3 Replies Latest reply on May 29, 2012 5:00 AM by Shawn Wallwork

# Retention computations through filtering or RAWSQL?

I am looking at time series of the number of users of a game in different countries divided by registration date and the date they played the game in Tableau.

Let's say I want to know how large a percentage of the users registered on a specific date (reg_date) are still playing on another date (play_date). The easy way to do that is just to make a plot where I use reg_date as a filter, and view the number of users along the play_dimension.

Now, I have created another calculated field which is the number of days a user has played the game (time_played = play_date-reg_date), and I would like to see how many users remain in the game after, say, a week, or a month, i.e #users(time_played=30)/#users(time_played=0). I have not found any way to implement this type of function in Tableau. I can view the number of users from time 0 to N as a line, but that is as far as I am getting.

• ###### 1. Re: Retention computations through filtering or RAWSQL?

Carl, post a sample packaged workbook and we'll take a look.

--Shawn

Follow Forum Folk: I'm not 'checking-out' this problem until I see the workbook, so feel free to jump in.

• ###### 2. Re: Retention computations through filtering or RAWSQL?

Here is a sample. This graph shows what fraction of users remain at time t in a line graph, as a table calculation. As it is made as a table calculation, I can not just filter it to look at time X, as it would just show 100% then.

What I would like to plot is the fraction of users that remain at time X (ideally a parameter, but can be a constant) as a bar plot, for the different countries and regions, i.e., only a single value per country or region (or whatever level i am looking at the results at).

• ###### 3. Re: Retention computations through filtering or RAWSQL?

Hi Carl, you're going to need to post a packaged workbook for us to take a look. Thanks,

--Shawn