Orders, baskets or events

How concentrated is revenue among customers?

The data. standard_rfm_v2.csv, the published sample file for RFM segmentation: 2,139 rows; columns Order ID, Customer ID, Order Date, Order Amount, Channel. Sample data, not a customer's.

The calculation. Python, standard library only. This exact code ran against the file when this page was built, and printed the output below it.

import csv
rows = list(csv.DictReader(open("standard_rfm_v2.csv")))
spend = {}
for r in rows:
    a = r["Order Amount"]
    if not a or float(a) <= 0:
        continue  # blank or refund rows are not purchases
    c = r["Customer ID"]
    spend[c] = spend.get(c, 0) + float(a)
top = sorted(spend.values(), reverse=True)
k = round(len(top) * 0.2)
print(len(rows), "rows,", len(spend), "customers with a purchase")
print(f"top 20% ({k} customers) bring",
      f"{sum(top[:k]) / sum(top):.1%} of revenue")

2139 rows, 599 customers with a purchase
top 20% (120 customers) bring 74.8% of revenue

What it says. The fifth of customers who spent most bring 74.8% of revenue in this order file. That concentration is why segmenting customers by recency, frequency and spend usually beats treating them as one average customer.

What it does not say. Spend to date is backward-looking: a big spender who has stopped buying counts fully here. Refund rows were left out rather than netted against purchases. The library analysis scores recency and frequency as well and names the segments.

Run RFM segmentation on this file or your own, or open its stored sample report.