Track relevant discovery, useful page actions, accepted inquiries and qualified conversations as separate stages. Connect analytics to durable inquiry records using identifiers, and keep personal details out of broad event properties.
Start with the business question
The first measurement question for a new Medicare marketing site is whether agency buyers discover useful pages and take an appropriate next step. Total page views cannot answer that alone. Separate agency-owner traffic from beneficiary, student or unrelated traffic where query and inquiry evidence allow it. Keep unknown audience cases explicit instead of automatically classifying every visitor as a buyer.
Write the funnel before installing events: discovery, page visit, relevant next-step action, accepted inquiry, reviewed fit, booked conversation and held conversation. Define each stage and its source of evidence. A browser event can show an interaction; a stored inquiry proves acceptance; a response owner’s disposition helps identify fit. Different systems answer different parts of the question.
A practical event dictionary
Use a small set of events such as article_view, tool_started, tool_completed, contact_started, inquiry_accepted and inquiry_error. These are proposed names for an implementation, not a claim about a package’s built-in features. Define when each fires and list allowed properties: page path, tool name, broad campaign labels, error category and non-identifying inquiry ID where appropriate.
Do not put names, email addresses, telephone numbers or free-text messages into analytics properties. The controlled inquiry system can retain the actual request under the appropriate access rules. Analytics should help identify which step works or fails without exposing every submission to everyone who can view a dashboard.
- Event: the exact behavior being measured.
- Trigger: the condition that makes it fire once.
- Properties: the minimum non-sensitive context needed.
- Evidence: browser interaction, server acceptance or reviewed outcome.
- Owner: the person responsible for interpreting and maintaining it.
Reconcile events against records
Test one accepted inquiry and inspect analytics, durable storage and downstream delivery. Then test a validation failure, duplicate retry and blocked analytics environment. A contact-button click should not count as an accepted inquiry. An accepted request should remain in the inquiry system even if analytics is unavailable. The acquisition report should explain these boundaries.
Compare accepted events with stored inquiries periodically and investigate differences. Duplicated events, retries and blocked scripts can make counts diverge. Use inquiry identifiers for reconciliation where appropriate, while keeping the dashboard free of the message body. Mark internal tests so they do not become evidence of market demand.
Read search and business outcomes together
Search Console and Bing reporting help explain discovery and query relevance. On-site analytics explains page behavior. Inquiry dispositions and appointment records explain whether relevant conversations followed. Do not force those counts to match; their definitions and collection boundaries differ. Record the reporting period, filters and missing data before comparing trends.
For the first monthly review, list what is known, what is functioning and what remains a hypothesis. A useful conclusion might be that relevant queries appeared and contact delivery passed testing, while inquiry volume remains too small to assess repeatability. Choose the next improvement from the weakest evidenced stage, such as a confusing contact action or a broken tool.
Questions agents ask
Is a form-start event a lead?
No. Report started forms separately from accepted inquiries and reviewed qualified requests.
What should analytics store about a contact?
Use only the minimum non-sensitive context and an appropriate identifier. Keep personal details in the controlled inquiry record.
Sources & further reading
Primary references for the rules and platform mechanics discussed here. Templates and recommendations are Medicare Marketing’s planning guidance. Examples are illustrative unless explicitly sourced.
Turn the idea into a next step.
Use the free toolkit to check your local foundation, build a content rhythm, or plan a webinar.
Explore the tools →