Capacity Forecast

Capacity Forecast(CF) chart aims to predict the near future based on the continuing condition of the recent past. Using the historical Throughput data, the CF analytic forecasts the best fit completion rates if a given number of additional cards are pulled onto the board. This forecasting can help you adjust the capacity allocation in your kanban system to pull work at a steady pace throughout the lifetime of the board. 

Filtering the Data Set

  1. Navigate to Board> ESP Module, and then select the Smart Lane and enter the start and end date for which you want to plot the chart.
  2. The chart is plotted based on the card count as Unit of Measurement (UoM) in the CF analytics page. 
  3. You can perform any of these activities to further refine your search and narrow down the scope of the chart:
    • Select any of these attributes: like Item Type, Priority, Size, or Class of Service to plot the chart. You can further refine your filter by selecting particular attribute values. For example, if you have selected ItemType as filtering attribute , then you can further select Change Request and Defect to narrow down the scope of the chart.
    • You can also define the Start and End lane to further filter your data on the chart.
      As you apply any of these filters, the chart will be refreshed and rendered automatically. 
  4. While calculating CF, the chart takes into account all the cards ranging between the start and end lane that you specify in the analytics page, which might include:
  • Cards entered or exited the Done Lane Type
  • Cards archived from any lane beyond (to the right) the Done Lane Type

Reading the Diagram


  1. In the CF analytics page, the CFD chart is placed at the extreme below. You can specify the date range in this chart by dragging the slider. In the image 1, the date range selected is 27 Oct to 27 Nov.
  2. As you specify the date range in image 1, the detailed Capacity Forecasting chart is rendered ( image 2) where you can view:
    • The flow of the actual throughput during the selected date range (at the interval of 2 days)
    • The predictive flow of throughput in the near future based on the continuing condition of the actual throughput. So, each overlapping line in the predictive section depicts the predictability of completing certain number of additional cards at a current velocity. So, in image 2, pulling 20 additional cards into the Kanban system have higher chance to be completed than pulling 40 cards or more.
      The same prediction has been plotted in the Frequency Distribution chart shown in image 4.  
  3. The column chart shows the relation between completion days (X-axis) and number of cards (Y-axis). For example, in the above chart, it shows that average completion days for 0-1 card is 21 days, 1-2 cards are 2 days and 2-3 cards are 1 day.  So, it is evident that lesser cards have consumed more time (<20 days) and 1-3 cards have consumed lesser than 4 days.
  4. The frequency distribution chart shows the relation between number of additional cards and probability percentage of completing those cards.
    For example, in image 4, the distribution chart predicts 56% chance of completing additional 4 cards if they are pulled into the system at the current throughput rate rendered in image 2. So, according to this chart, pulling lesser cards have higher chance of getting completed as per the schedule.

Interpreting the Diagram

A probabilistic forecast can help us make informed decision about pulling work into the system and adjusting the WIP limit. Using historical data for a particular date range can help an organization build a reliable, high quality forecast, efficiently mitigate project risk. By analyzing the predictive data, one can make incremental improvements in the capacity planning.