Automated Forecast (Item Variable vs Period type filter)

  • 29 November 2023
  • 0 replies
  • 62 views

Userlevel 2
Badge +3

Hi Pigment Community,

I’m currently working on building a conversion rate forecast in pigment and wanted to hear your thoughts on my calendar setup. 

Context: In our forecast model we are interested only in the current year and previous year. Information older than that is not being included in our forecast, visually and mathematically. Thus, I wanted to have an automated rolling calendar on an annual basis. This way if we are in Oct 2023 my model will show Jan 2022 - Oct 2023, rather than Oct 2022 - Oct 2023.

I know that in pigment you can use item variables to easily update variables, such as year, for all views in a model. However, I have opted for a different solution. Instead I have tweaked the ‘Period type’ dimension to include a 3rd item called ‘Actual old’. I then filter my metrics on all data where period type does not equal ‘Actual old’. Additionally, my period type is decided by my switch date, which is automatically chosen based on the max date in my data transaction.

Current pros and cons I have thought of
Item Variable

  • easy to adjust
  • no automation
  • not sure how win rates/close rates will be affected

Tweaked Period Type Filter

  • all yearly updates will happen according to the switchover date
  • time consuming to change in the future
  • more difficult for outsiders to understand

I have attached screenshots of my model for clarity. Personally, both solutions seem to be similar, with Item Variables applying a top-down approach, while my solution applies a bottom up approach. 

I would love to know you thoughts on the topic and whether or not you have experimented with similar designs, as well as any issues you imagine I could face in the future with a design like this.  

Adjust period type dimension
Month dimension with ‘Period type trailing’ which has updating period type
switchover date based on ‘All Deals’ transaction block
‘Period type trailing’ user to filter metric calculation

Thank you for your feedback!


0 replies

Be the first to reply!

Reply