Change from one week to the next
Week-over-week growth is the fastest signal a business has. It is also the noisiest, and the sections below are mostly about not being misled by it.
The formula
Worked example: 8,000 last week, 9,200 this week
- Change: +1,200
- Growth: 1,200 ÷ 8,000 = +15%
Why one week means little
Fifteen percent in a week compounds to over a thousand percent a year, which is not going to happen. Weekly figures swing on a public holiday, a payday, a single large order or a weather change, and a good week is followed by an ordinary one more often than by another good one.
| Week | Value | WoW change | 4-week average |
|---|---|---|---|
| 1 | 8,000 | — | — |
| 2 | 9,200 | +15.0% | — |
| 3 | 8,400 | −8.7% | — |
| 4 | 8,900 | +6.0% | 8,625 |
| 5 | 8,600 | −3.4% | 8,775 |
The weekly changes bounce between −8.7% and +15%. The four-week average moves from 8,625 to 8,775 — a 1.7% rise, which is the real trend.
Compare against the same week last year
For anything seasonal, the useful comparison is not last week but the same week a year ago. A garden centre up 15% week-over-week in April is seeing spring arrive, not a change in its business. Year-over-year on the same week removes the season.
Small bases exaggerate
From 20 to 23 is also +15%. Three extra units is not a trend, and a metric that starts near zero will show enormous percentages for months. Report the absolute change alongside the percentage until the base is large enough for the percentage to mean something.
What this calculator leaves out
Whether the two weeks are comparable at all. A five-day trading week against a four-day one is a 25% difference before anything happened.
When weekly data is the right tool
Weekly figures are useful for exactly one thing: catching a break in the pattern quickly. A launch, an outage, a pricing change or a competitor's move shows up in days, and waiting for the monthly figure means waiting three weeks longer than necessary.
Use it as an alarm, not as a report. Act on a weekly figure only when it moves outside the range the last twelve weeks established, and confirm with the month before drawing a conclusion.
Reporting it honestly
State the absolute figures next to the percentage, the same week last year alongside last week, and the number of trading days in each. A weekly report that shows only the percentage change is not wrong; it is incomplete in exactly the way that leads to bad decisions.
Negative weeks
A fall from 9,200 back to 8,000 is −13%, not −15%. The percentage is asymmetric because the base changed, and a metric that rose 15% then fell 13% is exactly where it started.
Anyone reporting week-over-week should add the two-week change against the original base, or a flat month can be presented as alternating good and bad news.
Related calculators
- Month-over-month — the same calculation on a less noisy interval
- Year-over-year — the comparison that removes seasonality
- Moving average — the trend underneath the weekly noise