Suppose there is a typical desktop RTS game without online multiplayer on a Windows platform, like the original Warcraft: Orcs & Humans. It still connects to a server to log basic anonymous info about the player's sessions. Now it would be interesting to see customer (player) retention (and other stats) in an industry-accepted way.

What are some typical retention (and other stats) available to game developers, relevant to offline desktop games in general?

Unlike other games, RTS implies longer gameplay sessions (40min+). How does it affect retention calculation? What are some useful retention metrics (and other stats) in this case?

How to calculate above retention parameters, what statistics values are needed to do it?


2 Answers 2


In a studio where I worked we use to track different type of retention: Day 1, Day 3, Day 5. This metric was calculated saving a UID of the user that started a game session for the first time. Each new log-in in the game, with that UID we could see after the first day how much UID were back. Then again after 3 day and again. This was the retention.

You can tweak you time span according to what do you think your critique part will be.

You can also record the mission that a player starts, in order to find some "holes" in your missions.

Another very useful metrics is the play session length that can help you to tweak your game experience around the session.

Metrics on the object of your game, to understand which are more used and which are not.

  • \$\begingroup\$ Could you please add more details about how did you calculate the retention for 1-3-5-N days? \$\endgroup\$
    – Kromster
    Commented Feb 23, 2017 at 10:49
  • \$\begingroup\$ Updated the answer. \$\endgroup\$
    – LaboPie
    Commented Feb 23, 2017 at 11:07

If the game is truly offline (no hidden connection to a server) then there aren't really any good "retention" statistics you can reliably track that aren't simply being estimated based on sales data and by monitoring community activity (forums, support, etc).

However, if you assume some kind of online account or client (IE: Steam) then getting retention metrics becomes trivial because Steam stores all that handy information for you in order to display it to the player.

The important information as a developer/company would be:

  • Weekly, monthly and quarterly sales
  • Daily, weekly and monthly logins
  • Average session length
  • Number of support requests
  • Average time to resolution of support requests

Those would be what I'd be interested in for mostly obvious reasons... sales tells you how well your game is doing in concrete numbers, logins tells you how popular your game is, session length tells you how engaging the game is, support tickets tell you about the health of the game, and resolution reflects the quality of your support.

Keep in mind that watching those numbers relative to patches, competition within (and outside) of your genre, and other factors over time will help you assess the direction you need to take the game or future releases.

For example, if you are selling 10k games per week and have 100k average logins with an average playtime of 45min per session and suddenly that doubles (or cuts in half) after a new patch then that gives you critical data moving forward about how the player-base feels in a way that community and feedback can't measure.

  • \$\begingroup\$ As said in the question^ "It still connects to a server to log basic anonymous info about the player's sessions" \$\endgroup\$
    – Kromster
    Commented Feb 23, 2017 at 16:01
  • 1
    \$\begingroup\$ I know, I just wanted to make it clear that offline games such as Warcraft never connected to a server so those retention analytics wouldn't have existed. If you just mean that "type" of game in a modern setting then that's fine but I didn't want there to be any confusion. I would say that any offline game that doesn't require an internet connection to play shouldn't be sending any information over the network at all. That would be a major red flag to me for invasion of privacy and I wouldn't recommend it. \$\endgroup\$
    – Aithos
    Commented Feb 23, 2017 at 16:21

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .