Unique users via logs

Count how many users you have in your structured logs

Published: Wednesday, May 4, 2022 Last modified: Sunday, Jan 29, 2023

First you need to finger print your user, e.g. by setting a cookie or hashing the UA string.

func addContextMiddleware(next http.Handler) http.Handler {
	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		cookie, _ := r.Cookie("visitor")
		logging := log.WithFields(
				"id":      r.Header.Get("X-Request-Id"),
				"country": r.Header.Get("Cloudfront-Viewer-Country"),
				"ua":      r.UserAgent(),
		if cookie != nil {
			cvisitor := context.WithValue(r.Context(), visitor, cookie.Value)
			logging = logging.WithField("visitor", cookie.Value)
			clog := context.WithValue(cvisitor, logger, logging)
			next.ServeHTTP(w, r.WithContext(clog))
		} else {
			visitorID, _ := generateRandomString(24)
			// log.Infof("Generating vistor id: %s", visitorID)
			expiration := time.Now().Add(365 * 24 * time.Hour)
			setCookie := http.Cookie{Name: "visitor", Value: visitorID, Expires: expiration}
			http.SetCookie(w, &setCookie)
			cvisitor := context.WithValue(r.Context(), visitor, visitorID)
			logging = logging.WithField("visitor", visitorID)
			clog := context.WithValue(cvisitor, logger, logging)
			next.ServeHTTP(w, r.WithContext(clog))

Once you are logging a unique visitor attribute in your logs, you can count your users!

Count distinct users

A Log insights query like:

fields @timestamp, fields.visitor
| sort @timestamp desc
| filter ispresent(fields.visitor)
| stats count_distinct(fields.visitor) by bin(1h)

Check the Amazon documentation for more sample queries.

Also do checkout: https://www.rehanvdm.com/blog/aws-serverless-you-might-not-need-third-party-monitoring