Track the funnel, not one vanity number
Start with four events: question or search, answer or article shown, source article opened, and support contact started. They show whether visitors find a path to information. A “did this help?” control is useful directional feedback, not a complete measure of resolution.
Use gaps as a leading metric
The most useful metric for a new help center is often unanswered questions. Group them by intent each week. If five visitors ask variants of one question, write or improve a source article, then observe whether future searches find it.
Report carefully
Say “visitors viewed a source-linked answer” or “contact starts for this topic changed compared with the prior period.” Do not claim every answer saved a ticket. Once several weeks are stable, compare similar cohorts with and without the help-center experience and document the assumptions.