Free Resource
The Content
Planning Toolkit
Frameworks for deciding what to measure, how to name things in GA4, and how to tell the difference between a useful metric and a distraction.
The Question-First Dashboard Framework
Most dashboards are built by adding metrics. This framework inverts that process. You start with decisions, identify questions those decisions require, and then find the minimal set of metrics that answer them.
List the decisions
Write down the decisions your team makes regularly that could benefit from data. "Should we publish more blog posts?" is a decision. "Is our landing page working?" is a decision. "Which channel should get more budget?" is a decision. Aim for four to six recurring decisions.
Turn decisions into questions
For each decision, write the specific question that, if answered, would inform the decision. The question should be narrow enough to have a measurable answer. "Is our blog working?" is too broad. "Are organic visitors from the blog reading more than one article per session?" is specific enough to measure.
Find the minimum metrics
For each question, identify the smallest number of metrics that actually answer it. If you need three metrics to answer a question, use three. If two work, use two. Resist the temptation to add related metrics that are "nice to have." Every extra metric adds cognitive load without adding decision-making value.
Build one section per question
In Looker Studio or whichever dashboard tool you use, create a labeled section for each question. Use the question itself as the section title. Add only the metrics identified in step three. Review the dashboard in three months and remove any section you haven't used to make a decision.
GA4 Event Naming Reference
Consistent event naming is the foundation of reliable GA4 data. Once you've named an event and it starts collecting data, renaming it effectively means starting over for that metric. Getting names right at the start saves time later.
Use lowercase with underscores
GA4 event names are case-sensitive. "Form_Submit" and "form_submit" are two different events. Always use lowercase letters and underscores to separate words. This matches Google's own naming convention for automatically collected events.
Action then object
Structure event names as verb_noun. The action comes first, the thing being acted on comes second. This creates alphabetically organized event lists that group similar actions together and makes event names easier to understand at a glance in reports.
Be specific enough to be useful
A single "button_click" event for every button on your site is not useful. You need to know which button. Either use parameters to carry that detail (button_text, button_location) or create distinct event names for distinct actions (download_report, request_quote).
Avoid reserved names
GA4 has a list of reserved event names that you cannot use for custom events because they're used internally by the platform. These include names like app_remove, error, first_open, and in_app_purchase. Attempting to use them will either fail silently or cause data conflicts. The full list is in Google's documentation.
Common custom event name patterns
Signal vs. Noise: Which Metrics Actually Matter
Not every metric in GA4 is equally useful for every type of site. This framework describes how to evaluate whether a metric is a signal (something that should influence decisions) or noise (something that fluctuates without meaning much).
Signals
A signal is a metric that, when it changes, indicates something real has changed about how your site or content is performing. Signals are actionable: a meaningful change in a signal tells you something specific to investigate or act on.
- Conversion rate on a key landing page
- Engagement rate from a specific channel
- Average engagement time for content-heavy pages
- Event count for a specific intended action
- Sessions from organic search for target content
Noise
Noise is a metric that fluctuates for reasons unrelated to your site's performance, or one that changes without indicating anything you can act on. Noise metrics aren't useless, but they shouldn't drive decisions on their own.
- Total pageviews without context
- Raw session count without segmentation
- Direct traffic (often unattributed)
- Bounce rate on pages designed for single-page reading
- Real-time user count on a page without time-sensitive decisions
The signal test
Before adding a metric to your dashboard, ask two questions. First: if this number went up significantly, would I know what to do differently? Second: if this number went down significantly, would I know what to investigate first? If the answer to both is yes, it's a signal. If the answer to either is "not really," it's probably noise for your current purposes.
