The task and original case
Write a landing-page comparison that distinguishes percentage points from relative change. The invented variants provide a descriptive calculation, not a statistical conclusion. The strongest work sample includes the unresolved experiment assumptions beside the appealing headline.
| Case element | Value or requirement |
|---|---|
| Inputs | Variant A 50 conversions/1000 visits; B 60/1000. |
| Build | An experiment readout with scope, assignment assumptions and unresolved inference. |
This is an original invented exercise. It is independent practice, with no employer assessment or open-vacancy claim. Download the inputs, attempt your own result, then open the separate answer key.
Data and worked inputs
Build a result another person can follow
Variant A records fifty conversions from 1000 visits; B records sixty from 1000. Define what a visit and conversion mean, the observation window and whether one person can contribute multiple visits. Calculate each conversion rate from its own numerator and denominator before comparing them.
Subtract the rates for the absolute difference; divide that difference by A's rate for the relative change. A PivotTable can summarize the numeric counts, with Sum and Count settings checked against Microsoft's guidance. It cannot establish random assignment, a causal effect or significance merely by displaying a larger percentage. Keep these inference requirements in the readout.
Sources[1]
Check the worked result
A's rate is 50 ÷1000 =5%. B's is 60 ÷1000 =6%. The absolute difference is one percentage point. The relative increase is(6% −5%) ÷5% =20%. The phrase “20-percentage-point increase” would be incorrect.
The ten additional conversions in B's observed group are descriptive. Without a declared assignment method, uncertainty analysis and appropriate experiment design, this arithmetic does not establish that B reliably outperforms A or that its design caused the difference.
Make the failure visible
Give B a different number of visits and check that you still compare rates rather than raw conversions. Test a baseline rate of zero; relative change is then undefined under this formula. Also flag changes in visit or conversion definitions across variants. A larger denominator cannot repair an inconsistent measurement rule.
Revise it and ask someone to reproduce it
Ask a peer to rewrite your headline using both the absolute and relative change correctly. For revision two, append another observation window without overwriting the original. Compare whether the definitions and assignment assumptions stayed consistent before pooling counts or recommending a follow-up test.
- Both rates reconcile to displayed conversion and visit counts.
- One percentage point and 20% relative change are labelled correctly.
- The descriptive result does not claim statistical or causal certainty.
Describe the work accurately
After completing your own version, you could adapt this project bullet: “Created an independent landing-page readout with explicit denominators, percentage-point and relative-change calculations, and unresolved inference requirements.”
Use Projects or Independent practice as the heading. Keep the input files, result and revision together. Describe only your own completed work; the invented numbers are model outputs, not employer results or measured business impact.