How to create a cohort analysis?

Simon

Hey, I would like to recreate a cohort analysis (see screenshot of excel below).

 

In the first column, we have the trailing past 30 days and in the first row the number of days that are added to the respective date until the customer is won.

 

Is there a way to replicate this in domo?

kshah008

Hi all,

Can anybody help @Simon out?

Thanks!

Godzilla

if you have the lead activation date and the age fields in your dataset, you might be able to do this in a Sumo card since it has the pivot capability. You might be able to do the conversion rate calc as a beast mode if you have something in the dataset identifying which are converted leads.

kshah008

@Simon, did Godzilla's reply help you out?

JeffNi01

Bump. Tried using a Sumo table for this and I don't think it worked. 

 

Anyone else face a similar issue? Would love to see if there's a way for me to track multiple cohorts over time with a running total.

Lisa-Chang

 

not exactly in a table form but this is what we've created as cohort. Yoou can toggle the metrics into table form. What we've done is create a beast mode for every intervals, eg 30-60-90 days. hope this will give you some ideas. 

 

 

 

parththakkar

I have faced this exact issue as well and I ahven't been able to figure it out! Anybody make any progress with this?

user06521

Anyone got an idea about this? Facing the same issue.

[Deleted User]

@AS @Valiant @ST_-Superman-_

 

Thoughts from our power members?

Thanks!

ST_-Superman-_

I have run into similar issues as this, although I have not tried to build a cohort.  I'm referring to requests to recreate a specific report in Domo.  I would tackle this by first asking what information the user wants to get from this data?  What business question is being asked and how do you measure success?  While Domo may not have a great way to recreate this exact visual, in my experience there are tools available to answer the business question.  Often, you can answer the question in a much more intuitive way.  

 

As for recreating this report, I would want to take a closer look at some data to see how it is structured.

Valiant

We've done something similar, but actually use R to create the analysis and use a heatmap card to display the results after we push the numbers back into Domo. 

 

Hoping that the upcoming R ETL functionality will allow us to streamline this.

 

Here's a guide to something like what we did in R:

https://www.r-bloggers.com/measuring-user-retention-using-cohort-analysis-with-r/

user061076

I am facing the same issue. Can't figure out how to calculate subtotals for cohort users and averages for each retention time period. I am not good with R, I will try ti sign up for the Beta jupyter notebook to push the results directly into the heat map card. I don't believe the averages will show, though.

jaeW_at_Onyx

try creating a dataset view.  you may need to have two.

 

i don't understand whether you need number of days since a point in time, OR number of days since a value in a column.

 

either way, you can structure this data if you take your transactional data and join it to a date dimension table.  You can do that in a dataset view or an ETL.  you can calculate averages using a second dataset view, or in analyzer.

 

i don't recommend Jupyter Notebook even though it's fun b/c it's not intended to be part of a data pipeline

 

this sounds like a fun project, if you post more details and a sample dataset i'll take a closer look.