User analytics in under five minutes

Use an events model for analytics--track page loaded, page left, and maybe a handful of other events (to taste). Spec out that the API (from the client JS) is something like:

  emitAnalyticsEvent( eventType, location, browserTime, optionalData )

On the server side, on receipt of the request, timestamp it again, and insert wholesale into a database table (in postgres, use the jsonb type to store the unstructured optionalData ). In optionalData, note the userAgent and IP from the headers. Additionally, note the session and user associated with the event.

Congratulations! You've now got a flexible (albeit barebones) setup.

Visualizing and analyzing the results may take more than an afternoon, but that's outside the collection scope. Even still, it's trivial to show user history, visitor IPs, and high-level traffic patterns.