SaaS Revenue Forecasting: A Step-by-Step Model

SaaS revenue forecasting predicts recurring revenue from the customers you already have, the customers you expect to add and the customers you expect to lose. Because subscription revenue repeats, the forecast is built as a roll-forward of customers or monthly recurring revenue (MRR), not as a growth rate applied to last year's sales. This guide builds a six-month customer and MRR model with arithmetic you can verify, shows the MRR waterfall that investors and boards expect, covers churn, expansion and unit economics such as lifetime value and payback, and explains when to move from a simple model to cohorts. It is part of budgeting and forecasting and an application of how to forecast revenue.

The core idea: a roll-forward

Each month follows the same identity:

Closing customers = Opening customers - Churned customers + New customers
Closing MRR = Opening MRR + New MRR + Expansion MRR - Contraction MRR - Churned MRR

Forecast the drivers (new customers, churn rate, price, expansion), and the revenue follows. This is easy to audit because each line has a clear definition and an owner: marketing and sales own new customers, customer success owns churn and expansion, and finance owns the model.

Step 1: Define the terms

Write definitions down and apply them consistently. A common source of confusion is mixing billing, revenue recognized and contract value. Check revenue recognition rules with your accountants.

Step 2: A six-month customer model

Assume a hypothetical company starts with 500 customers at an ARPA of $100 per month (MRR of $50,000), loses 2 percent of opening customers each month and adds 25 new customers each month.

Check month 1: 500 - (500 × 0.02) + 25 = 500 - 10 + 25 = 515. Check month 2: 515 - 10.3 + 25 = 529.7. Fractional customers are a modeling convenience; the MRR column is calculated from unrounded customers. MRR grows from $50,000 to $58,562, an increase of $8,562 or 17.1 percent in six months (8,562 / 50,000), even though 2 percent churn removes about 10 customers a month. The steady 25 new customers more than offset churn now, but as the base grows, churn grows with it and growth slows. At 1,250 customers, 2 percent churn removes 25 a month, which equals the new customers, and the base stops growing. That break-even point is churn rate divided into new customers: 25 / 0.02 = 1,250.

Step 3: Annual view of churn

Monthly churn compounds. A 2 percent monthly churn rate means:

Share of a customer cohort remaining after 12 months = (1 - 0.02)^12 = 0.785

So about 78.5 percent of a cohort remains and 21.5 percent is lost within a year. A monthly rate that looks small is a large annual loss. State churn in both forms so decision-makers see it.

Step 4: The MRR waterfall

Boards prefer to see revenue movements separated by cause. A hypothetical month:

Check: 50,000 + 2,500 + 500 - 250 - 1,000 = 51,750. Net new MRR = $1,750. Net revenue retention for the month = (50,000 + 500 - 250 - 1,000) / 50,000 = 49,250 / 50,000 = 98.5%. If expansion were large enough that NRR exceeded 100 percent, the installed base would grow even with no new customers.

Forecast each line: new MRR from the sales and marketing plan, expansion and contraction from history by segment, and churn from retention history.

Step 5: Unit economics

Recurring revenue businesses are judged on the cost of acquiring customers relative to their value.

(Illustrative: an 80 percent gross margin and a CAC of $600 are assumptions, not benchmarks.) These formulas assume constant churn and margin, which is a simplification; use them for comparison and direction, not precision. Check that gross margin includes hosting, support and payment fees. See unit economics.

Step 6: Build it in a spreadsheet

Layout:

1. Assumptions: opening customers, ARPA, new customers per month (by channel if possible), churn rate, expansion and contraction rates, price changes with dates, gross margin.
2. Monthly roll-forward: formulas as above, one column per month.
3. MRR waterfall and ARR: summarize by month and quarter.
4. Cash view: billing terms (monthly, annual in advance) change cash timing. Annual prepayment brings cash in earlier than revenue. See cash flow management.
5. Checks: closing balance equals next opening balance; customers never negative; waterfall sums.

Where the new customers come from

The 25 new customers a month in the model need a source. For a self-serve product, the usual path is a trial funnel, which works the same way as the ecommerce funnel in how to forecast ecommerce sales. Suppose the company starts 250 trials a month and 10 percent convert to paid: 250 × 0.10 = 25 new customers. With a CAC of $600, the acquisition spend is 25 × $600 = $15,000 a month, or $60 per trial. Now the forecast has a lever: doubling trial starts without changing conversion would double new customers, but only if the extra traffic converts as well as the existing traffic, which usually it does not. For a new plan or tier with no history, borrow conversion assumptions from the closest existing plan and test them, as described in forecasting sales for a new product.

Modeling a price increase

Price changes are one of the biggest levers in a subscription model, and one of the easiest to forecast badly. Suppose the company raises the price by $10 (10 percent) for 40 percent of its 500 customers, and expects 1 point of extra churn among the affected customers.

Net MRR rises by $1,780 a month, about 3.6 percent of the $50,000 base. The result is highly sensitive to the churn assumption: if extra churn were 8 points rather than 1, the loss would be 200 × 8% = 16 customers × $110 = $1,760, wiping out nearly the whole gain. Stress the churn assumption before deciding.

Refinements

Move to cohorts when churn clearly varies with customer age, when you have enough history (typically at least a year of monthly cohorts) and when the decision justifies the effort.

Presenting the forecast to investors and the board

• Show base, low and high cases, with the assumptions that differ.
• Show the MRR waterfall, not just total MRR.
• Show how forecast accuracy has been, and how the assumptions came from history. See how to measure financial forecast accuracy.
• Reconcile to cash and runway.
• Be consistent in definitions across periods.

Common mistakes

• Applying a growth rate to MRR instead of modeling churn and new customers.
• Forgetting that churn grows as the base grows.
• Quoting monthly churn without the annual effect.
• Counting bookings or contract value as revenue.
• Ignoring expansion and contraction.
• Computing LTV with an optimistic churn rate.
• Mixing customer churn and dollar churn.
• Not separating new customer acquisition costs from retention costs.

FAQs

How do you forecast SaaS revenue?

Roll forward customers or MRR each month: opening, plus new, plus expansion, minus contraction, minus churn, and multiply customers by ARPA.

What is MRR?

Monthly recurring revenue: the monthly value of active subscriptions.

How do I forecast churn?

Use your own history by segment and tenure, convert monthly rates to annual to see the full effect, and test sensitivity.

What is net revenue retention?

Opening MRR plus expansion, minus contraction and churn, divided by opening MRR for a cohort of customers.

When should I use cohorts?

When churn depends on how long a customer has been with you and you have enough history to measure it.