Most revenue forecasts give you one number: “we will be at 9,000 in a year”. The trouble is that nobody knows monthly growth in advance. A range is more honest. This article shows how a simulator turns three guesses about monthly growth into a range for your monthly recurring revenue (MRR), and what the range does and does not tell you.
The three guesses
You give the simulator three monthly growth rates:
- Low: a poor month, say 0%.
- Most likely: the rate you expect, say 5%.
- High: a good month, say 12%.
You also give a starting MRR (5,000 in our example) and the number of months (12).
The simulator then plays out the year 10,000 times. Each time it picks growth rates from between your low and high, leaning towards the most likely one, and works out where MRR ends up. It then sorts the 10,000 results and reports three of them.
What the three results mean
- Low (10th percentile): 10% of the runs ended below this figure.
- Median: half ended below, half above.
- High (90th percentile): 10% of the runs ended above this figure.
So 80% of the runs land between low and high. That is a statement about the simulation, built from your three guesses. It is not a promise that your real revenue will land there 80% of the time.
A note on labels. Some industries call these “P10” and “P90”, and in some of them P10 is the figure that 10% of outcomes exceed, which is the opposite end. To avoid being read backwards, we say low (10th percentile) and high (90th percentile).
Why not just use the middle guess?
Compounding 5% a month for 12 months from 5,000 gives 8,979.28. Compounding the low guess (0%) gives 5,000, and the high guess (12%) gives 19,479.88. Those three single-rate lines are far apart, and they are all extreme in a way: nobody grows at exactly 12% every single month.
A range built from the three guesses sits between them and tells you where most of the outcomes cluster.
Two ways growth can vary
The same three guesses give a different range depending on how growth is allowed to vary.
Different each month. Every month gets its own growth rate. Good months and bad months tend to cancel each other out over a year, so the results cluster around the middle.
Same for the whole period. The simulator picks one growth rate for each run and keeps it for all 12 months. A run with a high rate stays high all year, and one with a low rate stays low, so the results spread much more widely.
A worked example
Start 5,000. Low 0%, most likely 5%, high 12%. 12 months. 10,000 runs, with the simulator’s default seed (12345).
| MRR after 12 months | Different each month | Same for the whole period |
|---|---|---|
| Low (10th percentile) | 8,720.66 | 6,695.89 |
| Median | 9,650.75 | 9,579.26 |
| High (90th percentile) | 10,725.05 | 14,225.37 |
The medians are close (about 9,600), but the width of the range is very different. If growth changes from month to month, 80% of the runs end between about 8,700 and 10,700. If growth is one fixed rate for the year, 80% end between about 6,700 and 14,200.
Which one is right for you? It depends on what you think drives your growth. If your growth rate shifts with the month’s luck, the first is closer. If it is set by something that holds all year, such as the market you are in or how well your product fits it, the second is closer. Many real businesses sit in between, so looking at both is sensible.
What the simulator leaves out
- Anything outside your three guesses. The range cannot contain a month worse than your low or better than your high. A lost big customer, a price change or a new competitor are not in it.
- Churn and expansion separately. Growth is a single net rate each month.
- Seasonality. Every month draws from the same range.
- The shape. The triangular shape (most likely value most probable, tails tapering to the low and high) is our choice for turning three guesses into a range. It is a convenient assumption, not a finding about real growth.
- The random seed. The numbers above come from one seed. A different seed gives slightly different numbers, because 10,000 runs is a sample, not every possible outcome.
Try it with your own numbers
The revenue forecast range simulator lets you set the starting MRR, the three growth rates, the number of months and which way growth varies. It shows the low, median and high for every month in a table.
Sources
- Uranie (CEA): Triangular law (the triangular distribution)
- uqtestfuns: Triangular distribution (sampling method)
- DNV: P10, P50 and P90 (why the labels can be read in opposite ways)
W
Written by the Webtilly team