately rather than accepting all defaults blindly.
Recommended events are predefined event names that Google recommends you use for specific industries and interaction types. For e-commerce, these include add_to_cart, begin_checkout, purchase, and view_item. Using recommended event names unlocks pre-built GA4 reports and ensures compatibility with Google's machine learning features, including predictive audiences. Inventing your own event name, prodAddedCart instead of add_to_cart, means losing access to these built-in reports.
Custom events are anything you define yourself for interactions that don't fit Google's recommended taxonomy. These are fully valid but require custom Explorations or Looker Studio dashboards to report on, as they don't appear in standard reports automatically.
GA4's four event categories, only recommended and custom events require active GTM implementation, but all four need deliberate configuration decisions
BigQuery Export: Planning for Scale
If you're working with high-traffic properties, data science teams, or clients who need custom attribution modelling, it's worth knowing that GA4 offers a free native export to Google BigQuery. This gives you access to raw, unsampled, user-level event data that you can query with SQL, connect to your own data warehouse, or combine with CRM data for deeper analysis.
Enabling BigQuery export won't affect your standard GA4 reporting, but it future-proofs your measurement infrastructure considerably. Go to Admin → BigQuery Links to set it up. Even if you don't immediately use the exported data, having a historical record in BigQuery from day one is significantly more valuable than wishing six months later that you'd turned it on sooner. GA4's standard interface only retains Explorations data for 14 months at most. BigQuery export gives you unlimited retention by default.
Tool Recommendations
Google Tag Manager, Free, essential. The foundation of any serious GA4 implementation.
GA4 DebugView, Built into GA4. Use it every single time you make a configuration change.
Tag Assistant Companion (Chrome extension by Google), Confirms which tags are firing on each page in real time.
Cookiebot or CookieYes, Consent management with native GTM integration and Consent Mode v2 support. Both carry ICO-recognised consent documentation, which matters if you're ever audited.
Screaming Frog, Useful for auditing URL structures before configuring filters and exclusions.
Looker Studio, Google's free dashboarding tool. Connect directly to GA4 for client-ready reporting that goes beyond the standard GA4 interface.
GA4 Debugger (Chrome extension by Analytics Debugger), More detailed event inspection than Tag Assistant alone; useful for verifying parameter values alongside event names.
Key Takeaways
GA4 uses an event-based data model; every interaction is an event with parameters, and you must intentionally design what you measure
One property per website is almost always correct; over-segmenting properties creates reporting fragmentation
Google Tag Manager should be your deployment method in nearly every scenario
Change data retention to 14 months immediately, the default two months is inadequate for most business reporting needs
Internal traffic filters, Search Console links, and Google Ads connections are non-negotiable configuration steps
Cross-domain tracking must be configured if your user journey spans more than one domain
Consent Mode v2 is a legal requirement under UK GDPR, enforced by the ICO. It is not a nice-to-have
Use recommended event names for commercial interactions, they unlock GA4's built-in reports and machine learning features
Always build a Measurement Plan before configuring tracking, strategy before implementation
Consider enabling BigQuery export from day one, even if you don't need it immediately
Document everything; undocumented GA4 configurations become liabilities within months