Mega-Ultron-Super-Duper Connected Dataset to rule them all

BLUF: Should I join/append all of our data so we have one workable dataset?

This isn't a technical question, but rather theoretical. We've got data in many places (I'm sure this is common):

Budget in XYZ

Financials in ABC

Stats in DEF

Stats Historical (just switch systems so have a different datamart) in LMN

Everything that our end users do is taking pieces of each of these. I.e. How did we do this year (Stats) compared to last year (Stats Historical), and did that meet our expectations (Budget)? Is Revenue (Financials) in line with forcast (Budget)?

As it stands, we have to go to 4 different places to piece together the full story. My goal/hope/(maybe) pipe dream is to have all 4 sources of truth in one spot so these answers can be found simpler and more quickly.

What is the community's thoughts? Is this worth the time of development/implementation, or is it just "how things are"? Should I press into this as an initiative, or accept where we are at?

Any input is appreciated - I appreciate this community!

Best Answer

  • MichelleH
    MichelleH Coach
    Answer ✓

    @Data_Devon I would suggest going the append route for this data and adding a Category column to indicate the source of each row (Budget, Financials, etc.). Then you can use beast modes to present the output in a single card like this:

    Budget Revenue: sum(case when Category = 'Budget' then Revenue else 0 end)

    Actual Revenue: sum(case when Category = 'Financials' then Revenue else 0 end)

    Variance:

    sum(case when Category = 'Financials' then Revenue else 0 end) -

    sum(case when Category = 'Financials' then Revenue else 0 end)

    You should also check out @GrantSmith's writeup about how to do period-over-period analysis, which may be helpful for this type of data:

Answers

  • MichelleH
    MichelleH Coach
    Answer ✓

    @Data_Devon I would suggest going the append route for this data and adding a Category column to indicate the source of each row (Budget, Financials, etc.). Then you can use beast modes to present the output in a single card like this:

    Budget Revenue: sum(case when Category = 'Budget' then Revenue else 0 end)

    Actual Revenue: sum(case when Category = 'Financials' then Revenue else 0 end)

    Variance:

    sum(case when Category = 'Financials' then Revenue else 0 end) -

    sum(case when Category = 'Financials' then Revenue else 0 end)

    You should also check out @GrantSmith's writeup about how to do period-over-period analysis, which may be helpful for this type of data: