“Connect analytics” can mean page views, contact clicks or checking received requests. These are different tasks. Without separating them, a report may show activity while leaving the owner unsure what to change on the website.
This guide helps write a brief for future measurement. It does not assume GA4, advertising tags or attribution are already running. Search readiness and first-month observations have separate purposes; here we agree definitions and data acceptance.
Start with a decision, not a metric list
Write a few questions whose answers would change the team’s actions. For example: do service page visitors reach contact, does submission complete, and do received requests concern the intended service? Give each an owner and a possible decision.
Do not collect every available metric just in case. If nobody will review the report or knows what to do with its findings, clarify the task. Views may describe attention to a page but do not themselves prove enquiry quality or sales.
Agree what counts as an enquiry
Separate opening contact, attempting submission, system acceptance and receipt by the responsible person. A phone click does not confirm a conversation, nor does opening a messaging app confirm a new message. For each channel, record what the website can confirm and what needs another source.
The GA4 recommended event reference includes generate_lead for a generated lead. An event name does not prove your implementation measures that outcome: agree and test its trigger.
Separately define a suitable request: the intended service, area and ability to continue the conversation. A manager’s assessment and a browser signal are different sources. Do not call an accepted form a completed order or assign an arbitrary monetary value.
Map events and their limits
The following is an educational service-website example, not a description of an analytics system already connected on XEVOR. Names and events depend on the agreed process; this table is not ready-to-use code for every tool.
| Observation | Agreed moment | What it does not prove |
|---|---|---|
| Service view | The page opens under the tool’s defined rule | All terms were read |
| Contact click | The person activates a contact link | A call or received message |
| Submission attempt | The form is passed for processing | Acceptance or delivery |
| Accepted request | The system confirms acceptance under an agreed rule | A manager read it or a sale occurred |
| Suitable enquiry | The team reviews it against defined criteria | Payment for an order |
Define necessary data and limits
Comparing pages may require page type, language or a fixed service code. Give each field a purpose and allowed values. A customer’s free-text description is unnecessary for counting a contact click; do not make an analytics event a copy of the enquiry.
Google’s PII prevention guidance excludes email addresses and personal phone numbers from Analytics; private data can also arrive through URLs. Check event fields and addresses, not just the tool’s name.
Before connecting collection, agree visitor information, necessary consent settings, access and retention with the responsible people. Specific requirements depend on the collection and business circumstances; this article does not replace legal review. For visit sources, define an allowed collection method first rather than adding arbitrary labels to every link.
Test events with controlled scenarios
For each event, write the expected result: when it should appear, when it should not and which fields are allowed. Use safe test data and an agreed environment. Separately check error correction, repeated clicks and refresh so one action does not become several results.
- An agreed successful scenario produces the expected event.
- An error or unconfirmed submission is not counted as success.
- Retries and refresh follow the agreed rule without accidental duplicates.
- Languages and contact channels differ only by necessary attributes.
- Consent restrictions and blocked tools are reflected in the findings.
- Test events can be separated from live data.
For an already connected GA4 setup, DebugView helps observe test events and parameters. Results also depend on privacy and consent settings. This checks tool signals, not receipt of an enquiry by a person.
Avoid conclusions from incomplete collection
Record the collection start date, time zone, report source and known gaps. Compare data with consistent event definitions and periods. If settings change mid-month, note that before discussing growth or decline.
A zero in a report can mean no actions or a measurement problem. Repeat the control scenario first. A mismatch with an enquiry log also requires checking definitions and collection limits rather than automatically concluding the website is ineffective.
For submission problems, start with form fields and states; check an awkward journey using the mobile scenario. Measurement helps frame a question but does not fix the interface.
What should remain in the brief
- Business question, owner and the decision supported by the result.
- Definitions of enquiry, success, error and suitable request.
- Event map with triggers and allowed fields.
- Selected tool, access, privacy and collection limits.
- Acceptance scenarios, separating tests and the start date.
- Review format and known limitations.
Agree measurement before implementation if it is part of the project. A prepared brief is not a connected system, and a tool name in an estimate does not replace verification. For an existing website, record current collection before planning changes.
Discuss business website requirements · Send a task description
Measurement plan FAQ
Is counting button clicks enough?
It may help assess intent. Counting received enquiries needs different confirmation and a separate definition.
Do I have to choose GA4?
Agree the task and limits first. Choose a tool for the required collection, access and working arrangements, not popularity alone.
Will new analytics restore past events?
Do not assume so. Data depends on what was previously collected and retained; record when new collection starts.
Do more events mean more sales?
No. Traffic, measurement rules or retries may change. Confirm sales with separate business records.