A weekly dashboard should answer one question in under a minute: is the business healthier than last week, and if not, where did it slip? Most retail dashboards fail that test by showing thirty charts. These eight numbers are enough. We calculated each one on a year of real orders of a UK online retailer (Dec 2010 – Nov 2011) so you can see what they look like in practice.
The eight numbers
| KPI | Formula | Dec 2010 – Nov 2011 |
|---|---|---|
| 1. Revenue | sum of quantity × price, completed sales | $9,633,406 |
| 2. Orders | count of distinct invoices | 18,957 |
| 3. Average order value | revenue ÷ orders | $508 |
| 4. Active customers | customers with at least one order in the period | 4,293 |
| 5. New-customer share | first-time buyers ÷ active customers, per month | 19.5% in Nov 11 |
| 6. Repeat purchase rate | customers with 2+ orders ÷ active customers | 63.9% |
| 7. Cancellation rate | cancelled value ÷ revenue | 3.2% |
| 8. Top-10% customer share | revenue of top 10% customers ÷ revenue from identified customers | 60.5% |
The first four describe volume: how much, how many, how big. The last four describe quality: where growth comes from, whether customers return, how much is lost to cancellations and how exposed you are to a few large buyers.
Revenue: watch the week, judge the trend
Source: UCI Online Retail II (Chen, 2019), sheet Year 2010-2011, CC BY 4.0; our calculation
Weekly revenue is noisy: single weeks swing by a third against their neighbours. One bad week is not a signal; three in a row is. The week ending January 2 shows zero because the store recorded no sales over the holiday break. A dashboard should label such gaps, or someone will spend Monday morning investigating a closed warehouse.
The seasonal shape is the real story. Monthly revenue grew from $670,439 in Jan 11 to $1,452,116 in Nov 11. Any week-on-week comparison in the autumn will look good; compare with the same week last year once you have it.
Orders and order value move separately
Average order value ranged from $417 to $620 across the twelve months, while the number of orders grew from 1,081 in Jan 11 to 2,751 in Nov 11. Revenue growth here came from more orders, not bigger ones. That matters for what you do next: more orders point to acquisition and reach; bigger orders point to pricing, bundles and minimum order values.
Customers: new versus returning
Source: UCI Online Retail II (Chen, 2019), sheet Year 2010-2011, CC BY 4.0; our calculation
The share of first-time buyers fell from about half in early 2011 to a fifth by summer and stayed between a fifth and a quarter through the autumn peak. The base is maturing: most of the busy season’s revenue came from customers who had bought before. 63.9% of customers ordered more than once during the year.
Concentration: how much rides on a few customers
The top 10% of identified customers brought 60.5% of their revenue. Revenue per customer averaged $1,916. High concentration is not bad in itself, but it changes priorities: losing ten large accounts would hurt more than losing a thousand small ones. Our RFM segmentation article shows who those customers are.
Geography is concentrated too:
Source: UCI Online Retail II (Chen, 2019), sheet Year 2010-2011, CC BY 4.0; our calculation
What the weekly view looks like
The layout that works: one row of eight tiles, each with the current week, the change against the four-week average and a small trend line; below it, revenue by week and the new-versus-returning split. Everything else goes one click deeper.
- Tiles first, charts second. If all eight tiles are in their normal range, the meeting is over.
- Compare with a baseline, not with last week alone. A four-week average absorbs the noise.
- Definitions on the dashboard. “Revenue” with or without cancellations, “customer” with or without guest orders: write it under the number.
- One source of truth. If finance and marketing get different revenue, fix that before building any chart.
Limits
- The data covers one year, so there is no year-over-year comparison here; your dashboard should have one.
- Revenue is gross of cancellations; the cancellation rate is shown separately.
- Customer KPIs use only orders with a customer ID.
Data: UCI Online Retail II (Chen, 2019), licensed CC BY 4.0. All numbers on this page are calculated by our script on that dataset and loaded into the page from its output.