For this simple example, let's assume the audience is a government water use planning group for the State of Colorado. The group wants to understand how many weather stations are reporting precipitation information and a sense of how well the information is being captured. The following steps show how this can be made into a question tree:

• Step 1: List out questions the audience wants to ask of the analytics:
  • Did the number of stations reporting precipitation change from last month?
  • How many stations in total have sent precipitation numbers?
  • Where are the stations located?
  • How many have reported a significant rainfall?
  • Is there anything weird in the daily sums for each station that might indicate a problem?
  • Which areas of the state had some good rain last month?
• Step 2: Generalize the questions and consolidate:
  • What is the trend of stations reporting usable precipitation data?
  • How many stations report data?
  • Where are the stations that reported usable data in the period?
  • Did a station report significant rainfall in the period?
  • Are there abnormal precipitation values for a specific station?
  • Where are the stations that reported usable data in the period? (consolidate with the similar question earlier in the list)
• Step 3: Identify starting point questions:
  • Based on conversations with the water planning group, you determine their first thought is to simply check the number of stations that report 15-minute precipitation data.
  • Starting point question: How many stations report data?
• Step 4: Organize the questions into a question tree diagram: