ROI and CAGR explained
Return on investment and compound annual growth rate answer the same question — how much did something grow — from two different starting points. ROI begins with an amount put in and an amount received back. CAGR begins with a value at two dates. This guide walks through what each one computes, why the plain total and the per-year rate differ, and why annualizing an ROI turns out to be the very same arithmetic as a CAGR. Every figure here is computed when the page is built by the same engine behind the ROI calculator and the CAGR calculator, on fictional inputs labelled as examples. It is one of Sumvia's business-math guides.
Two framings of the same growth
Both tools measure how a quantity changed, and both can report a yearly rate. The difference is what you hand them. ROI takes the money that left and the money that came back and scores the outlay as a percentage of itself. CAGR takes a value seen at a start date and an end date and reports the steady yearly rate that connects them. Neither looks inside the period: ROI ignores everything between the two totals, and CAGR ignores everything between the two dates. That is what makes them comparable across situations, and also what they leave out.
ROI: what came back against what went in
The return-on-investment formula is a single ratio: ROI% = (total received − amount invested) ÷ amount invested × 100. The denominator is the whole point — ROI measures against what you put in, not against revenue or against the profit alone. Put in 8,000 and receive 12,000 back, and the engine reports a net gain of 4,000 and an ROI of 50%. Enter the 4,000 profit in the received-back field by mistake and the same formula would read it as "spent 8,000, got 4,000 back" — a heavy loss — which is the most common way the number is misread.
A gain below the outlay produces a negative ROI: a loss, not an error. The scale bottoms out at −100%, the point at which nothing at all came back. Anything in between quantifies a partial loss, and the calculator reports it rather than refusing it.
Adding time: the annualized rate
A plain ROI says nothing about how long the money was at work, and that silence is where comparisons quietly go wrong — 50% earned in one year and 50% earned over a decade read identically. Supplying a holding period fixes this. Held for four years, the 8,000 into 12,000 example annualizes to 10.67% a year. Notice that this is not 12.5% — dividing the total return by the year count overstates the rate, because it ignores that each year's growth compounds on the one before. The engine instead takes the fourth root of the growth ratio, the rate that, applied four times in succession, lands exactly on the total.
The annualized form is defined only when something came back: it takes a root of the ratio between the gain and the cost, and a total loss — a gain of zero — has no yearly growth rate, so the engine returns an explicit error there rather than a number. A partial loss annualizes normally to a negative yearly rate.
CAGR: the same root, starting from two values
Compound annual growth rate asks the per-year question directly, from a start value and an end value instead of from money in and money out: CAGR% = ((end ÷ start)1 ÷ years − 1) × 100. Revenue of 50,000 reaching 80,000 over three years is 60% of total growth, and the cube root of that growth ratio gives 16.96% a year. Apply 16.96% three times over and you arrive back at 80,000, even though the real year-to-year path was almost certainly nothing so smooth.
Declines are fine. A value that falls from 200 to 150 over three years has a well-defined negative rate of -9.14% a year. The formula only breaks down at zero, which is why both the start and the end value must be above it — a geometric mean of a zero or a negative value has no meaning.
Why total return and annualized return differ
The gap between a total figure and a per-year figure is the work compounding does, and it is worth seeing plainly. The example's 50% total return is not 12.5% a year spread evenly; it is 10.67% a year compounded. The same holds for a value series: 60% of total revenue growth over three years is 16.96% a year, not a third of the total. Dividing a total by the number of years always answers a larger number than the rate that actually repeated, because division pretends each year started from the same base, when in truth every year started from a base the previous year had already enlarged.
Annualizing an ROI is a CAGR
The two tools are not merely similar; the annualized ROI and the CAGR are the identical computation. Annualizing an ROI takes the growth ratio between the amount invested and the amount received and raises it to one over the number of years — which is exactly what CAGR does to the growth ratio between a start value and an end value. The 8,000 into 12,000 over four years annualizes to 10.67%; feed the same 8,000 and 12,000 into the CAGR formula over four years and it returns 10.67% — the same number, because it is the same root of the same ratio. Choose the tool by the shape of your inputs: money out and money in points to ROI; a value at two dates points to CAGR.
Comparing outlays of different lengths
This is where the annualized rate earns its keep. Two outlays sit on the table. The first turns 10,000 into 20,000 over eight years; the second turns the same 10,000 into 13,000 in two. On plain ROI it is no contest — the first scores 100% against the second's 30%. Annualize them and the ranking inverts: the first compounds at 9.05% a year, the second at 14.02%. The larger total return belongs to the slower annual pace, because it was earned over four times as long.
Which figure should decide is not something the arithmetic settles — that depends on what happens to the money the faster outlay hands back years earlier, and whether it can be put to work again at a similar rate. The calculator's job is only to report both the total and the annualized figure so the question is visible rather than buried inside one tidy percentage. The total answers "how much, in all"; the annualized rate answers "how fast, per year"; they are different questions and a short run can win one while losing the other.
What these rates cannot see
Both measures are two-point summaries, and reading more into them than two points can carry is the recurring error. ROI compares two totals and sees nothing of the timing between them, so a deal that returns most of its cash early and one that pays everything at the end can post the same percentage. CAGR sees two endpoints and the clock and nothing else, so money added or withdrawn along the way is mistaken for growth — a series that was topped up inflates its own CAGR. Annualizing a very short window stretches the problem further: 2,000 growing to 2,200 across half a year is 10% of actual growth, yet entered as 0.5 years it annualizes to 21%, on the assumption that the half-year pace holds for a full year with compounding. The number is not wrong; it is an extrapolation wearing the clothes of a measurement, and short windows should be read as projections to challenge rather than rates to report.
A correct rate can also be the wrong rate for a decision. Neither figure knows anything about the risk a return carried, the fees that trimmed it, or the tax due on the way out — those belong in the inputs, folded into the amount invested and the amount received, before the percentage means what it appears to mean. Choosing which calculation fits a real situation is judgment the arithmetic cannot supply.
Frequently asked questions
What is the difference between ROI and CAGR?
They are two framings of the same growth. ROI starts from money out and money in; CAGR starts from a value observed at two dates and always expresses a yearly rate. They meet at the annualized figure: annualizing an ROI gives exactly the CAGR of the invested amount growing into the total received. The example — 8,000 into 12,000 over four years — is both a 50% total ROI annualizing to 10.67% and a CAGR of 10.67%: the identical number, reached from both directions.
Why is annualized ROI not just the total ROI divided by the number of years?
Because each year's growth compounds on the last, so the rate that repeats to the total is smaller than a straight division suggests. The example's 50% total return over four years annualizes to 10.67% a year, not 12.5%. The engine takes the fourth root of the growth ratio rather than dividing, which is why the two numbers never agree except over a single year.
Can ROI and CAGR be negative?
Yes. ROI runs down to −100%, which means nothing came back at all; anything between 0 and −100 is a partial loss. CAGR turns negative whenever the end value sits below the start — a decline from 200 to 150 over three years is a CAGR of -9.14% a year. The CAGR formula is defined only for start and end values above zero, since a geometric mean has no meaning at or below zero.
Do these figures account for when the cash moves?
No. ROI compares two totals and sees nothing in between; CAGR sees two endpoints and the clock and nothing else. A project that returns most of its money early and one that pays out only at the end can post the same ROI, and a series that was added to along the way inflates its own CAGR. Both are two-point measures by construction, not cash-flow analyses.
Can I use a period that is not a whole number of years?
Yes — the exponent is fractional, so 2.5 years, 18 months (1.5), or a single quarter (0.25) all work. Be wary of annualizing very short windows, though: 2,000 growing to 2,200 across half a year is real growth of 10%, but entered as 0.5 years the formula reports 21% a year, because it assumes that half-year pace repeats with compounding. Treat an annualized short window as a projection, not a measurement.
Why does a total loss have no annualized rate?
Annualizing takes a root of the ratio between what came back and what went in, and that root is only defined when something came back. If the gain is zero, the engine returns an explicit error rather than a number: a total loss has no yearly growth rate. A partial loss — any gain above zero — annualizes normally to a negative rate.
Every figure on this page is computed at build time from labelled fictional inputs by the same tested engine behind the calculators, and the tools themselves run entirely in your browser. The calculator implements the general mathematical definition of each rate; your jurisdiction's, contract's, or accounting conventions may differ, and this guide is not investment, tax, or accounting advice. Implementation and rounding details are on the methodology page.