Short answer: The top 5 MRR forecasting mistakes are: 1) Using a single average churn rate instead of cohort-based, 2) ignoring expansion revenue, 3) treating new sales as linear, 4) neglecting seasonality and macro trends, and 5) failing to update forecasts with actuals. Fixing these builds a reliable forecast.
Key takeaways
- Use cohort-based churn not averages.
- Include expansion revenue in forecast.
- Model new sales with growth curve.
- Account for seasonality and macro events.
- Re-forecast monthly with actual data.
- Shorten forecast horizon to reduce error.
What you will find here
Your MRR forecast sets your hiring plan, your marketing budget, and your runway. One wrong assumption and you spend money you don’t have, or leave growth on the table. I’ve seen SaaS teams make the same five mistakes over and over. Here they are, and how to fix them.
1. Using a Single Average Churn Rate
Most startups calculate churn as total lost MRR divided by starting MRR. That’s a mistake. Average churn hides huge differences between cohorts. Customers who signed up during a high-spend promotion might churn faster than your core base. If you project forward with the average, you’ll overestimate retention.
The fix: Build cohort-based churn analysis. Group customers by month of acquisition. Calculate the churn rate per cohort per month. Use the specific cohort’s churn curve in your forecast. For a deeper dive on metrics that underpin this, check out Beginner’s Guide to SaaS Revenue Metrics.
2. Ignoring Expansion Revenue
Many SaaS teams forecast only new sales and churn. They forget that existing customers upgrade, add seats, or buy add-ons. Expansion MRR can be 30% or more of total new revenue for mature companies. Ignore it and you consistently underestimate your future MRR.
The fix: Track expansion MRR separately. Calculate a net dollar retention rate (NDR). Use NDR to project expansion on the existing base. If your NDR is above 100%, your base grows even without new sales. This is a powerful force in your forecast.
3. Treating New Sales as Linear
New sales rarely follow a straight line. Early growth is lumpy. A big enterprise deal in month 3 can warp your projection. Linear extrapolation makes you miss both surges and slumps. You end up with a forecast that looks smooth but is wrong every month.
The fix: Use a bottom-up forecast. Break your pipeline into stages with conversion rates. Weight each deal by probability. For new customers, model a growth curve based on historical ramp patterns. Be explicit about your assumptions for each segment.
4. Neglecting Seasonality and Macro Trends
Example: Many B2B SaaS companies see Q4 dips and Q1 surges. Others get spikes in back-to-school months. If your forecast is a flat monthly number, you will miss these patterns. Worse, you ignore macroeconomic shifts like a recession that slow enterprise buying cycles.
The fix: Go back at least 12 months of data. Calculate monthly seasonality indices. Factor in known events (product launches, pricing changes). For macro trends, use external benchmarks like SaaS industry reports. Build downside scenarios to stress-test your forecast.
To improve your accuracy, also revisit your ARR calculation — it’s the foundation of your MRR projection.
5. Setting and Forgetting Your Forecast
A forecast is a living document. Many teams build an annual forecast in January and never change it. Reality diverges, but the budget stays fixed. This leads to bad decisions all year.
The fix: Re-forecast monthly. Compare actual MRR to forecasted MRR. Update assumptions for churn, expansion, and new sales. Shorten your forecast horizon — 12 months out is less reliable than 3 months. Roll forward every month with the latest data.
For a clear breakdown of MRR vs ARR and when to use each, see ARR vs MRR: Key Differences Every SaaS Founder Must Know.
How to Build a Better MRR Forecast
Here’s a step-by-step process to avoid these mistakes:
- Collect historical data — at least 12 months of MRR by type (new, expansion, churn, contraction).
- Calculate cohort-based churn for each acquisition month. Use that in your base retention projection.
- Compute expansion rate from past 6 months. Apply to the existing base, reduced by churn.
- Build a pipeline-driven new sales model with weighted stages. Don’t just average past sales.
- Add seasonality factors and macro adjustments. Create base, optimistic, and pessimistic scenarios.
- Re-forecast monthly. Compare actuals, update assumptions, and communicate changes to the team.
This process takes work, but it beats flying blind.
Common Questions About MRR Forecasting
How often should I re-forecast? Monthly, at minimum. If your business is volatile, do a mid-month check.
What’s a good forecast error range? Under 10% for the next quarter is great. Over 20% signals you need to fix your model.
Should I include one-time fees? No. MRR is recurring. Book one-time revenue separately to avoid distorting your forecast.
How far out should I forecast? 12 months is the standard. But treat months 9-12 as directional, not precise.
What if I don’t have enough history? Start with industry benchmarks as placeholders. Replace them with your own data as soon as you can.
Frequently asked questions
How often should I re-forecast MRR?
Re-forecast monthly at minimum. If your business is volatile — for example, you rely on large enterprise deals — do a mid-month check. Compare actual MRR to forecasted MRR, update churn and expansion assumptions, and communicate changes to the team. A rolling 12-month forecast updated every month keeps you agile.
What is a good forecast error range for MRR?
A forecast error under 10% for the next quarter is excellent. Under 15% is good. Over 20% indicates your model or data needs work. Track mean absolute percentage error (MAPE) monthly. If error grows in later months, shorten your forecast horizon.
Should I include one-time fees in MRR forecast?
No. MRR is recurring revenue only — subscriptions and recurring add-ons. One-time fees (setup, consulting) distort the recurring picture. Book them separately in your revenue model. Forecasting them as MRR inflates your baseline and leads to bad decisions.
How far out should I forecast MRR?
12 months is the standard for SaaS companies. But treat months 9–12 as directional, not precise. Your 3-month forecast should be accurate within 10–15%. Use a shorter horizon for operational decisions (hiring, spend) and longer for strategic planning.
What if I don’t have enough history for cohort churn?
Start with industry benchmarks for churn rates by month of subscription. For example, use a 5% monthly churn for month 1, declining to 3% by month 6. Replace these with your own data as soon as you have at least 6 months of cohort history. Benchmarks are a starting point, not a permanent solution.