Welcome! This guide covers the basics of descriptive statistics and how they apply to real eCommerce data - delivery times, shipping performance, and A/B test results.
By the end you'll be able to:
Understand key terms like average, median, and percentiles
Interpret patterns in your delivery and checkout data
Spot anomalies and act on them
1. What are descriptive statistics?
Descriptive statistics summarize large data sets so you can see patterns, trends, and areas to improve. In eCommerce you'll use them to analyze:
🚚 Delivery times - how long orders usually take to arrive
🛒 Order values - average revenue per checkout
📉 Test results - which shipping option wins an A/B test
Example: if your SLA promises 4-day delivery, descriptive stats let you back it up -"80% of our orders arrive within 4 business days."
2. Central tendency: mean, median, mode
These describe what's typical in your data:
Mean (average) - good for metrics like Average Order Value, but sensitive to extreme values.
Median (middle value) - the best measure of the typical customer's experience; not thrown off by outliers.
Mode (most frequent value) - what shows up most often, e.g. your most popular shipping destination.
Where these three fall depends on the shape of your data. In a symmetric (normal) distribution they line up; when data is skewed, the mean gets pulled toward the long tail while the median stays closer to the centre - which is exactly why the median is often the more reliable "typical" value.
Where mean, median, and mode fall depends on the shape of your data.
3. Variability: how spread out is the data?
Spread shows how consistent (or inconsistent) your operations are:
Range (min to max) - a quick best-vs-worst snapshot of delivery times.
Standard deviation (SD) - how tightly values cluster around the average. Lower SD = more consistent performance.
Outliers - any data point more than 3 standard deviations from the mean is treated as an anomaly (a system issue, a regional delay, or a one-off error worth investigating).
Values beyond ±3 standard deviations are treated as outliers.
4. Percentiles & the TP80 metric
Percentiles split your data into equal parts and set performance benchmarks:
TP80 (80th percentile) - the time within which 80% of orders arrive. Widely used in SLAs and delivery promises.
TP90 (90th percentile) - captures edge cases (the slowest 10%), useful for understanding operational risk.
Try it: calculate TP80 and TP90 on your delivery data - what does it say about your performance?
5. Key takeaways
✅ Central tendency (mean, median, mode) shows typical performance - and skew tells you which to trust.
✅ Variability (range, SD) reveals consistency.
✅ Percentiles (TP80/TP90) set realistic SLAs.
✅ Outliers (the 3 SD rule) are red flags for deeper investigation.
✅ The same tools apply to conversion rates, CTR, CVR, and more.
Need help applying these to your store's performance? Reach out - we're here to help.