Four worked examples show how a dashboard can give a plausible number and still miss the business question. Use these checks before you make a decision with the numbers.
A join can multiply a total
Weekly order revenue jumps after someone adds line items to the report. One order appears twice because it has two items. Both joined rows carry its full order total.
Check the source
Find where the amount is recorded and what one row represents. Compare the joined amount with the total from the orders table, where each order ID appears once.
Rows after joining orders to line items
Order
Line item
Order total
A101
A101-1
$120
A101
A101-2
$120
A102
A102-1
$80
Sum of joined rows
$320
$120 + $120 + $80
Sum once per order ID
$200
$120 + $80
Ask DewPoint Check completed order revenue at one row per order. Compare it with the chart total and explain any difference introduced by joining to line items.
A team-wide rate needs team-wide counts
The operations review asks what share of all qualifying calls met the target. Taking the average of the two queues’ percentages gives each queue equal weight, even though one handled ten times as many calls.
Check the source
Use the same eligibility and target rules for every queue. Sum the calls that met the target and divide by all qualifying calls.
Calls meeting the response target
Queue
Met target
Qualifying
Rate
General
90
100
90%
Specialist
1
10
10%
Average queue rate
50%
(90% + 10%) ÷ 2
Share of qualifying calls
82.7%
(90 + 1) ÷ (100 + 10)
Ask DewPoint Show the overall share of qualifying calls that met the response target. Include the numerator and denominator beside each queue’s rate and verify the overall result against the source counts.
Zero percent needs eligible events
Three teams display 0% late shipments. That value describes a team with eligible shipments and zero late ones. A team with no eligible shipments has an undefined rate. A team with missing shipment data needs a coverage check.
Check the source
Show the eligible count next to the rate. Check whether an empty feed is expected before interpreting an empty denominator, and suppress the percentage when the count is zero or missing.
Late deliveries by operating site
Site
Late
Eligible
North
0
12
South
0
0
West
Missing
Missing
Blank rates coerced to zero
0% each
Different source conditions look identical.
Rates with coverage
0% in North
South has no eligible shipments. West has missing data.
Ask DewPoint Show late deliveries and eligible deliveries by site. Show the rate when the eligible count is positive and flag sites with zero eligible deliveries or missing coverage.
A partial week needs a fair comparison
Wednesday’s report says ticket volume fell 40%. It compares three days of intake with the previous week’s five working days. Comparing Monday through Wednesday for both weeks shows volume rose.
Check the source
Use the event timestamp, the reporting time zone and two equivalent windows. Confirm both windows are complete through their cutoff.
Tickets opened, measured in the same time zone
Window
Tickets
Previous Mon–Fri
200
Previous Mon–Wed
100
Current Mon–Wed
120
Against five days
−40%
(120 − 200) ÷ 200
Against the same days
+20%
(120 − 100) ÷ 100
Ask DewPoint Compare tickets opened Monday through Wednesday with the same days last week. Use our reporting time zone, confirm the latest day is complete and show both counts with the percentage change.
Follow a figure back to the data.
Explore a full dashboard question or install the skill and check a metric in a workspace of your own.