GROWTH Function in Excel (6 Easy Examples)

Excel’s GROWTH function returns predicted values from an exponential curve fitted to existing data.

It suits data that grows by a roughly consistent percentage, rather than by the same amount each period. You can estimate missing values or project later periods.

The predictions depend on that exponential pattern remaining a reasonable fit. They describe the fitted curve, not a guarantee of future growth.

In this article, I’ll show you how to project future values, estimate missing observations, and compare exponential growth with a linear trend.

Table of Contents

GROWTH Function Syntax in Excel

GROWTH fits an exponential curve to known values and returns estimates at the positions you specify.

=GROWTH(known_y's, [known_x's], [new_x's], [const])

known_y’s (required): The observed values, such as monthly seller counts. Every value must be greater than zero.
known_x’s (optional): The matching positions, such as month numbers. If omitted, Excel assumes 1, 2, 3, and so on.
new_x’s (optional): The positions you want estimates for. If omitted, Excel uses known_x's and returns fitted values at the existing positions.
const (optional): TRUE or omitted lets Excel estimate the multiplier b. FALSE forces b to 1.

The curve has the form y = b * m^x. Here, m is the growth factor per period, and b is the fitted value at x = 0.

GROWTH returns projected values, not a percentage growth rate. You don’t need to supply a rate because Excel estimates the curve from your observations.

When to Use GROWTH Function

Project future counts when the historical pattern looks like steady percentage growth.
Estimate a missing observation within a series that follows an exponential pattern.
Compare an exponential projection with a straight-line projection before choosing a model.
Explore how extending a compounding pattern changes the projected values over time.

Example 1: Project the Next Three Months

Let’s start with a marketplace whose seller count has been growing each month.

Below is the dataset. Columns A:C contain January through September, their month numbers, and seller counts. Columns E:F list the three future months and their numbers.

We want to project the marketplace seller count for October, November, and December.

Here is the formula to enter in G2:

=GROWTH($C$2:$C$10,$B$2:$B$10,F2:F4)

=GROWTH($C$2:$C$10,$B$2:$B$10,F2:F4) in G2

The formula spills into G2:G4, displaying 9,121, 9,941, and 10,835 for the three future months.

How this formula works:

$C$2:$C$10 supplies the nine observed seller counts.
$B$2:$B$10 supplies their corresponding month numbers.
F2:F4 asks for estimates at months 10, 11, and 12.
With const omitted, Excel fits both the growth factor and the starting multiplier.

Excel uses all nine observations to fit the curve. It doesn’t take September’s count and repeat only the most recent month’s percentage change.

Pro Tip: Enter the formula only in G2 and leave G3:G4 empty. In Excel 2021 and later, pressing Enter returns all three results. Occupied output cells cause #SPILL!.

Example 2: Estimate a Missing Month

Now let’s fill a gap inside the historical series.

Below is the dataset. Columns A:C list website sessions with May missing. Cells E2:F2 identify May and its month number, 5.

We want to estimate May’s website sessions using the surrounding months.

Here is the formula to enter in G2:

=GROWTH(C2:C9,B2:B9,F2)

The result displays as 23,917 sessions. Because F2 contains one requested month number, the formula returns one value in G2.

This is interpolation: estimating a point inside the observed time span. The estimate uses the fitted curve across all eight observations, not an average of April and June.

Notice that B2:B9 contains 1, 2, 3, 4, 6, 7, 8, and 9. That gap preserves June’s actual position after the missing month.

Omitting known_x's would make Excel assume positions 1 through 8. June would become position 5, shifting the timeline and producing an incorrect estimate for May.

Pro Tip: Keep the real period numbers when observations are missing. Don’t renumber the remaining rows to remove the gap, and label the returned value as an estimate rather than an observed count.

Example 3: GROWTH vs TREND on Revenue

Let’s compare two projections that use exactly the same historical data.

Below is the dataset. Columns A:B contain annual revenue from 2018 through 2025, and D2:D4 contains the future years 2026, 2027, and 2028.

We want to see how an exponential projection differs from a straight-line projection for this compounding revenue series.

Here is the GROWTH formula to enter in E2:

=GROWTH($B$2:$B$9,$A$2:$A$9,D2:D4)

=GROWTH($B$2:$B$9,$A$2:$A$9,D2:D4) in E2

The results spill into E2:E4: $7,026,123, $8,304,296, and $9,814,989 for 2026 through 2028.

And here is the TREND formula to enter in F2 for comparison:

=TREND($B$2:$B$9,$A$2:$A$9,D2:D4)

This spills into F2:F4, displaying $6,159,286, $6,736,905, and $7,314,524 for those same years.

GROWTH fits a curve with a constant percentage change per period. As the base gets larger, the dollar increases get larger too.

TREND fits a straight line with a constant dollar increase per year. That’s why its projections pull further below GROWTH as the years advance.

The choice depends on the pattern you’re modeling. A higher answer doesn’t make GROWTH more accurate, and a lower answer doesn’t automatically make TREND safer.

For a series that adds a similar amount each period, TREND or FORECAST.LINEAR is more appropriate. GROWTH suits a pattern that multiplies by a similar factor.

Example 4: What const FALSE Changes

Here’s an optional argument that can push a reasonable projection far too high.

Below is the dataset. Columns A:B contain eight weeks of loyalty signups, and D2:D4 lists future weeks 9, 10, and 11.

We want to compare the default projection with one that sets const to FALSE.

Here is the default formula to enter in E2:

=GROWTH($B$2:$B$9,$A$2:$A$9,D2:D4)

It spills into E2:E4, returning 352, 403, and 461 signups. These follow the historical series, which ends at 309.

Now enter the formula with const set to FALSE in F2:

=GROWTH($B$2:$B$9,$A$2:$A$9,D2:D4,FALSE)

=GROWTH($B$2:$B$9,$A$2:$A$9,D2:D4,FALSE) in F2

The results jump to 5,449, 14,174, and 36,866 in F2:F4. That first week is roughly 15 times the default projection of 352 signups.

Setting const to FALSE forces b to 1 in y = b * m^x. The curve must therefore pass through a value of 1 at period zero.

That constraint doesn’t suit these signups, which already reach 120 in week 1. Excel changes the fitted growth factor to accommodate the forced starting point.

This sets a multiplier to 1. It doesn’t set an additive intercept to zero, as a straight-line model’s constant option would.

Pro Tip: Leave const omitted unless your model specifically requires a value of 1 at period zero. If you want to inspect the fitted m and b coefficients, LOGEST returns them directly.

Example 5: Project Beyond the Observed Data

Let’s see what happens when we keep extending the same curve.

Below is the dataset. Columns A:B contain 12 months of app downloads, and D2:D7 lists future months 13, 14, 15, 18, 24, and 36.

We want to compare nearby projections with the results much further beyond our observed data.

Here is the formula to enter in E2:

=GROWTH($B$2:$B$13,$A$2:$A$13,D2:D7)

=GROWTH($B$2:$B$13,$A$2:$A$13,D2:D7) in E2

The formula spills six estimates into E2:E7, one for each requested month:

Month 13: 27,945 downloads.
Month 14: 30,493 downloads.
Month 15: 33,274 downloads.
Month 18: 43,233 downloads.
Month 24: 72,984 downloads.
Month 36: 207,997 downloads.

The final observed month contains 25,600 downloads. Treat the month-36 result of 207,997, roughly eight times that, as implausible rather than a dependable forecast.

The formula keeps applying the fitted growth factor indefinitely. It knows nothing about market size, advertising changes, or whether the app can sustain that pattern.

Exponential projections can rise sharply as you extend the horizon. GROWTH doesn’t return an error when a mathematically valid result stops making business sense.

Pro Tip: Keep your projection horizon tied to what the available data can support. For repeating seasonal peaks and dips, consider FORECAST.ETS instead of assuming one smooth compounding pattern.

Example 6: Fix Common GROWTH Errors

Finally, let’s compare a working formula with three deliberate mistakes.

Below is the dataset. Column A contains month numbers, B contains online orders, and C repeats those orders with a zero in C4. Column E labels each formula case.

We want to project month 9 and identify which argument problems prevent GROWTH from returning an answer.

Here is the correct formula in F2:

=GROWTH(B2:B9,A2:A9,9)

It displays 453.54. Both historical ranges contain eight entries, the order counts are positive, and the requested month is numeric.

Here is the deliberate zero-value error in F3:

=GROWTH(C2:C9,A2:A9,9)

This returns #NUM! because C4 contains zero. GROWTH fits the curve using logarithms, so zero and negative observed values aren’t allowed.

Correct the source value if it’s a data-entry mistake. If zero is a real observation, reconsider the model rather than replacing it with an invented positive count.

Here is the deliberately mismatched range formula in F4:

=GROWTH(B2:B9,A2:A8,9)

This returns #REF! because there are eight order counts but only seven month numbers. Extend the historical x range to A2:A9, as in the correct formula.

The historical ranges must pair up. The future range can contain a different number of positions, as the earlier examples demonstrate.

Here is the deliberate text-argument error in F5:

=GROWTH(B2:B9,A2:A9,"Nine")

This returns #VALUE! because "Nine" is text. Use the number 9, as in F2, or reference a cell containing that number.

These are four separate formulas. The errors in F3:F5 are intentional demonstrations, not failed parts of a spilled result.

Tips & Common Mistakes

Keep the timeline intact. Supply known_x's when periods are missing or unevenly spaced. Omitting it assumes consecutive positions starting at 1.
Check positive observed values. Zero or negative known_y's values cause #NUM!. Don’t hide a genuine model problem with an error-handling wrapper.
Use Enter in modern Excel. In Excel 2021 and later, multiple results spill automatically. Excel 2019 and earlier require preselecting the output range and confirming with Ctrl+Shift+Enter.
Leave spill space clear. An occupied output cell causes #SPILL!. Edit a spilled formula at its starting cell; G2# references the entire spill from Example 1.
Watch for implicit intersection. An @ operator can reduce an array result to one value. Check for it if a formula expected to return several projections shows only one.
Distinguish formatting from rounding. The whole-number formats in these examples control the displayed counts. GROWTH still calculates decimal estimates, as Example 6’s two-decimal display illustrates.
Choose the model by the pattern. No newer Excel function replaces GROWTH. TREND and FORECAST.LINEAR fit additive change; FORECAST.ETS addresses seasonal time series.

The missing-month and comparison examples give you ways to check the fit before extending it.

Keep observed values separate from estimates, and question a projection when its assumptions no longer match the business.

List of All Excel Functions