GA4 only records page views by default, but what actually guides action is key events: how far the user scrolled, how long they stayed, whether they completed the action you wanted. Without key events marked, the report is just vanity metrics like visits — no matter how much you look, you can’t tell whether content is worth it, and budget gets split by gut feel.
Here’s the bottom line: mark scroll depth, meaningful engagement, and conversion actions as key events, and GA4 turns from a visit report into a data source that guides decisions. Three types of events are enough to answer whether a piece of content got read to the end, got trusted, and drove action. For the full chain from click to lead capture, the click-to-lead article is a good reference.
Decide what to record first — don’t mark everything
More tracking isn’t better. Too many events and the report is all noise, with the key metrics drowned out. Before starting, list your business goals: content sites usually have “read to the key position”, “stay long enough”, and “click out or leave a lead” — each goal maps to one event, keep it under five.
Naming should also be unified before you start. Event names use lowercase with underscores (e.g. cta_click), parameter names have fixed meanings, and once live they never change — renaming breaks historical data and year-over-year comparisons become meaningless.
Three types of events worth marking
Scroll depth: mark 50% and 90% to distinguish “casually swiping” from “actually reading to the end”. Meaningful engagement: use engaged sessions to see whether users stayed over 10 seconds or interacted. Key clicks: mark the end-of-article CTA, outbound link clicks, and form submissions as events, linking straight to conversion.
| Event | Meaning | Question it answers |
|---|---|---|
| scroll_50 / scroll_90 | Scroll depth | Did the content get read into |
| engaged_session | Engaged session | Were users willing to stay |
| cta_click / lead | Key clicks and lead capture | Did the content drive action |
Building it step by step in the GA4 console
Scroll and dwell are mostly auto-collected via enhanced measurement: go to Admin → Data Streams, select your website data stream, turn on enhanced measurement, and check “scroll” to get 90%-depth events with no code. For custom depths like 50%, you need GTM’s scroll depth trigger.
Custom actions like CTA clicks and form submissions need manual tracking: send them directly with gtag, or build a trigger in GTM bound to a button selector. After tracking, go to Admin → Events and mark them as key events; for ones that should count as conversions, check them in conversion settings so the data enters the attribution model. The most common miss for beginners is exactly this step — the event is collected but never marked, so its contribution never shows in reports, and the tracking is wasted.
| Step | Where | Purpose |
|---|---|---|
| Turn on enhanced measurement | Admin → Data Streams | Auto-collect scroll and dwell |
| Create custom events | GTM or Events page | Collect CTA and form submissions |
| Mark as key events | Admin → Events | Enter analysis and exploration reports |
| Set as conversions | Conversion settings | Enter attribution model calculation |
After tracking, self-test with DebugView first: turn on debug mode, click around the page, confirm event names and parameters report as expected, then wait 24 hours to check the standard report. Don’t assume the data is wrong right away.
Watch event rates, not event totals
Looking at event totals alone gets distorted by traffic scale — a high-PV article naturally has more events, but that doesn’t mean the content is good. Look at rates instead: scroll_90 share, CTA click rate, engaged session rate, sorted across articles horizontally, to find the ones that truly keep people.
Feed these rates into the weekly report — it’s less work than screenshots every week. The approach of pulling data from the GA4 API into a sheet with auto-charts can be copied straight from the GA4 + Sheets weekly dashboard.
Layer content with events to decide where resources go
Cross 90% scroll rate and CTA click rate to split articles into three layers, each with different actions:
- High read, high convert: core assets. Add internal links, FAQ, and on-site recommendation slots to amplify them.
- High read, low convert: good content, weak guidance. Add CTAs, swap hooks, move conversion entry points forward.
- Low read: title and content misaligned. Rewrite the title, rewrite the opening, or merge into a related long-form article.
The layered result reads more accurately alongside content funnel analysis: the funnel tells you which stage people drop at, event rates tell you which specific articles are dropping. Look at both together and you have evidence for which article to change and which part.
Closing the loop back to SEO
Cross-reference GA4 key events with Search Console impressions and clicks and the problems surface by themselves. An article with high GSC clicks but low scroll_90 means the title lured people in but the content didn’t deliver — fix the opening and structure; the reverse — low clicks but high read-to-end — means ranking or snippet isn’t landing, so optimize the title and description.
Go a step further and make key-event rates the content KPI replacing PV. For multi-channel contribution, attribution models split it apart — a lot of planting-style content gets badly undervalued under last-click attribution and only shows its worth with data-driven attribution. What you report becomes “this article drove this many effective reads and this many leads” instead of “this many views” — the former negotiates budget, the latter only negotiates buzz.
Five things to run through before landing
Turn on enhanced measurement in the data stream settings and confirm scroll events are collected; track cta_click and form submission with GTM, locking names once; mark them as key events on the Events page and set conversion counting in parallel; self-test with DebugView and let it run a full week after confirming parameters; next week, rank articles by scroll_90 rate and CTA click rate and push resources to the top few. Run through these five and GA4 truly starts speaking for you.
