Forum Discussion

Anonymous's avatar
Anonymous
Not applicable
4 years ago
Solved

Customer Repurchase Cycle

Hi Community and PowerBi Masters,   I am currently working on an analysis for which I try to understand the consumer behavior in detail.  The corresponding data that I have is (name of table: All_...
  • colacan's avatar
    colacan
    4 years ago

    Anonymous  Hi, considering your issues I made measures as below. I hope this helps you.

     

    //[NumOfTotPurchase]: This counts all purchases even if order numbers are the same. e.g. if Emil purchased 4 macbooks on the same day under same oerder number, it still counts 4.

    NumOfTotPurchase = countrows(All_Platforms)   

    //[NumOfVisit]:  This counts only number of visit regardless number of purchases e.g. If Emil purchased 4 macbooks, if counts only 1 as long as all purchases are done under same order number
     
    NumOfVisit =
    VAR currentuser = MAX( All_Platforms[User Name] )
    VAR NumOfVisit =
    COUNTROWS(
    DISTINCT(
    SELECTCOLUMNS(
    FILTER( All_Platforms, All_Platforms[User Name] = currentuser ),
    "OrderNum", All_Platforms[Order Number],
    "Name", All_Platforms[User Name]
    )
    )
    )
    Return NumOfVisit
     
    //[NumOfReVisit]:  This is the same as above [NumOfVisit] but it excludes the first visit. hence the formular is simply [NumOfVisit]-1
     
    NumOfReVisit = [NumOfVisit]-1
     
    //[AVG_Time_Revisit_Indv]:  This is similar to the measure [AverageTime_Revisit] which I provided previously but it is revised to ignore bulk purchase under the same order number. But this measure cannot be used universal purpose(e.g. KPI card) since it needs specific "Name".  Just make this measure in your model but don't use this. this measure will be used internally to make the final new measure
     
    AVG_Time_Revisit_Indv =
    VAR currentuser = MAX( All_Platforms[User Name] )
    VAR NumberOfDates =
    CALCULATE(
    DATEDIFF(
    FIRSTNONBLANK( All_Platforms[OrderTime], All_Platforms[OrderTime] ),
    LASTNONBLANK( All_Platforms[OrderTime], All_Platforms[OrderTime] ),
    DAY
    ),
    All_Platforms[User Name] = currentuser
    )
    VAR NumberOfRePurcases =
    COUNTROWS(
    DISTINCT(
    SELECTCOLUMNS(
    FILTER( All_Platforms, All_Platforms[User Name] = currentuser ),
    "OrderNum", All_Platforms[Order Number],
    "Name", All_Platforms[User Name]
    )
    )
    ) - 1
    VAR Avg_Time_Interval =
    CALCULATE(
    DIVIDE( NumberOfDates, NumberOfRePurcases ),
    ALLSELECTED( all_Platforms )
    )
    Return Avg_Time_Interval
     
    //[Avg_Revisit_Interval]:  This is the one you can use to show revisit interval for individual as well as overall time interval (for KPI card)
     
    Avg_Revisit_Interval =
    AVERAGEX(
    ADDCOLUMNS(
    VALUES( All_Platforms[User Name] ),
    "AvgTime", [AVG_Time_Revisit_Indv] ),
    [AVG_Time_Revisit_Indv]
    )
     
    //[Time_till_next_visit ]:  This is the measure you can use along with speciifc date which shows the number of dates till next visit. e.g. if you select the date when Emil visited, this measure will show the number of dates till Emil's next visit
     
    Time_till_next_visit =
    VAR current_date = max(All_Platforms[OrderTime])
    VAR Next_visit =
        CALCULATE(
            min(All_Platforms[OrderTime]),
            All_Platforms[OrderTime] > current_date)
    RETURN DATEDIFF(current_date,Next_visit,DAY)
     
    //[Avg_Revist_Time_From]:  This is similar to above [Time_till_next_visit ] which you can use along with speciifc date. While [Time_till_next_visit ] shows the number of dates till next visit, this measure show average number of dates for revisit calculating till last visit date. You need to create the measure [NumOfVisit] first to create this measure
     
    Avg_Revist_Time_From =
    VAR current_date = max(All_Platforms[OrderTime])
    VAR Fianl_visit =
        CALCULATE(
            MAX(All_Platforms[OrderTime]),
            All_Platforms[OrderTime] > current_date)
    VAR DateDifference = DATEDIFF(current_date,Fianl_visit,DAY)
    VAR Num_visit =
        CALCULATE(
            [NumOfVisit],
            All_Platforms[OrderTime] > current_date)
    RETURN DIVIDE( DateDifference,Num_visit)
     
     
    Please mark this as solution if this helped you.
colacan's avatar
colacan
Resolver II
4 years ago

Anonymous  Hi, Jannnn04

 

1. I was creating columns instead of measures before (why is that so?).

Calculate column evaluates each expression under row context. and filter context does not exist in calculate column.

COUNTROWS function, like SUM( ) or AVERAGE( ), conducts it's calculation under Filter context.  since there is no filter context in calculate column, it calculates over entire table. that's why you got total number.

But measure is different. both CALCULATE and CALCULATETALBE invoke Context Transition which means, these two function change Row context to Filter context (in other words, context transition makes the 1 row as it's filter). one important thing is that all measures are wrapped by CALCULATE function inside the engine regardless you used calculate function for your measure or not. Therefore, if you create a measure, it will colcuate for the current row only.

Having say that, to create a column with the same result, you can simply wap the expression with CALCULATE as below.

 

Measure: NumberOfPurchase = countrows(All_Platforms)

Calculated column: NumberOfPurchase = CALCULATE(countrows(All_Platforms))

 

1. Average time with specific time:

As I expalined previously, to calculate average time, you need at lease two transactions because the average time is the Time Gap between 2 transactions.

If you want to show the earler time and the average time in the same row in the report, you need to create a table for it.

 

2. I can not depict the average repurchase time in a KPI card, in a gauge or in a bar chart

The measure (average time) needs customer as it's filter context because we cannot define re-purchase without "WHO" but the gauge does not have customer filter unless you select specific "customer" from outter filter. (e.g. selecting "Emil" with slicer).

If you want to show overall "avg. revisiting time" regardless customer (which means average of revisiting time for all customers who purchased item at least twice), then it will be another measure which can be combined with eailer one (the indivisual average time).

In that case, the 2 measures can be combined as below:

  IF indivisual is selected, return the value from eailer measure,

  else return value from new measure.

(By the way, those customer who purchase only once shoule be excluded for the new measure.)

 

I hope you can handle your issues with above guideline.  I may make the code when I find time but not sure when. 

so If you are looking for the solution urgently, please find help from others.

Thanks.

Anonymous's avatar
Anonymous
Not applicable
4 years ago

Hello colacan,

 

wow, thanks a lot, again. The explanations are really, really helpful! 🙂

I tried to create the code myself but so far I got stuck with both issue 1 and 2. Would be great if you can help me with that again. Really brilliant from you, thanks!

 

 

I am super sorry to further bother but I discovered another problem.

Issue

I now edited the data and have for one specific day, with one order number, many purchases from one customer (emil) for one product (macbook in category laptop). That decreased the the average time between purchases from that user. However, the real average time between purchases has not changed. He just purchased more items within one order. 

How can I change the code to treat a single, large order (with one order number) as a single transaction?

 

Obviosuly also happy if anyone else would want to assist colacan with his great work so far:)

Thanks a lot and best regards!

Jan

 

  • Anonymous's avatar
    Anonymous
    Not applicable
    4 years ago

    I still did not manage to solve that issue and would be super grateful if someone can assist! 🙂 

    Thanks, community!

    • colacan's avatar
      colacan
      Resolver II
      4 years ago

      Anonymous  Hi, considering your issues I made measures as below. I hope this helps you.

       

      //[NumOfTotPurchase]: This counts all purchases even if order numbers are the same. e.g. if Emil purchased 4 macbooks on the same day under same oerder number, it still counts 4.

      NumOfTotPurchase = countrows(All_Platforms)   

      //[NumOfVisit]:  This counts only number of visit regardless number of purchases e.g. If Emil purchased 4 macbooks, if counts only 1 as long as all purchases are done under same order number
       
      NumOfVisit =
      VAR currentuser = MAX( All_Platforms[User Name] )
      VAR NumOfVisit =
      COUNTROWS(
      DISTINCT(
      SELECTCOLUMNS(
      FILTER( All_Platforms, All_Platforms[User Name] = currentuser ),
      "OrderNum", All_Platforms[Order Number],
      "Name", All_Platforms[User Name]
      )
      )
      )
      Return NumOfVisit
       
      //[NumOfReVisit]:  This is the same as above [NumOfVisit] but it excludes the first visit. hence the formular is simply [NumOfVisit]-1
       
      NumOfReVisit = [NumOfVisit]-1
       
      //[AVG_Time_Revisit_Indv]:  This is similar to the measure [AverageTime_Revisit] which I provided previously but it is revised to ignore bulk purchase under the same order number. But this measure cannot be used universal purpose(e.g. KPI card) since it needs specific "Name".  Just make this measure in your model but don't use this. this measure will be used internally to make the final new measure
       
      AVG_Time_Revisit_Indv =
      VAR currentuser = MAX( All_Platforms[User Name] )
      VAR NumberOfDates =
      CALCULATE(
      DATEDIFF(
      FIRSTNONBLANK( All_Platforms[OrderTime], All_Platforms[OrderTime] ),
      LASTNONBLANK( All_Platforms[OrderTime], All_Platforms[OrderTime] ),
      DAY
      ),
      All_Platforms[User Name] = currentuser
      )
      VAR NumberOfRePurcases =
      COUNTROWS(
      DISTINCT(
      SELECTCOLUMNS(
      FILTER( All_Platforms, All_Platforms[User Name] = currentuser ),
      "OrderNum", All_Platforms[Order Number],
      "Name", All_Platforms[User Name]
      )
      )
      ) - 1
      VAR Avg_Time_Interval =
      CALCULATE(
      DIVIDE( NumberOfDates, NumberOfRePurcases ),
      ALLSELECTED( all_Platforms )
      )
      Return Avg_Time_Interval
       
      //[Avg_Revisit_Interval]:  This is the one you can use to show revisit interval for individual as well as overall time interval (for KPI card)
       
      Avg_Revisit_Interval =
      AVERAGEX(
      ADDCOLUMNS(
      VALUES( All_Platforms[User Name] ),
      "AvgTime", [AVG_Time_Revisit_Indv] ),
      [AVG_Time_Revisit_Indv]
      )
       
      //[Time_till_next_visit ]:  This is the measure you can use along with speciifc date which shows the number of dates till next visit. e.g. if you select the date when Emil visited, this measure will show the number of dates till Emil's next visit
       
      Time_till_next_visit =
      VAR current_date = max(All_Platforms[OrderTime])
      VAR Next_visit =
          CALCULATE(
              min(All_Platforms[OrderTime]),
              All_Platforms[OrderTime] > current_date)
      RETURN DATEDIFF(current_date,Next_visit,DAY)
       
      //[Avg_Revist_Time_From]:  This is similar to above [Time_till_next_visit ] which you can use along with speciifc date. While [Time_till_next_visit ] shows the number of dates till next visit, this measure show average number of dates for revisit calculating till last visit date. You need to create the measure [NumOfVisit] first to create this measure
       
      Avg_Revist_Time_From =
      VAR current_date = max(All_Platforms[OrderTime])
      VAR Fianl_visit =
          CALCULATE(
              MAX(All_Platforms[OrderTime]),
              All_Platforms[OrderTime] > current_date)
      VAR DateDifference = DATEDIFF(current_date,Fianl_visit,DAY)
      VAR Num_visit =
          CALCULATE(
              [NumOfVisit],
              All_Platforms[OrderTime] > current_date)
      RETURN DIVIDE( DateDifference,Num_visit)
       
       
      Please mark this as solution if this helped you.