Why is percentage of new users not 100% on day 1?

Question

Looking at the first week we started using Google Analytics (this is a few years ago, so all numbers have been processed), I see these numbers:

• Overview:
  • Users: 1689
  • Sessions: 2533
  • Percentage new sessions: 46.55%
• New vs Returning:
  • New users: 1180
  • New visitor sessions: 1180
  • Returning visitor sessions: 1353

Very confusing. Is there is difference between "new users" (1180 users) and "new visitors" (1180 sessions)? And between "new sessions" and "new visitor sessions" (46.55% = 1180 / 1353)?

But most importantly, why isn't the number of "New users" (1180) not equal to the number of "Users" (1689) during the first reporting period? Has Google secretly been registering who visited our site, before we started using Analytics?

Answer 1

First, Google changed their algorithm in early 2017 (https://support.google.com/analytics/answer/2992042?hl=en) since you didn't give what time frame your data comes from, this is one possible reason. In the previous algo, Google used sampling If the date range for your report includes data from prior to September 2016, then sampling occurs.

Google also states Analytics uses two different techniques for calculating Users for different kinds of report requests. As a result, you may notice discrepancies in Users in different reports. so it's difficult to compare counts between reports.

You may want to read https://support.google.com/analytics/answer/3123662 which says:

In an Analytics implementation without the User-ID feature, a unique user is counted each time your content is accessed from a different device and each time there’s a new session. For example, a search on a phone one day, purchase on a laptop three days later, and request for customer service on a tablet a month after that are counted as three unique users in a standard Analytics implementation, even if all those actions took place while a user was signed in to an account. While you can collect data about each of those interactions and devices, you can’t determine their relevance to one another. You only see independent data points.

So, ultimately, you can't compare user counts from different reports.