Forum Discussion

Drors's avatar
Drors
Icon for Resolver III rankResolver III
8 years ago
Solved

Funnels in Power BI

Hi All,

 

Im not sure how to display that kind of information in power bi and if powerbi have relevent charts to that kind of information.

 

I have a data of users and 5 actions they can do in any order they want with the datetime they actually did that

it looks like

 

User | ActionA | ActionB | ActionC | Action D

111    26/8/18   25/8/18   22/8/18   21/8/18

112    21/8/18   22/8/18   24/8/18   23/8/18

...

 

I want to create a funnel chart that will show me the percents of users go from A to B to C to D

and from

A to B to D to C

A > C > B > D

and so on..

 

anyone have good advice?

or some other idea how i can display these usfull informaiton?

 

Google Analytics have butiful chart (I think it call snake chart) exactly to that kind of information but i want to create it in my organization dashboards

 

Thanks

Need Help
Tips and Tricks
  • v-yulgu-msft's avatar
    v-yulgu-msft
    8 years ago

    Hi Drors,

     

    First, unpivot source table in Query Editor mode. And rename columns.

     

    In data view, add calculated columns in data table.

    Rank =
RANKX (
    FILTER ( Dataset1, Dataset1[User] = EARLIER ( Dataset1[User] ) ),
    Dataset1[Date],
    ,
    ASC,
    DENSE
)

Action Order =
RIGHT (
    CALCULATE (
        LASTNONBLANK ( Dataset1[Action Type], 1 ),
        FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 1 )
    ),
    1
)
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 2 )
        ),
        1
    )
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 3 )
        ),
        1
    )
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 4 )
        ),
        1
    )

     

    New a calculated table.

    Table_1 =
ADDCOLUMNS (
    FILTER (
        CROSSJOIN (
            CROSSJOIN (
                CROSSJOIN (
                    VALUES ( Dataset1[Action Type] ),
                    SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType2", [Action Type] )
                ),
                SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType3", [Action Type] )
            ),
            SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType4", [Action Type] )
        ),
        [Action Type] <> [ActionType2]
            && [Action Type] <> [ActionType3]
            && [Action Type] <> [ActionType4]
            && [ActionType2] <> [ActionType3]
            && [ActionType2] <> [ActionType4]
            && [ActionType3] <> [ActionType4]
    ),
    "Action Order", RIGHT ( [Action Type], 1 ) & "-"
        & RIGHT ( [ActionType2], 1 )
        & "-"
        & RIGHT ( [ActionType3], 1 )
        & "-"
        & RIGHT ( [ActionType4], 1 )
)

     

    Establish a one to many relationship between above two tables.

     

    Place 'Table_1'[Action Order] and 'Dataset1'[User] into Funnel chart.

     

    Best regards,

    Yuliana Gu

3 Replies

  • v-yulgu-msft's avatar
    v-yulgu-msft
    Icon for Microsoft Employee rankMicrosoft Employee
    8 years ago

    Hi Drors,

     

    First, unpivot source table in Query Editor mode. And rename columns.

     

    In data view, add calculated columns in data table.

    Rank =
RANKX (
    FILTER ( Dataset1, Dataset1[User] = EARLIER ( Dataset1[User] ) ),
    Dataset1[Date],
    ,
    ASC,
    DENSE
)

Action Order =
RIGHT (
    CALCULATE (
        LASTNONBLANK ( Dataset1[Action Type], 1 ),
        FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 1 )
    ),
    1
)
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 2 )
        ),
        1
    )
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 3 )
        ),
        1
    )
    & "-"
    & RIGHT (
        CALCULATE (
            LASTNONBLANK ( Dataset1[Action Type], 1 ),
            FILTER ( ALLEXCEPT ( Dataset1, Dataset1[User] ), [Rank] = 4 )
        ),
        1
    )

     

    New a calculated table.

    Table_1 =
ADDCOLUMNS (
    FILTER (
        CROSSJOIN (
            CROSSJOIN (
                CROSSJOIN (
                    VALUES ( Dataset1[Action Type] ),
                    SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType2", [Action Type] )
                ),
                SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType3", [Action Type] )
            ),
            SELECTCOLUMNS ( VALUES ( Dataset1[Action Type] ), "ActionType4", [Action Type] )
        ),
        [Action Type] <> [ActionType2]
            && [Action Type] <> [ActionType3]
            && [Action Type] <> [ActionType4]
            && [ActionType2] <> [ActionType3]
            && [ActionType2] <> [ActionType4]
            && [ActionType3] <> [ActionType4]
    ),
    "Action Order", RIGHT ( [Action Type], 1 ) & "-"
        & RIGHT ( [ActionType2], 1 )
        & "-"
        & RIGHT ( [ActionType3], 1 )
        & "-"
        & RIGHT ( [ActionType4], 1 )
)

     

    Establish a one to many relationship between above two tables.

     

    Place 'Table_1'[Action Order] and 'Dataset1'[User] into Funnel chart.

     

    Best regards,

    Yuliana Gu

    • Drors's avatar
      Drors
      Icon for Resolver III rankResolver III
      8 years ago

      v-yulgu-msft

      Wow its looks amaizing, thank you.

       

      I have another question about that, If some of my clients just made 3 actions out of 4, for example

      A>B>C  (didnt do D)

      and I want to see the discrepencies ( in order to know which action is more complicate and the users dont want to do it)

      what do you recommend me to do?

      • v-yulgu-msft's avatar
        v-yulgu-msft
        Icon for Microsoft Employee rankMicrosoft Employee
        8 years ago

        Hi Drors,

         

        It looks like this is a new requirement different from the original post, right? If so, I would suggest you create a new thread on forum so that more community members can see it and provide advice. Please remember to post dummy data and desired result.

         

        By the way, if my above advice is helpful to your scenario, would you please kindly mark it as an answer?

         

        Thanks,

        Yuliana Gu