3 Replies Latest reply on Dec 10, 2012 7:29 AM by Ekaterina K

    Making pie-charts, timecharts on customers groups: active/inactive, new/regular/react

    Ekaterina K

      Hi everyone,

      We are trying to reproduce  some key business metrics that we operate, to estimate convenience and suitableness of Tableau for our company's needs.


      It is unclear how to make the following and we ask for your kind assistance.

      I attach workbook with example data on the basis of fields purchasetime and userid.


      1. Divide customers into Active/Inactive groups, count them and make a pie-chart.

      Active customer - is that who made a purchase in the latest month. Inactive - respectively had no purchases in the latest month.

      What we tried:

      Create calculated field 'Active_Inactive': IF datediff('day', window_max(max([purchasetime])),now())>31 then 'Inactive' else 'Active' end

      It is created as a measure only, with no way to convert to dimension.

      Then I select pie-marks, drag Active_Inactive to Color field and CNT(userid) to Angle.

      The pie shows just one color.

      p.s. I saw the question http://community.tableau.com/message/192334#192334 but it still doesn't work for us as there can be more than one purchase and when you add 'max' function to purchasetime the evaluated field becomes a measure and pie doesn't work correct. 


      2. Make time distribution for active/inactive customers as columns by months.

      Like it is described here: http://community.tableau.com/thread/121961


      3. Divide customers into New/Regular/React groups, count them and make a pie-chart.

      New - are those who made their first purchase in latest 2 weeks.

      Regular - are those who made a purchase in the latest month and in the previous month.

      React - are those who made a purchase in the period (eternity;2) months ago & had no purchases in the period [2;1] months ago & had a purchase in the latest month.


      Thanks in advance and hope to solve these issues as they are really crucial for our decision.


        • 1. Re: Making pie-charts, timecharts on customers groups: active/inactive, new/regular/react
          Joe Mako



          Attached is an example for each of your questions, there was some logic that I did not quite understand from your description, so I just took guesses. with more details, I am sure your exact desired logic can be computed in Tableau.


          The downside of the attached is that it requires substantial knowledge about Tableau, specifically table calculations in order to produce these results. Also because of densification, the stacked bars takes a few seconds to compute.


          If you have the ability to process your data prior to Tableau, assigning the status to each userid, then that lowers the bar, eliminating the need for complex table calculations, making is easy to analyze and produce visualizations from.


          Also there are a great deal of other routes to produce these same results, so if you have additional requirements or constraints, other routes may be better.


          As you are considering Tableau, I highly recommend http://www.tableausoftware.com/videos/zen to better understand what situations Tableau is a great a fit for, and you are welcome to contact me if you have other questions.

          • 2. Re: Making pie-charts, timecharts on customers groups: active/inactive, new/regular/react

            Hi Ekaterina,


            1st, Joe has given you an excellent example showing it can be done in Tableau, but it is likely a poor choice for applying all levels of complex logic like this for your data.  Chief amongst the reasons:


            -Speed to compute (one of Joe's views required 80 seconds to compute, with about 20k customers, not that large considering many retailers have millions or tens of millions),


            -Ease of maintenance since these calcs are opaque and don't automatically work in every view without understanding Tableau fairly well,




            If you have a data architect, DBA or experienced SAS programmer around, I would push these complex, repeating business rules to their world (analytic data prep) where they can apply these and many other rules at the database level, past the data warehouse.  This work would allow the analysts to rapidly consume this information with Tableau and explore it from many directions.  This is exactly the approach I have applied at many companies including Netflix, REI and while teaching on the Faculty of The American Marketing Association.


            Other examples of things that belong at the analytic data prep layer include customer segmentation, lifetime value estimation, propensity to buy based on an offer or category, etc.  This would hold true with other analytic and BI tools, not just Tableau. 


            I hope that helps you move forward to a successful solution for your situation.


            Best regards,

            Stephen McDaniel

            Director of Analytic Product Management

            Tableau Software

            1 of 1 people found this helpful
            • 3. Re: Making pie-charts, timecharts on customers groups: active/inactive, new/regular/react
              Ekaterina K

              Joe, Stephen, thank you very much for the real help. Such quick and accurate solving of customer's request shows the whole company as a trusted companion. We really appreciate this. All results and comments are now being discussed at our side.