Examples
Data quality examples
Projects where the finding was in how the data was collected: duplicate records, abandoned surveys, and one headline rate that was three events added together.
- In this subject
- 6 of 11
- Subject
- Data quality
- Every one opens
- in full
Featured project
Customers · 1,857 rows
The apparent 0.42-point gap fell to 0.07 once partial responses were included.
Customer survey: Q2 · 2 files, opened in full: the finding, the number behind it, and the trace.
Is the difference between customer groups real?Responses
1,840
Completion rate
71.2%
Significant effects
2
Rows analysed
1,857
Money · 441 rows
Contractor spend rose $412,000 while headcount spend fell $448,000, leaving only a $4,000 net change.
What actually changed between these budget versions?
-$4,000Net change
Customers · 24,680 rows
5,510 people were stored as 6,214 records, depressing reported retention by 10.9 points.
Why does donor retention look worse than it is?
58.1%Retention, by person
Customers · 26,066 rows
The 22.4% rate combined three different events; the unrecovered no-show rate was 11.4%.
Why is our no-show rate 22%?
11.4%Unrecovered no-shows
Money · 68,892 rows
Fourteen-month-old recipe cards overstated the menu's margin by 6.1 points (64.5%, not 70.6%), and nine dishes were selling below cost.
Is the menu as profitable as the recipe cards say?
64.5%Contribution margin
Marketing & spend · 15,737 rows
Three platforms claimed 8,412 conversions but described 4,318 orders. 2,580 of the claims were the same order counted again.
How many orders did our ad platforms actually drive?
$92.26Blended cost per order
A project can answer more than one kind of question, so these also appear under other subjects. See all 11.