GA4 KPI Framework: Choosing the Right Key Events for Your Business (2026)

How do I choose the right GA4 key events for my business?

Choose GA4 key events by starting from business outcomes, not from available events. For every candidate key event, ask: "Does a 10% increase in this event count make the business more or less money?" If the answer is "yes, directly" — it's a macro conversion (key event).

If the answer is "it depends" or "indirectly" — it's a micro conversion (tracked as a custom event, not marked as a key event). The most common mistake: marking too many events as key events. GA4's key event count metric adds up every event marked as a key event, so marking page_view, scroll, and purchase all as key events produces a meaningless aggregate.

A GA4 property with more than 5–7 key events usually has key event count inflation that makes the metric useless for reporting.

The tiered conversion hierarchy

Tier 1 — Macro conversions (mark as GA4 key events)

Events that directly represent business value:

Mark these as key events. Import them to Google Ads as primary conversion goals. Use them in Smart Bidding. Report them to stakeholders as the primary success metric.

Maximum macro conversions per GA4 property: 2–4. More than this and you're either tracking multiple distinct conversion types (which is valid) or you've inflated the definition of "macro."

Tier 2 — Micro conversions (track but don't mark as key events)

Events that indicate progress toward a macro conversion:

Track these as custom events. Use them in Explorations for funnel analysis. Use the highest-intent ones (add_to_cart, begin_checkout) as audience triggers for Smart Bidding signal. Do not mark them as key events in GA4 Admin — this keeps your key event count clean.

Tier 3 — Engagement signals (auto-collected events)

GA4's auto-collected events (scroll, click, page_view, session_start) provide engagement context but are not conversion signals. Use them in Explorations for content analysis. Never mark them as key events.

Business-type KPI frameworks

E-commerce KPI framework

Primary KPIs (GA4 key events):

• purchase — the only mandatory macro conversion
• refund (optional, negative signal) — mark to track if refund rate is a reporting requirement

Secondary KPIs (tracked, not key events):

• add_to_cart — funnel health indicator
• begin_checkout — checkout entry rate
• add_to_wishlist — intent signal for remarketing

Reporting KPIs (derived metrics in Looker Studio):

• Cart-to-purchase rate (add_to_cart ÷ purchase)
• Checkout completion rate (begin_checkout ÷ purchase)
• Revenue per session (total revenue ÷ sessions)
• Average order value (total revenue ÷ purchase count)

Smart Bidding import:

• Primary: purchase with transaction value (for Target ROAS)
• Secondary micro-conversion: add_to_cart or begin_checkout (for properties under 50 monthly purchases)

B2B Lead Generation KPI framework

Primary KPIs (GA4 key events):

• generate_lead — contact form, demo request, or trial signup submission
• qualified_lead (if implemented via audience trigger) — lead with qualifying attributes

Secondary KPIs:

• pricing_page_view — high-intent signal
• case_study_download — content engagement with purchase-consideration intent
• free_trial_started — if applicable

Reporting KPIs:

• Lead volume by channel (generate_lead by session_default_channel_group)
• Lead quality rate (qualified_lead ÷ generate_lead, if both are tracked)
• Cost per lead (requires Google Ads data blending in Looker Studio)

Smart Bidding:

• Import generate_lead as primary conversion
• If lead quality varies by channel, use qualified_lead (audience trigger) as primary and generate_lead as secondary

SaaS KPI framework

Primary KPIs (GA4 key events):

• free_trial_started — trial signup (top of monetisation funnel)
• subscription_started — paid conversion (the actual revenue event)

Intermediate KPIs (tracked, not key events):

• feature_X_activated — product activation milestone (feature usage indicating likely retention)
• trial_day_7_active — users still active 7 days after trial start (leading indicator of paid conversion)
• upgrade_page_view — intent signal for upsell campaigns

Reporting KPIs:

• Trial-to-paid conversion rate (subscription_started ÷ free_trial_started)
• Feature activation rate by cohort
• Monthly active users (proxy: users with session_start in last 30 days)

Content/Media KPI framework

Primary KPIs:

• newsletter_sign_up — email list growth
• subscription_started — paywall conversion (if applicable)

Engagement KPIs (tracked, not key events):

• article_completed (custom event, fired at 90% scroll depth on article pages)
• video_25pct, video_50pct, video_75pct, video_complete (auto-collected with enhanced measurement)
• share — social amplification signal

Reporting KPIs:

• Email capture rate (newsletter_sign_up ÷ sessions from organic search)
• Paywall conversion rate by content category
• Article completion rate by author or topic

The vanity metric problem

The most common KPI framework mistake: reporting page views, sessions, and bounce rate as primary KPIs. These are vanity metrics — they go up and down with traffic volume and require no business outcome to improve. An SEO campaign that drives 3x more traffic from irrelevant keywords produces:

• Higher page views ✓
• Higher sessions ✓
• Better bounce rate (if new landing pages have slightly better engagement) ✓
• Zero increase in revenue or leads ✗

A sound KPI framework reports business outcomes (purchases, leads, trial signups) as primary metrics and uses traffic and engagement metrics only as diagnostic context.

GA4 key events and Smart Bidding: the signal quality principle

For Smart Bidding to work effectively, your imported key events should be:

High frequency — at least 30 conversions per campaign per month (ideally 50+). Below this, Smart Bidding doesn't have enough signal to find patterns.

Directly tied to revenue — Smart Bidding optimises toward the conversion signal you give it. If you give it page_view as a conversion, it will optimise for page views, not revenue.

Accurate — a purchase event with duplicate transaction IDs, or a lead event that fires on page load rather than form submission, corrupts Smart Bidding's training data. Bidding degrades.

Frequently updated — don't import conversion goals and then ignore them. If your business redefines what constitutes a lead, update the conversion event accordingly.