hello, can i check with you why even though i followed all your steps, for the last step when your graph splits into the different weeks, mine does a cumulative value instead. so instead of 100% as the starting point for week 0 for every cohort, the week 0 is split by proportion for all the cohorts to = to 100%
Hey Jessica. I think you're using the percentage of total table calculation instead of dividing by window_max. You want to use the window max function to divide the number of users that are left at each point by the total number you started with. I hope that helps. Thanks for watching.
is there any way to select a cohort automatically for every conversion? I mean some way to represent correct retention for every recur without selecting right cohorts manually
Hey Kateryna. Yes, you could do this by working backwards and starting with a retention curve which includes all available data, and then segmenting the lines by the relevant cohort grouping (signup month for example). This will result in a multi-line graph where each line represents a cohort. The issue is that such views can be difficult to read sometimes, and may create confusion since not all cohorts will be equal in age.
@@devanshbhardwaj5955 To build such a cohort table you need to create a calculated field which uses the datediff and IF functions in Tableau. You create custom buckets of X days (between 0 and 6 equals 1 would be the first bucket of 7 days, 7 - 13 would be the second bucket etc) and then set the calculated field as a dimension. When you place the field on the columns you will get columns for each bucket like shown in my video. You then need a date field for the rows, and then can use either countd for absolute values, or countd/window_max to calculate percentages.
@@stayriddi5094 The formula I used to create the columns in the table is made up of datediff and fixed. Calculated field #1: {fixed user_id: min(event_created_at)} --> This gives you the first event performed by the user. Calculated field #2: datediff('week',CALCULATED FIELD #1,event_created_at) --> This gives you the number of weeks between the first event and the current event. You'll then turn this into a dimension and add it to the columns shelf. You then use countd(user_id)/window_max(countd(user_id) for your values. This will give you the percentage of the users which did activity in the given cell, compared to the total # of users. I'll just add that this is a tough concept to grasp and you're better off trying to understand the approach first before trying to build it. Try it first on paper or in excel and once you grasp it, building it in Tableau isn't too difficult. Good luck. -Justin
hello, can i check with you why even though i followed all your steps, for the last step when your graph splits into the different weeks, mine does a cumulative value instead. so instead of 100% as the starting point for week 0 for every cohort, the week 0 is split by proportion for all the cohorts to = to 100%
Hey Jessica. I think you're using the percentage of total table calculation instead of dividing by window_max.
You want to use the window max function to divide the number of users that are left at each point by the total number you started with.
I hope that helps. Thanks for watching.
is there any way to select a cohort automatically for every conversion? I mean some way to represent correct retention for every recur without selecting right cohorts manually
Hey Kateryna.
Yes, you could do this by working backwards and starting with a retention curve which includes all available data, and then segmenting the lines by the relevant cohort grouping (signup month for example).
This will result in a multi-line graph where each line represents a cohort.
The issue is that such views can be difficult to read sometimes, and may create confusion since not all cohorts will be equal in age.
can you please tell how you made that retention graph of starting
cohort i mean
@@devanshbhardwaj5955 To build such a cohort table you need to create a calculated field which uses the datediff and IF functions in Tableau.
You create custom buckets of X days (between 0 and 6 equals 1 would be the first bucket of 7 days, 7 - 13 would be the second bucket etc) and then set the calculated field as a dimension. When you place the field on the columns you will get columns for each bucket like shown in my video.
You then need a date field for the rows, and then can use either countd for absolute values, or countd/window_max to calculate percentages.
@@theanalystsplaybook Would it be possible to be more specific on the exact formula you used?
@@stayriddi5094 The formula I used to create the columns in the table is made up of datediff and fixed.
Calculated field #1: {fixed user_id: min(event_created_at)} --> This gives you the first event performed by the user.
Calculated field #2: datediff('week',CALCULATED FIELD #1,event_created_at) --> This gives you the number of weeks between the first event and the current event. You'll then turn this into a dimension and add it to the columns shelf.
You then use countd(user_id)/window_max(countd(user_id) for your values. This will give you the percentage of the users which did activity in the given cell, compared to the total # of users.
I'll just add that this is a tough concept to grasp and you're better off trying to understand the approach first before trying to build it. Try it first on paper or in excel and once you grasp it, building it in Tableau isn't too difficult.
Good luck. -Justin
i need yo help like rn 1
yo bro please reply man