How to get an accurate aggregate average of a percentage on a line chart?

Sam1
Sam1 ⚪️
in Card Building

I have a line chart that shows a percentage across categories (opens/delivered graphed by email campaign). I would like to show which campaigns are above or below average with a line on the graph. Because of the different sizes of the email campaigns I cannot just average the ratios - instead I need to divide all the opens by all the deliveries for filtered data. It seems that the default average in Domo won't do this, but I was able to get the overall average to show as the summary number with this beast mode:

sum(`UniqueOpens`) / sum(`NumberDelivered`)

However I can't seem to get this to show as line on the graph. The blue line is what I want to show while the black line is the average of the ratios.

percentage average.png

I want the graphed average to change as the dataset is filtered by end users, so if they want to compare a subset of the campaigns they could see which ones are above/below average of the subset.


Thank you!

Best Answer

  • GrantSmith
    GrantSmith Indiana 🔴
    Accepted Answer

    Hi @Sam1

    Sounds like you want an average across the entire dataset to compare against. You can utilize a window function to get this and then compare your average to to it.

    Overall percentage:

    SUM(SUM(`UniqueOpens`)) OVER () / SUM(SUM(`NumberDelivered`)) OVER ()

    Putting it all together:

    CASE WHEN SUM(`UniqueOpens`)/SUM(`NumberDelivered`) > SUM(SUM(`UniqueOpens`)) OVER () / SUM(SUM(`NumberDelivered`)) OVER () THEN 'Above'
WHEN SUM(`UniqueOpens`)/SUM(`NumberDelivered`) < SUM(SUM(`UniqueOpens`)) OVER () / SUM(SUM(`NumberDelivered`)) OVER () THEN 'Below'
ELSE 'Same'
END


Answers

  • GrantSmith
    GrantSmith Indiana 🔴

    One thing to note, window functions aren't available out of the box, you need to talk with your CSM to get them enabled in your instance.

  • jaeW_at_Onyx
    jaeW_at_Onyx Budapest / Portland, OR 🟤

    careful... unique opens cannot be aggregated over time. that's not really the definition of unique. I think i would push back to the business on this one.

    If you measure Unique opens daily. and i open the email 5 times on separate days in the week. do i count once or five times?

Categories