What this calculator does
Day-over-day growth is the fastest and noisiest of the standard growth comparisons: useful for live dashboards and same-day monitoring, but the single worst window for judging an underlying trend, since almost any daily fluctuation looks dramatic in isolation.
This calculator deliberately shows the annualised equivalent of a daily rate specifically to make that noise visible. A day that is up 5% sounds unremarkable; the same rate compounded over 365 days produces a number in the billions of per cent, which is the calculator’s way of saying: do not extrapolate a single day.
The formula
The daily rate is the plain percentage change between today and yesterday. Annualising it compounds that rate 365 times over: (1 + DoD)^365 − 1. Because a full year is 365 compounding steps, even a very ordinary-looking daily rate produces an annualised figure with no real-world meaning: the size of the number is itself evidence that the comparison window is too short to extrapolate from.
| Term | Meaning |
|---|---|
| Today | The current day’s figure. |
| Yesterday | The immediately preceding day’s figure. |
| DoD growth | (Today − yesterday) ÷ yesterday × 100. |
| Annualised rate | What the DoD rate compounds to over 365 days, equal to (1 + DoD)^365 − 1. |
The inputs explained
| Field | What to enter |
|---|---|
| Today’s value | Today’s figure. |
| Yesterday’s value | Yesterday’s figure. |
When to use it
Monitoring a live dashboard
Operational metrics, orders, active sessions, error rates, are often tracked day over day because a live system needs same-day visibility, not a week’s wait for a smoother trend to emerge.
Catching an incident immediately
A sharp negative DoD figure on a metric like successful transactions or site availability is often the first signal something has broken, hours before a weekly or monthly report would surface it.
Explaining why a daily headline growth rate is meaningless on its own
A viral social post might show a genuine 50% DoD spike. Annualising it produces an astronomically large, obviously impossible number: a fast way to show why that day’s rate says nothing about the following 364 days.
Deciding whether a metric needs a longer window
If the annualised DoD figure swings wildly from one day to the next, that alone is a sign the metric needs to be read weekly or monthly, not daily, before drawing any conclusion about direction.
Worked examples
Every figure in the tables below is produced by this page’s own calculator at build time, so the numbers and the tool always agree. Select any row to load that scenario.
How the DoD figure changes as today’s result varies
A fixed prior-day figure of 400 against a range of outcomes for today.
| Today | DoD growth | Change | If sustained, annualised rate |
|---|---|---|---|
| 380 | -5.00% | -20.000 | -100.0% |
| 400 | 0.000% | 0 | 0.000% |
| 420 | 5.00% | 20.000 | 5,421,184,057.8% |
| 440 | 10.0% | 40.000 | 1.2833e+17% |
| 480 | 20.0% | 80.000 | 7.9644e+30% |
How the same result reads against different starting points
Today’s figure held at 420 while the prior day it is compared against varies.
| Yesterday | DoD growth | Change | If sustained, annualised rate |
|---|---|---|---|
| 350 | 20.0% | 70.000 | 7.9644e+30% |
| 400 | 5.00% | 20.000 | 5,421,184,057.8% |
| 420 | 0.000% | 0 | 0.000% |
| 440 | -4.55% | -20.000 | -100.0% |
| 500 | -16.0% | -80.000 | -100.0% |
Questions
Why do the annualised figures get so extreme?
Because 365 compounding steps magnify even a small daily rate enormously. It is the same mathematics as weekly or monthly compounding, just carried much further, which is why daily rates should almost never be annualised for real decision-making.
Should I ever actually use the annualised DoD figure?
Rarely as a forecast: mainly as an illustration of scale, to show that a headline daily percentage cannot be extrapolated. For real trend-reading, a weekly or monthly window is almost always more informative than daily.
Why is DoD so much noisier than WoW?
A single day captures whatever happened on exactly that day, including day-of-week effects (weekday versus weekend) and one-off events. A week averages across seven days and so cancels out much of that noise; a single day cannot.
What is a better way to smooth daily data?
A rolling average, for example, comparing today’s 7-day average to yesterday’s 7-day average, keeps the daily cadence while damping single-day noise, at the cost of reacting a little more slowly to a genuine sudden change.
Does a 0% DoD reading mean nothing happened?
It means today exactly matched yesterday for this particular metric: worth noting if the metric is usually volatile, since an unusually flat day can itself be a signal worth investigating.
For a steadier view that cancels out day-of-week effects, see the week-over-week growth calculator. For a fully custom comparison window, use the period-over-period growth calculator.