Google Analytics 4 has been the default Google Analytics since July 2023 when Universal Analytics shut down. Three years on, most UK SMB GA4 implementations are still half-configured — basic page-views and traffic-source reporting working, but the conversion configuration, ecommerce parameters, consent-mode and BigQuery export missing. This guide is the complete UK 2026 playbook for the configuration most SMB sites should have but typically do not.
What GA4 actually is
GA4 is a fundamentally different analytics product from Universal Analytics. Event-first rather than session-first — every interaction is an event with structured parameters. Cross-platform by default — a single property tracks web, iOS app and Android app under one identity model. Privacy-first by design — IP anonymisation is implicit, the user model relies less on persistent identifiers, consent-mode integrates more deeply. Predictive metrics native — purchase probability, churn probability, revenue prediction built into the platform. The conceptual shift from Universal Analytics is meaningful and most teams underestimate how much of their UA knowledge does not transfer.
1. Property and data-stream structure
Start with one GA4 Property per organisation, with one Data Stream per platform (web, iOS app, Android app). Avoid the common mistake of creating multiple Properties for what is really one organisation — cross-Property analysis is limited and you lose the unified user view that GA4’s identity model is built around. The Data Stream is where the measurement ID lives; the gtag.js script or Google Tag Manager configuration references the Data Stream measurement ID.
2. Enhanced Measurement and the automatic events
GA4’s Enhanced Measurement automatically tracks several event types: page_view, scroll (90% scroll depth), click (outbound links), site_search (URL query parameters), video_start / video_progress / video_complete (YouTube embeds), and file_download (clicks on common file extensions). Enable Enhanced Measurement on the Data Stream but be specific about which automatic events you want — some create noise more than signal. Scroll events at 90% are typically not the right threshold; consider whether you want them. Click events on outbound links are usually valuable; site_search is usually valuable if your site has search.
3. Custom event configuration
Beyond the Enhanced Measurement automatics, define the events that matter to your business. Common SMB custom events: form_submit (every lead form), generate_lead (every qualified form submission with the lead value where calculable), contact_call (every tappable phone-number click), purchase (every transaction with full ecommerce parameters), add_to_cart (every basket add), view_item (every product page view). Each event carries parameters describing the action (form_id, lead_value, product_id, etc.). Send events via gtag.js, Google Tag Manager, or directly via the Measurement Protocol for server-side use cases.
4. Conversion configuration
GA4’s "Conversions" (renamed "Key Events" in late 2024) are the events that count as primary objectives. Mark the events that matter as conversions inside the Admin → Conversions panel (or, in newer terminology, Admin → Events → toggle "Mark as conversion"). Common SMB conversions: generate_lead, purchase, contact_call. Each conversion carries its own conversion-rate, conversion-by-source, conversion-by-campaign reports automatically.
5. Consent-mode v2 for UK GDPR / PECR
Consent-mode v2 is the GA4 mechanism for respecting user consent under UK GDPR and PECR while still capturing useful analytics. The flow: the consent banner captures user choice (analytics_storage = granted or denied, ad_storage = granted or denied). gtag.js or GTM passes the consent state to GA4 before any event fires. Where consent is denied, GA4 receives pinged events but stores no identifiers; the data feeds modelled conversions in Ads but does not create individual user profiles. The implementation is technical and easy to misconfigure — common mistakes include firing events before consent is established, or not passing consent updates when the user changes their choice.
6. Ecommerce setup
For ecommerce sites, the GA4 ecommerce parameters need to be sent on every relevant event. view_item with currency, value, items[] (each item with item_id, item_name, item_brand, item_category, item_variant, price, quantity). add_to_cart with the same items[] structure. begin_checkout, add_payment_info, add_shipping_info, purchase. Each event needs the correct parameter shape; getting them right unlocks the full GA4 ecommerce reporting suite (purchase journey, checkout funnel, product performance, revenue attribution). Shopify, BigCommerce and most ecommerce platforms emit these natively where the GA4 integration is configured.
7. BigQuery export
BigQuery export is free on the standard GA4 (was paid on the legacy GA360 Universal Analytics). Configure it once and GA4 streams every event into BigQuery for SQL analysis. For SMB sites this is often unused — the in-platform reporting covers the basic needs — but for any site with material data analysis ambitions, BigQuery export is genuinely useful: custom funnel analysis, cohort analysis, multi-touch attribution, joining analytics data with offline CRM data. The setup is one toggle inside Admin → BigQuery Linking.
8. Cross-domain tracking
For sites that span multiple domains (the marketing site at example.com and the booking flow at app.example.com, or the main brand at brand.co.uk and the secondary brand at brand.de), cross-domain tracking links them under a single user identity in GA4. Configure inside Admin → Data Streams → Configure tag settings → Configure your domains. Add every domain that is part of the user journey. Without this, every domain hop starts a new session and the funnel reporting breaks.
9. Internal traffic filtering
Filter out internal traffic so the analytics reflects real visitors. Inside Admin → Data Streams → Configure tag settings → Define internal traffic. Define the IP addresses or IP ranges your team uses. Then Admin → Data Filters → toggle the Internal Traffic filter from Testing to Active. Without this, your office and home IPs distort the conversion-rate analysis.
10. The audit checklist
Eight checks every GA4 implementation should pass. (1) Real-time report shows your traffic correctly when you load the site in incognito. (2) Conversions panel lists the right key events marked as conversions. (3) Ecommerce reports show transactions correctly if you have ecommerce (test purchase if necessary). (4) Consent-mode shows the expected granted/denied breakdown matching your consent banner data. (5) Internal traffic is filtered. (6) BigQuery export is configured if you need it. (7) Cross-domain tracking is configured if you span multiple domains. (8) Google Ads and Google Search Console are linked for the cross-product reporting.
Common mistakes
Five recurring failures in UK SMB GA4 audits. (1) Universal Analytics tag still firing alongside GA4 — wasted bandwidth, conflicting data. (2) Conversions configured at the page-URL level rather than event level — misses the actual interaction. (3) Consent-mode not configured at all (or configured incorrectly), producing either over-collection (privacy risk) or under-collection (modelled-conversion accuracy loss). (4) Ecommerce parameters partially correct (purchase events firing but without the items[] array), breaking the funnel reports. (5) BigQuery export never enabled, leaving meaningful analysis depth on the table.