SaaS revenue forecasting: how to project MRR and ARR from movement rates
SaaS revenue forecasting is projecting future MRR and ARR by modeling how revenue moves each month, not by multiplying last year by a growth number. The method that survives a board review is bottom-up: take your current MRR, then add forecasted new business and expansion and subtract contraction and churn, month by month, with the churn and expansion rates anchored to your own cohort retention.
Use top-down (market size times target share) only as a ceiling check for a fundraising story. Bottom-up is what drives hiring, spend, and the number you actually defend.
Single-multiplier forecasts fail for one reason: “we grew 30 percent last year, so model 30 percent again” hides the four or five forces that produced that 30 percent.
A board member will ask what happens if churn ticks up a point. A multiplier has no answer. A movement-rate model does.
Bottom-up vs top-down forecasting (and when to use each)
Bottom-up builds the forecast from your actual revenue mechanics. Top-down starts from the size of the market. They answer different questions, and mixing them up is how founders end up defending a plan they can’t operate.
| Approach | How it works | Best for | Where it fails |
|---|---|---|---|
| Bottom-up | Current MRR plus new, expansion, reactivation, minus contraction, churn | Planning, hiring, board forecasts, spend pacing | Slower to build, needs clean historical data |
| Top-down | TAM times target share times average contract value | Fundraising ceilings, TAM narrative | Useless for operations, encourages magical thinking |
When the two numbers diverge, trust bottom-up. Top-down tells you the room is big. It says nothing about whether you can fill it next quarter.
Top-down: fast, good for fundraising ceilings, dangerous for planning
Top-down forecasting multiplies your total addressable market by a target share and an average contract value. A $4B TAM, a 1 percent target share, and a $20K ACV implies $40M of reachable ARR.
That math is fine for a pitch deck. It is dangerous the moment you use it to plan headcount, because it assumes demand you have not proven and ignores the churn that erodes every dollar you win. Treat top-down as a ceiling, never a plan.
Bottom-up: the board-defensible default
Bottom-up starts from the revenue you already have and rolls it forward using rates you can measure. You begin with current MRR, apply your retention curves to the existing base, add new business at your real run rate, layer in expansion, and subtract contraction and churn.
Every input ties to an operational lever. Want more growth in the model? You raise new-business bookings or cut churn, and you can see exactly what each does.
That traceability is why bottom-up holds up when someone pushes on it. For the full version, see how to build a SaaS financial model.
The only revenue formula you need: the five MRR movements
Every dollar of MRR change comes from five movements. Get these right and the forecast builds itself.
Ending MRR = Starting MRR + New MRR + Expansion MRR + Reactivation MRR − Contraction MRR − Churned MRR
Here is what each term means:
- New MRR is recurring revenue from customers who signed this period.
- Expansion MRR is upsells, added seats, and usage growth from existing customers, tracked as expansion revenue.
- Reactivation MRR is revenue from customers who left and came back, tracked as reactivation MRR.
- Contraction MRR is downgrades and seat reductions, tracked as contraction MRR.
- Churned MRR is revenue from customers who cancelled outright.
Net the additions against the subtractions and you get net new MRR:
Net New MRR = (New + Expansion + Reactivation) − (Contraction + Churned)
That single line is your growth engine for the month. Forecast it well and the rest is arithmetic.
NRR, GRR, Net New MRR, and Quick Ratio with 2025 benchmarks
Four ratios summarize the movements. Know where you sit on each before you forecast a single month.
Net revenue retention (NRR) measures how the existing base grows or shrinks before new logos, expansion included. Gross revenue retention (GRR) measures the same base with expansion stripped out, so it caps at 100 percent.
The gap between them is your expansion engine. The full breakdown lives in gross vs net revenue retention.
NRR = (Starting MRR + Expansion + Reactivation − Contraction − Churn) / Starting MRR
GRR = (Starting MRR − Contraction − Churn) / Starting MRR
The SaaS quick ratio tells you how efficiently you grow: dollars added per dollar lost.
Quick Ratio = (New MRR + Expansion MRR) / (Churned MRR + Contraction MRR)
| Metric | 2025 median (private B2B SaaS) | Healthy / top-quartile | Source |
|---|---|---|---|
| Net revenue retention | 101–102% | Above 120% best-in-class | Benchmarkit 2025; SaaS Capital 2025 ($1M+ ARR) |
| Gross revenue retention | ~88% (down from ~90% three years prior) | 90%+ | Benchmarkit 2025 |
| NRR, $3–15M ARR stage | — | 99% top quartile | ChartMogul 2025 |
| Annual growth rate | 24–25% median | Bootstrapped ~23%, equity-backed ~25% | SaaS Capital 2025 (1,000+ cos) |
| CAC payback | ~20 months | 12–14 months historically | KeyBanc 2024 |
| SaaS quick ratio | — | 4.0+ ($4 gained per $1 lost) | Standard benchmark |
A quick ratio below 1.0 means the business is shrinking. Below 100 percent NRR is a flag. These are the anchors you sanity-check your own assumptions against.
Run your own with the SaaS quick ratio calculator and the net revenue retention calculator.
How to forecast each movement
Forecast each movement separately. Lumping them into one growth rate is exactly the mistake that breaks the model.
New business MRR
Forecast new MRR from your run rate and pipeline, then haircut for how expensive customers are to acquire right now. Take the trailing three-to-six-month average of new MRR, adjust for known pipeline and seasonality, and stress-test against sales capacity.
The constraint in 2025 and 2026 is payback. Median CAC payback stretched to roughly 20 months, up from a historical 12 to 14. New-business MRR ties up cash longer before it pays back, so a forecast that assumes you can spend your way to more logos needs a runway check.
See the CAC payback period guide for how to model it, or use the CAC payback calculator.
Expansion MRR
Anchor expansion to your NRR, then forecast it as a percentage of the existing base. If your existing base is $100K and your monthly expansion runs 2 percent, that is $2,000 of expansion MRR before you count a single new logo.
Median NRR sits around 101 to 102 percent in 2025. If you lack history, model expansion conservatively. SMB-heavy products land lower (90 to 105 percent NRR), enterprise higher (115 to 125 percent). Expansion is the cheapest growth you have, so model it deliberately rather than folding it into “growth.”
Forecasting churn the right way: cohort retention curves, not a flat percentage
Do not use a single flat churn percentage. It overstates retention because early-life customers churn far faster than tenured ones, so a blended rate flatters your forecast.
Build cohort retention curves instead: for each signup month, track what share of revenue survives at months 1, 3, 6, and 12, then apply the curve by customer age. A cohort six months old decays slower than one signed last week. The cohort analysis guide walks through building these, and cohort analysis defines the method.
Segment by plan or ACV, because the spread is enormous:
| Segment (ACV) | Monthly gross churn | Source |
|---|---|---|
| SMB (below $10K) | 3–5% | Optifai 2025 (N=939) |
| Mid-market | 1.5–3% | Optifai 2025 |
| Enterprise (above $100K) | 1–2% | Optifai 2025 |
| Best-in-class | Below 1% | Optifai 2025 |
Forecast each cohort’s decay separately and sum. The cohort approach is the single biggest accuracy lever in the whole model, because a flat rate buries the high early-life churn that actually shapes the curve. For the mechanics, see SaaS churn rate calculation and benchmarks or the churn rate calculator.
Worked 12-month example
Here is the full bottom-up forecast for a company starting at $100,000 MRR, adding $8,000 of new MRR per month, with 2 percent monthly expansion and 3 percent monthly churn on the existing base.
The per-month rule is simple:
MRR(t) = MRR(t−1) × (1 + 0.02 − 0.03) + 8,000 = MRR(t−1) × 0.99 + 8,000
| Month | Starting MRR | Expansion (2%) | Churn (3%) | New | Ending MRR |
|---|---|---|---|---|---|
| 1 | 100,000 | 2,000 | 3,000 | 8,000 | 107,000 |
| 2 | 107,000 | 2,140 | 3,210 | 8,000 | 113,930 |
| 3 | 113,930 | 2,279 | 3,418 | 8,000 | 120,791 |
| 4 | 120,791 | 2,416 | 3,624 | 8,000 | 127,583 |
| 5 | 127,583 | 2,552 | 3,827 | 8,000 | 134,307 |
| 6 | 134,307 | 2,686 | 4,029 | 8,000 | 140,964 |
| 7 | 140,964 | 2,819 | 4,229 | 8,000 | 147,554 |
| 8 | 147,554 | 2,951 | 4,427 | 8,000 | 154,079 |
| 9 | 154,079 | 3,082 | 4,622 | 8,000 | 160,538 |
| 10 | 160,538 | 3,211 | 4,816 | 8,000 | 166,933 |
| 11 | 166,933 | 3,339 | 5,008 | 8,000 | 173,263 |
| 12 | 173,263 | 3,465 | 5,198 | 8,000 | 179,531 |
Ending MRR at month 12 is $179,531. Multiply by 12 for the ARR run rate:
ARR = $179,531 × 12 = $2,154,367
Note what happens inside the table. New MRR is flat at $8,000, but churn grows every month because it is a percentage of a rising base.
By month 12, churn ($5,198) is eating most of the new business plus expansion. That is the engine that produces an equilibrium.
Equilibrium MRR = monthly new business / net monthly churn rate = $8,000 / 0.01 = $800,000
At $800K MRR, with these rates, growth stalls completely. New business exactly replaces churn. If the net rate doubles from 1 percent to 2 percent, that ceiling drops to $400,000.
See this dynamic in the maximum MRR calculator, or build the projection yourself with the MRR forecaster.
Why forecast error compounds, and how far out you can forecast
A small rate error does not stay small. It compounds every period, which is why a forecast that looks tight at month 3 can be badly wrong by month 12.
The (1+e)^n drift
A forecast that is off by a rate e each month drifts by roughly (1 + e) to the power of n over n months. Take the example above and change one input: actual churn is 4 percent, not the 3 percent you modeled. A one-point miss.
Rerun the same 12 months with 4 percent churn and ending MRR lands at $164,585 instead of $179,531. ARR comes in at $1,975,020 versus the $2,154,367 you forecast. That is an 8.3 percent miss from a single point of churn you got wrong.
Isolate the compounding on the surviving base and it is starker. Your modeled net decay was 1 percent a month (0.99 to the 12th power); the real decay was 2 percent (0.98 to the 12th). The ratio is 1.13, so the retained base alone is off by about 13 percent before you count the expansion those churned dollars would have produced.
Stretch the horizon to 18 months, run a higher-churn SMB book, or miss expansion too, and the swing reaches 15 to 25 percent. The mechanism is the same every time: errors multiply, they do not add.
Accuracy by horizon
A well-built bottom-up model holds roughly 82 to 88 percent accuracy out to 12 to 18 months, with plus or minus 12 to 20 percent variance. Past about 24 months, accuracy degrades sharply because the compounding above takes over.
Investors flag sustained plus or minus 25 percent variance as a red flag, because it means the model has no predictive grip. So forecast in monthly cohorts, present a range, and re-forecast every quarter as actuals land. Do not lock a single three-year number and defend it.
Build a range, not a point, and re-forecast quarterly
Forecast three scenarios, not one. A point forecast is wrong the day you publish it. A range tells the board where the real risk sits.
- Base case: your trailing rates, lightly adjusted for known pipeline.
- Upside: churn improves a point, expansion ticks up, new business hits plan.
- Downside: churn worsens a point, a key deal slips, expansion flattens.
Because error compounds, the spread between upside and downside widens the further out you go. That is correct: a forecast that shows a narrow band 18 months out is lying to you. Re-forecast every quarter, swap modeled rates for actuals, and watch which scenario you are tracking.
The SaaS revenue growth rate calculator helps you back-check the growth your scenarios imply, and CAGR is the right way to express multi-year growth as one comparable number.
2025–2026 benchmark cheat sheet
Anchor your assumptions to your own history first. Use these as the sanity check when you lack data or want to know whether your numbers are credible.
| Metric | Benchmark | Source |
|---|---|---|
| NRR (private B2B SaaS) | 101–102% median; 120%+ best-in-class | Benchmarkit / SaaS Capital 2025 |
| GRR | ~88% median | Benchmarkit 2025 |
| Annual growth | 24–25% median; KeyBanc 19–21% | SaaS Capital / KeyBanc 2025 |
| Monthly churn, SMB | 3–5% | Optifai 2025 (N=939) |
| Monthly churn, enterprise | 1–2% | Optifai 2025 |
| CAC payback | ~20 months median | KeyBanc 2024 |
| Quick ratio | 4.0+ healthy; below 1.0 shrinking | Standard benchmark |
| Forecast accuracy | 82–88% at 12–18 months | Industry horizon benchmark |
NRR is the lever that moves growth most. SaaS Capital found that moving NRR from the 90 to 100 percent band into the 100 to 110 percent band adds about 5 points of growth, and the highest-NRR firms grow roughly 83 percent faster than the median. More numbers live in the state of SaaS metrics 2026 and on the benchmarks page.
How Mowt automates this
Everything above assumes you have clean historical movement rates and cohort retention curves. Most teams do not, because pulling new, expansion, contraction, and churn out of Stripe by hand is slow and easy to get wrong.
Mowt computes those five movements live from your Stripe data and builds the cohort retention curves automatically, across more than $1B in tracked revenue for 500-plus companies. That turns the manual arithmetic in this article into a forecast that updates itself as customers sign, expand, and churn, so the rates you forecast from are your real ones. See forecasting and cohort analysis, or connect Stripe to start.
If you want the manual version first, the primer on how to calculate MRR and the MRR forecaster cover the inputs, and MRR vs ARR explains why you model in MRR and report in ARR.
FAQ
How do you forecast SaaS revenue?
Forecast bottom-up from MRR movement rates. Start with current MRR, then for each future month add forecasted new and expansion MRR and subtract contraction and churn, using rates from your own last 6 to 12 months and your cohort retention curves. Roll it forward month by month, multiply ending MRR by 12 for ARR, and re-forecast quarterly as actuals come in.
What is bottom-up vs top-down forecasting?
Bottom-up builds the forecast from your actual revenue mechanics (current customers, retention curves, new-business run rate, expansion), so it ties to operations and survives board scrutiny. Top-down starts from market size times target share times ACV. Use bottom-up for planning and board forecasts and top-down only as a ceiling check; when they diverge, trust the bottom-up number.
How do you forecast churn?
Do not use a single flat churn percentage, because early-life customers churn far faster than tenured ones and a blended rate overstates retention. Build cohort retention curves (what share of each signup month’s revenue survives at months 1, 3, 6, 12) and apply the curve by customer age. Segment by plan or ACV, since SMB churn (3 to 5 percent a month) dwarfs enterprise churn (1 to 2 percent).
How far out can you forecast SaaS revenue?
Reliably about 12 to 18 months with a well-built bottom-up model, typically 82 to 88 percent accuracy. Beyond roughly 24 months accuracy degrades sharply because small churn and expansion errors compound every period. Present a base, upside, and downside range and re-forecast every quarter rather than locking a single three-year number.
Should I forecast in MRR or ARR?
Model in MRR, because that is where movement happens monthly, then derive ARR as ending MRR times 12. MRR gives the high-frequency signal you need for hiring and spend pacing; ARR is the headline number for the board and valuation. Forecasting straight to ARR hides the monthly churn and expansion dynamics that decide whether you hit it.
What NRR and GRR should I assume in a forecast?
For 2025, median private B2B SaaS NRR is around 101 to 102 percent and median GRR around 88 percent. Anchor to your own numbers, but if you lack history, model GRR in the high 80s and NRR near 100 percent for a mid-market product; assume lower for SMB-heavy (90 to 105 percent NRR) and higher for enterprise (115 to 125 percent).
About the Author
Matt Smith
Serial entrepreneur and former big 4 consultant turned SaaS operator. Built and scaled analytics and data warehouses platforms at multiple enterprise Stripe companies before founding Mowt. Passionate about making complex metrics accessible to every founder.