Statistic Test ENG

 

1. “Z-test in Zalando – Size & Fit”

Scenario	# Groups	Metric	Question	Test
Show size recommendation badge	A/B	Conversion rate	Does conversion increase?	Z-test
New size prediction model	A/B	Return rate due to size	Does return rate decrease?	Z-test
“Find your size” button	A/B	Click-through rate	Does CTR increase?	Z-test
Helpful popup after purchase	A/B	Yes rate	Does Yes rate increase?	Z-test
Longer size form	A/B	Completion rate	Does completion drop?	Z-test
Virtual fitting feature	A/B	Adoption rate	Does adoption increase?	Z-test
New size guide page	A/B	Bounce rate	Does bounce rate decrease?	Z-test

Z-test pattern

• 2 groups

• Metric = proportion / rate

• Core question: “Is the rate different?”

---

2. “T-test in Zalando – Size & Fit”

Scenario	# Groups	Metric	Question	Test
New size chart	A/B	Avg order value	Does average order value increase?	T-test
New size guide layout	A/B	Avg time on page	Does average time increase?	T-test
Outfit recommendation	A/B	Avg items per order	Does avg items increase?	T-test
Fit satisfaction survey	A/B	Avg fit score	Does avg satisfaction increase?	T-test
New return process	A/B	Avg processing time	Does avg time decrease?	T-test
Compare 2 size models	A/B	Avg size error	Does avg error decrease?	T-test
New PDP layout	A/B	Avg revenue per session	Does avg revenue increase?	T-test

T-test pattern

• 2 groups

• Metric = continuous mean

• Core question: “Is the mean different?”

---

3. “Chi-square in Zalando – Size & Fit”

Scenario	# Groups	Metric	Question	Test
Return reasons	A/B	Return reason categories	Is the distribution different?	Chi-square
Fit perception	A/B	Too small / Just right / Too big	Is the fit distribution different?	Chi-square
Size funnel steps	A/B	Step1 / Step2 / Step3 / Complete	Is the funnel distribution different?	Chi-square
Device distribution	A/B	Mobile / Desktop / Tablet	Are group and device independent?	Chi-square
Purchased size distribution	A/B	XS / S / M / L / XL	Is the size distribution different?	Chi-square
Feedback levels	A/B	4-level helpfulness	Is the feedback distribution different?	Chi-square

Chi-square pattern

• Outcome = categorical (multiple categories)

• Data = count table

• Core question: “Is the category distribution different?”

---

4. “ANOVA in Zalando – Size & Fit”

Scenario	# Groups	Metric	Question	Test
3 size recommendation UIs	A/B/C	Avg conversion score	Is any group different?	ANOVA
4 ML models	A/B/C/D	Avg size error	Is any model better?	ANOVA
3 size guide layouts	A/B/C	Avg time	Is any layout different?	ANOVA
3 outfit strategies	A/B/C	Avg items per order	Is any strategy better?	ANOVA
3 size form versions	A/B/C	Avg time	Is complexity different?	ANOVA
3 confidence displays	A/B/C	Avg trust score	Is trust different?	ANOVA

ANOVA pattern

• More than 2 groups

• Metric = continuous mean

• Core question: “Is at least one group different?”

---

5. Final “Cheat Sheet” Table (All Tests Together)

This is the one table interviewers love.

Test	# Groups	Metric Type	Core Question
Z-test	2	Proportion / Rate	Is the rate different?
T-test	2	Continuous Mean	Is the mean different?
Chi-square	≥2	Categorical	Is the distribution different?
ANOVA	>2	Continuous Mean	Is any group different?