Custom dimensions or properties are used for audience segmentation

Without user properties, every analytics report averages free users, paid users, enterprise accounts, and trial signups into a single number that describes no actual cohort. You cannot compare activation between plans, cannot isolate power-user behavior, cannot segment by signup date to measure cohort quality, and cannot feed PLG motion with account-level data. Anonymous event streams produce vanity metrics; identified event streams produce business intelligence.

After authentication succeeds, identify the user with stable properties in lib/auth.ts:

analytics.identify(user.id, {
  plan: user.plan,
  created_at: user.createdAt,
  company_size: user.companySize,
})

Use opaque database IDs, never email addresses or names, to stay within PII boundaries. Set properties for plan changes and churn events so cohort reports reflect the current state, not the signup state.