SaaS ARR Forecasting: The Bottom-Up Model That Actually Works for $10K–$500K MRR

Most ARR forecasts at the $10K–$500K MRR stage are wrong in a predictable direction: they overestimate ending ARR because they underestimate the churn drag on the existing base.

A company at $200K MRR growing new customer acquisition at $20K new MRR per month looks like it will cross $500K MRR in about 15 months: $200K + ($20K × 15) = $500K. But this ignores that the existing $200K MRR base is churning. At 2% monthly churn, the base loses $4K in month 1, $3.9K in month 2, and so on. Over 15 months, the cumulative churn drag on the base is approximately $50K of MRR — which means the company crosses $500K ARR in 18 months, not 15. A 20% forecast error from a simple omission.

This guide covers the bottom-up ARR forecasting methodology that accounts for this: the MRR schedule approach, cohort-based churn projection, pipeline-to-ARR conversion, expansion modeling, and the three-scenario framework that separates good planning from wishful thinking.

Why Top-Down ARR Forecasting Does Not Work

Top-down ARR forecasting looks like this:

• "Our TAM is $5B. We can capture 0.1% in 3 years = $5M ARR."
• "We are growing at 15% MoM. At that rate, we will be at $1M MRR in 14 months."

Both approaches have the same flaw: they extrapolate from current trend without modeling the mechanics that produce the trend. The 15% MoM growth rate is a net number that already includes churn. If churn rate increases, the growth rate drops — and the top-down model does not detect this until it has already happened.

Bottom-up forecasting builds the projection from the components:

• How many new customers will we acquire per month, and at what average MRR?
• How much will existing customers expand?
• How much will churn reduce the base?

The bottom-up model makes the assumptions explicit. When actual results deviate from forecast, you can trace the deviation to its component (lower acquisition, higher churn, slower expansion) rather than just observing that "growth came in lower than expected."

The Building Blocks of a Bottom-Up ARR Forecast

Component 1: Starting MRR

Use your current month-end MRR as the starting point. This is the only input that should come directly from your billing data — everything else is a projection.

From your SaaS metrics tracking spreadsheet Tab 1, the closing MRR for the most recent closed month is your starting point.

Component 2: New MRR Projection

New MRR projection is a product of your acquisition funnel model:

New Customers per Month = Top of Funnel Leads × SQL Conversion % × Trial Conversion % × Paid Conversion %

New MRR per Month = New Customers per Month × Avg MRR per New Customer

Building the acquisition model:

Use 3-month trailing averages for each conversion rate. Do not use last month's best performance or a hoped-for future rate — use the rate that reflects current repeatable performance.

Growth assumption in the acquisition model: If you are scaling a channel (hiring SDRs, increasing paid search budget), model the ramp:

• Month 1–2: current performance
• Month 3–4: 10% increase (new SDR fully ramped at Month 3)
• Month 5+: 20% increase (second SDR adds 50% of first SDR's output)

Make ramp assumptions explicit. A new SDR takes 60–90 days to full productivity. A content marketing investment takes 6–12 months to compound. Modeling instant ramp is how acquisition forecasts become 40–60% wrong.

Component 3: Expansion MRR Projection

Expansion MRR comes from two sources:

1. Usage-based natural expansion: Customers naturally grow into higher tiers as their usage increases
2. Proactive expansion: CS-driven upsell and cross-sell motions

For the forecast:

Monthly Expansion MRR = (Starting MRR - Churned MRR) × Monthly Expansion Rate %

Use your trailing 6-month average expansion rate. For most SMB SaaS products, this is 0.5–1.5% per month of the retained base. For mid-market products with seat-based expansion, 1.5–3% per month.

Expansion MRR is the most under-forecasted component. Founders who project only new customer acquisition underestimate their ARR potential from the existing base. At $200K MRR with a 1% monthly expansion rate and 90% gross retention, expansion adds $1,800/month in the first month alone — compounding as the retained base grows.

For the full expansion revenue framework, see expansion revenue scoring.

Component 4: Churn and Contraction Projection

This is the most commonly mishandled component.

The flat rate mistake:

Monthly Churned MRR = Starting MRR × Monthly Churn Rate

Applying a flat 2% monthly churn to $200K starting MRR gives $4K churned in Month 1. Applying the same rate to $230K MRR in Month 6 gives $4,600 churned. The churn grows in absolute dollar terms as MRR grows.

This is correct for the aggregate model. What it misses is cohort heterogeneity — newer cohorts churn at higher rates than mature cohorts.

The cohort-adjusted churn model:

From your cohort retention data (Tab 4 of the spreadsheet), you can see that Month 1–3 cohorts churn at 3.5% per month while Month 12+ cohorts churn at 1.2% per month. As you grow, you accumulate more of the high-churn new cohorts.

A simple approximation:

Effective Churn Rate in Month N = (Churn Rate of Mature Cohorts × % of MRR from mature cohorts) + 
                                   (Churn Rate of New Cohorts × % of MRR from new cohorts)

As you grow faster, a higher proportion of MRR is from newer, higher-churn cohorts — your effective blended churn rate increases even if per-cohort churn rates are stable. Model this explicitly in a fast-growth scenario.

Contraction MRR projection:

Monthly Contraction MRR = Starting MRR × Monthly Contraction Rate %

Use trailing 6-month average contraction rate. For most SaaS products, this is 0.3–1.0% per month of the base.

The MRR Schedule: Putting It Together

At 12 months, this model produces $3.6M ARR — compared to the $4.4M that a naive "current MRR + $12K/month × 12" calculation would produce. The churn drag is $800K of ARR over the forecast period.

Pipeline-to-ARR Conversion

For sales-led SaaS, your pipeline provides an additional input to the new customer acquisition forecast.

Basic pipeline-to-ARR conversion:

Forecasted New ARR from Pipeline = 
  (Early Stage Pipeline × Early Stage Win Rate × Avg ACV) +
  (Mid Stage Pipeline × Mid Stage Win Rate × Avg ACV) +
  (Late Stage Pipeline × Late Stage Win Rate × Avg ACV)

Use historical win rates by stage from your CRM. Apply a sales cycle lag — if your average sales cycle is 45 days, pipeline closed in the next 30 days comes from deals currently in late stage.

Pipeline-to-ARR accuracy: Pipeline forecasts are more accurate at short horizons (30 days) and less accurate at long horizons (90+ days). For 12-month ARR forecasting, use pipeline coverage for the next 60 days and fall back to the acquisition model for months 3–12.

Pipeline coverage ratio benchmark: A healthy sales pipeline has 3–4x your quarterly new ARR target in total pipeline value. If you need $300K in new ARR in Q3 and have $1.2M in qualified pipeline, coverage is 4x — healthy. Below 2x coverage in the current quarter is a demand generation warning signal.

Three-Scenario Framework

Build three scenarios by varying three inputs:

Example at $200K MRR (Base inputs: 36 new customers/mo at $336 avg MRR, 2% churn, 0.9% expansion):

What the scenarios tell you:

The conservative-to-base gap ($3.1M vs. $3.6M at 12 months) is the cost of current churn and acquisition underperformance. The base-to-optimistic gap ($3.6M vs. $4.3M) is the upside from improved acquisition efficiency and reduced churn.

The planning use case: Do not hire to the optimistic scenario. Do not cut to the conservative scenario. Hire to a plan that breaks even in the base scenario and accelerates in the optimistic scenario. If the conservative scenario materializes, you have 3–6 months of early warning to adjust.

For the full three-scenario financial model that extends this to P&L and runway, see SaaS financial model template.

Common ARR Forecasting Mistakes

Mistake 1: Ignoring the Churn Tail

Described in the introduction: applying a flat churn rate to a growing MRR base without recognizing that the compounding churn drag accumulates over time. A $200K MRR company with 2% monthly churn loses $48K of ARR in the first year from customers who were active at forecast start — separate from future churn on newly acquired customers.

Fix: Build the MRR schedule monthly, calculating churned MRR as a percentage of starting MRR in each period. Do not subtract churn once from starting MRR — subtract it every month from that month's starting MRR.

Mistake 2: Overweighting Pipeline in 12-Month Forecasts

Current pipeline is a reliable input for the next 30–60 days of new ARR. It is not reliable for Month 7–12 projections because most of that pipeline does not exist yet. Founders who build 12-month ARR forecasts by projecting current pipeline coverage forward are effectively assuming that today's pipeline converts fully — ignoring that deals expire, lose to competition, and require continuous refilling.

Fix: Use pipeline for near-term accuracy (next 60 days); use acquisition model for medium-term forecast (months 3–12).

Mistake 3: Not Modeling Expansion ARR Separately

Many founders' ARR forecasts include expansion only implicitly (in the net churn rate). But expansion MRR is a different motion from retention — it requires different inputs (CS headcount, expansion playbook maturity, product surface area for upsell) and different assumptions.

Fix: Model expansion ARR as a separate line item with its own rate assumption. This forces you to think explicitly about whether your CS team has the capacity and playbook to generate the expansion forecast.

Mistake 4: Using a Single Churn Rate for All Cohorts

Newer cohorts churn faster. Applying one blended churn rate to all cohorts produces ARR overestimates in high-growth periods (when newer, higher-churn cohorts are a larger proportion of the base).

Fix: If you have cohort retention data (from Tab 4 of the spreadsheet), segment the churn rate by cohort age: 0–3 months, 4–6 months, 7–12 months, 12+ months. Apply different rates to each cohort in the MRR schedule.

Connecting the ARR Forecast to Operating Decisions

An ARR forecast is only useful if it drives decisions. The key questions:

Hiring: Given the base-case 12-month ARR trajectory, what is the maximum headcount we can add while maintaining Burn Multiple below 1.5x? Use the SaaS financial model to stress-test hiring plans against the conservative scenario.

Fundraising timing: If the base-case trajectory puts you at $4M ARR in 12 months, and the conservative case puts you at $3.2M, your fundraising should target a round that does not require $4M ARR to justify. Raise when the conservative scenario still produces a fundable outcome.

Channel investment: If the optimistic scenario requires acquisition rates 30% above current — what would it take to achieve that? More SDRs? Higher paid search budget? Content compounding from last year's investment? The forecast makes the question specific.

Burn Multiple tracking: Your ARR forecast provides the denominator for Burn Multiple planning. If Net New ARR in Month 6 is forecasted at $1.2M (annualized), and you want to maintain a 1.5x Burn Multiple, maximum net burn = $1.2M × 1.5 / 12 = $150K/month. This sets a hard ceiling on operating expense growth.

Reconciling Forecast to Actuals

At the end of each month, compare the forecast to actuals across the four components:

A -$3,400 miss on net new MRR is 37% below forecast — significant. But the component breakdown tells you exactly where it came from: paid search conversion down, contraction up, churn up. Each has a different diagnosis and intervention.

This reconciliation discipline, done monthly, turns the ARR forecast from a plan document into an operating feedback loop.

Connecting the Forecast to Your Metrics Dashboard

The ARR forecast provides the forward view. The B2B SaaS KPI dashboard provides the current view. Together:

• Daily pulse: Am I on track this week vs. this month's forecast?
• Weekly review: Is the month-to-date actuals pace matching the forecast trajectory?
• Monthly close: Full variance analysis — which components hit, which missed, why?

For the ARR vs MRR distinction that underpins the forecast methodology, and the NRR calculation that validates retention assumptions, these guides provide the definitional foundation.

Use the calculator to model specific scenarios — what happens to 12-month ARR if you improve trial conversion from 25% to 35%? If churn drops from 2% to 1.5%? The calculator handles the compounding math instantly.

Conclusion

A bottom-up ARR forecast for a $10K–$500K MRR company is a monthly MRR schedule built from four components: new customer acquisition (from the funnel model), expansion (from the CS motion), contraction (from the historical rate), and churn (cohort-adjusted, not flat rate).

The work is in getting the inputs right — using actual conversion rates, actual cohort churn data, and realistic ramp assumptions for new channels. The result is a forecast you can use to make hiring decisions, set Burn Multiple targets, and time your fundraise — rather than a growth rate extrapolation that tells you what you want to believe.

Build it. Update it monthly. Reconcile it to actuals. And when the variance appears — and it will — use the component breakdown to diagnose whether it is a GTM problem, a retention problem, or an expansion problem. That is how an ARR forecast becomes an operating system.

Frequently Asked Questions

How do you forecast ARR for a SaaS company?

Build a monthly MRR schedule for 12–18 months: start with current MRR, add projected new MRR from your acquisition model, add projected expansion MRR from existing customers, subtract projected churn and contraction MRR using cohort-based churn rates, and multiply ending MRR by 12 to get forward ARR. This is the bottom-up method — more work than a growth rate extrapolation, but 3–5x more accurate.

What is the difference between ARR and an ARR forecast?

ARR (Annual Recurring Revenue) is a point-in-time metric: current MRR × 12. An ARR forecast is a projection of what ARR will be at a future date based on modeled acquisition, churn, and expansion assumptions. ARR is what you have; the ARR forecast is what you will have if your model inputs hold.

How accurate should a 12-month ARR forecast be?

A well-built bottom-up ARR forecast for a $50K–$500K MRR company should come within 15–20% of actual at 12 months. The main sources of error are churn rate changes (which compound over 12 months), unexpected new customer acquisition acceleration or deceleration, and large single-account events (one big churn or one big expansion). At 6 months, accuracy should be within 10%.

What is the biggest mistake in SaaS ARR forecasting?

Ignoring the churn tail. A company at $200K MRR with 2% monthly churn will lose approximately $40K MRR per month from existing customers (declining as the base shrinks). Over 12 months, the cumulative churn drag is $350K+ of ARR that was on the books at the start of the forecast period. Founders who do not model this explicitly dramatically overestimate ending ARR.