Key takeaways
- One smoothed rate. CAGR replaces a jagged series with a constant annual rate between two endpoints.
- Count intervals, not data points. Fiscal 2022 to fiscal 2024 is two years, not three.
- It ignores volatility and cash flow. Contributions, withdrawals, and drawdowns never appear.
- Beginning value must be positive. A zero ending value gives exactly −100%.
- Report the endpoints alongside the rate. A CAGR without its inputs cannot be audited.
What CAGR answers
CAGR answers a narrow question: at what constant annual rate would the beginning value have to grow to reach the ending value in exactly n years? It summarises two values and a duration, which makes it useful for comparing investments, revenue lines, and market forecasts of different sizes and lengths.
It does not answer what happened in between. A business that fell 30% and then tripled can post the same CAGR as one that grew steadily, and a fund with a single strong year can look consistent over a decade. Treat CAGR as a description of the endpoints, and read the yearly series before drawing conclusions about performance.
The formula and the period count
CAGR = (ending ÷ beginning)^(1 ÷ n) − 1, where n is the number of years.
Total growth is ending ÷ beginning − 1. The two are related but not interchangeable: 100% total growth over five years is 14.87% a year, while 100% total growth over ten years is only 7.18% a year.
Count intervals, not observations. Revenue in fiscal 2022 and fiscal 2024 spans two years, so n = 2, even though there are three annual reports. Eighteen months is 1.5 years. Using the wrong n is the most common source of overstated growth: doubling in 18 months is 58.7% a year, not 41.4%.
The beginning value must be positive. If it is zero, CAGR is undefined; if it is negative, the ratio has no economic meaning. An ending value of zero gives exactly −100%, however long the period.
Worked examples
| Beginning | Ending | Years | Total growth | CAGR |
|---|---|---|---|---|
| $10,000 | $20,000 | 5 | 100% | 14.87% |
| $50,000 | $80,000 | 10 | 60% | 4.81% |
| $1,000 | $500 | 3 | −50% | −20.63% |
The first row doubles, so the rate is the fifth root of two minus one. The third row halves, so the rate is the cube root of one half minus one, which is negative. A negative CAGR is a real result, not an error: it means the ending value sits below the beginning value.
Doubling time is a useful cross-check. The rule of 72 gives 72 ÷ 14.87 = 4.84 years against an exact 5.0, close enough for a mental estimate. At higher rates the rule of 72 drifts, so use logarithms when precision matters.
Same CAGR, different paths
Two funds both turn $10,000 into $20,000 in two years, a CAGR of 41.4%. Fund A gains 41.4% in each year. Fund B loses 30% in the first year, falling to $7,000, then gains 185.7% to reach $20,000. The CAGR is identical; the experience is not.
This is the central limitation. CAGR discards the sequence of returns, and sequence matters for anyone who adds or withdraws money. An investor who sold after Fund B’s first year locked in a 30% loss that the CAGR never shows. Report CAGR next to the yearly values, the maximum drawdown, and the period length, and it becomes informative rather than misleading.
CAGR versus the arithmetic mean and IRR
The arithmetic mean of Fund B’s two yearly returns is (−30% + 185.7%) ÷ 2 = 77.9%, which is not a rate anyone earned. The geometric mean, which is CAGR, is 41.4%. For volatile series the arithmetic mean always exceeds the geometric mean, and the gap grows with volatility.
CAGR also assumes no cash enters or leaves. If an investor puts in $10,000 at the start and another $5,000 at the beginning of year four, then ends with $20,000 after five years, the endpoint CAGR is still 14.87%, but the money-weighted return is about 7.3%, because part of the ending value came from the later contribution. Use IRR for cash flows and CAGR for a single value that grows undisturbed.
Using CAGR on revenue and users
Revenue CAGR is standard in plans and investor updates, and the same cautions apply. Use consistent fiscal years, constant currency, and the same revenue definition at both ends. Mixing GAAP and adjusted revenue, or changing the scope of what is counted, turns CAGR into a measure of accounting changes rather than growth.
For user or unit metrics, be explicit about what is counted. Monthly active users, paying accounts, and seats are different bases, and a CAGR computed across a definition change is meaningless. If the definition changed, restate the earlier figure or disclose the break.
Short periods exaggerate. A 20% gain in one quarter annualises to 107%, which is arithmetically correct and practically useless. CAGR is most defensible over three or more full years, and it should be labelled with the exact start and end dates.
Where CAGR stops being useful
CAGR cannot describe a series that crosses zero or changes sign. It also hides seasonality, one-off events, and acquisitions. A revenue CAGR that includes an acquisition measures combined growth, not organic growth. Public companies separate the two for exactly this reason, and a private business comparing itself to a public peer should do the same.
Inflation is another omission. A 14.87% nominal CAGR with 3% inflation is 11.52% real, computed as (1.1487 ÷ 1.03) − 1. Over long periods the difference compounds: ten years at 14.87% nominal grows a dollar to $4.00, while its purchasing power grows to about $2.98.
Common mistakes
- Dividing by the number of data points instead of the number of intervals.
- Annualising a sub-year period without labelling it as a projection.
- Comparing a CAGR that includes contributions with one that does not.
- Using CAGR to describe a series with zero or negative values.
- Reporting a smoothed rate without the underlying yearly figures.
- Mixing fiscal years, currencies, or revenue definitions between endpoints.
Practical caveats
Round the rate, not the inputs. Computing 14.87% and then rounding the ending value to $20,000 can shift the result by a few basis points; keep full precision until the final display. State the period in years, including fractions, so a reader can reproduce the number.
Currency conversion needs a stated convention. Converting the beginning value at one rate and the ending value at another produces a CAGR that reflects the exchange rate as much as the business. Either use a constant rate or report the currency effect separately.
Finally, CAGR is a single number and should never travel alone. A table with beginning value, ending value, years, total growth, and CAGR is auditable; a rate quoted in isolation is not. When the rate is surprising, the table usually shows why.
Frequently asked questions
Can CAGR be negative?
Yes. If the ending value is below the beginning value, the rate that connects them is negative. A $1,000 value that falls to $500 over three years has a CAGR of −20.63%.
What is the difference between CAGR and average annual return?
The average annual return usually means the arithmetic mean of yearly returns, which overstates compound growth when returns vary. CAGR is the geometric mean and reflects the actual change in value.
Does CAGR include dividends?
Only if the series you use is a total-return series. A price-only series produces a price CAGR, which understates the return to an investor who reinvested dividends.
How many years should I use?
Full periods, ideally three or more. Short windows amplify noise, and partial years should be expressed as fractions rather than rounded up.
Is CAGR the same as annualised return?
For a single lump sum with no cash flows in or out, yes. With contributions or withdrawals, the annualised money-weighted return (IRR) will differ.
What if the ending value is zero?
CAGR is −100%. The business or investment lost its entire value, and no positive rate can describe the result.
Further reading
Keep exploring
Decision path