TL;DR: Most SaaS companies measure churn wrong and forecast it worse. They track logo churn when revenue churn matters more, they average across all customer segments when behavior varies wildly, and they treat churn as random when it’s actually predictable. Accurate churn forecasting requires cohort analysis, leading indicators that predict cancellation 30-60 days early, and segmentation showing which types of customers leave and when. Get this right and you’ll see revenue problems months before they hit your P&L.
We’ve watched dozens of SaaS companies confidently project 36-month revenue growth based on their “stable” 3% monthly churn rate, only to watch it spike to 6% eighteen months in and destroy their entire financial model.
Churn is the silent killer in SaaS. New logo growth is visible and celebrated. Marketing spend is tracked obsessively. Sales pipeline gets reviewed weekly. But churn lives in the background, quietly eroding everything you’ve built.
A SaaS company adding $100K in new MRR monthly looks like it’s growing fast. If churn is eating $60K monthly, net growth is $40K. If churn spikes to $90K for three months due to a product issue or economic downturn, net growth goes to $10K and suddenly the company is scrambling to understand what happened.
The companies that forecast churn accurately can see these problems coming and either fix them or adjust growth plans before they’re forced to lay people off or miss board projections.
Most SaaS companies start by tracking logo churn (the percentage of customers that cancel each month). This is easy to measure and easy to understand. If you have 1,000 customers and 30 cancel, that’s 3% monthly logo churn.
The problem is logo churn doesn’t tell you about business impact. If your 30 cancellations were all $50/month customers and you had 5 expansions from $500/month to $2,000/month customers, your revenue actually grew despite the logo losses.
What you actually need to track is net revenue retention: starting MRR from a cohort, plus expansions, minus contractions and churn, divided by starting MRR. A healthy SaaS company has net revenue retention above 100%, meaning the customers you already have are growing faster than they’re churning.
Calculate this monthly for each customer cohort. The January 2024 cohort started with $50K MRR. By July, they’re at $54K after expansion and churn. That’s 108% NRR for that cohort. Track this across all cohorts to see if retention is improving or degrading over time.
Here’s what this looks like in practice. A Series B SaaS company we worked with had 2.5% monthly logo churn and thought they were doing fine. When we calculated revenue retention by cohort, we discovered a problem. Their small customers (under $200/month) were churning at 5% monthly. Their mid-market customers ($500-2K/month) churned at 1.8% and expanded 20% annually. Their enterprise customers (over $5K/month) had 0.5% churn and expanded 35% annually.
Knowing this completely changed their strategy. They stopped spending acquisition dollars on small customers, doubled down on mid-market, and built enterprise features to accelerate that segment. Revenue retention improved from 98% to 112% within a year.
Aggregate churn numbers lie. They average together fundamentally different customer behaviors and give you a single number that masks what’s actually happening.
Cohort analysis means tracking each group of customers that signed up together and watching how they behave over time. The March 2024 cohort has different characteristics than the September 2023 cohort. Maybe you changed pricing. Maybe you targeted a different segment. Maybe your product improved. All of that shows up in cohort retention curves.
Build a cohort retention table showing every signup month and their retention at month 1, 3, 6, 12, 24. You’ll immediately see patterns. Maybe there’s a spike in churn at month 4 for most cohorts (onboarding problem). Maybe newer cohorts retain worse than older ones (product-market fit is degrading). Maybe there was a shift in month 18 where retention improved (you fixed something important).
We worked with a company that couldn’t understand why their churn was creeping up despite improving their product. Cohort analysis revealed the answer: customers acquired through paid channels churned at 2x the rate of customers from organic channels. As they’d scaled paid acquisition, they’d brought in worse-fit customers who looked the same initially but had much higher long-term churn. They adjusted targeting and improved paid channel churn by 40%.
Cohorts also show you the shape of your retention curve. Most SaaS products have high early churn in months 1-3 as bad-fit customers self-select out, then flattening retention after month 6 once customers are onboarded and getting value. If your curve doesn’t flatten, you have ongoing product or value issues. If it has secondary churn spikes at specific months, something in the customer journey is broken.
By the time a customer cancels, you’ve lost them. The only way to reduce churn is catching warning signs early when you can still intervene.
Build a predictive model based on customer behavior patterns. Track metrics like:
– Days since last login
– Feature usage depth (are they using core features or just surface tools?)
– Support ticket velocity (increased tickets often predict churn)
– Payment failures (failed credit cards lead to involuntary churn)
– Contraction signals (reduced seats, downgraded plan)
– Team member changes (if their champion leaves, churn risk spikes)
A well-built health score combines these into a single metric that predicts 30-60 day churn risk. Customers with health scores in the bottom quartile might have 15% monthly churn while top quartile customers have 0.5% churn.
The specific indicators vary by product, which is why you need to analyze your own data. Pull a list of customers who churned in the last 6 months, then look at their behavior in the 60 days before cancellation. What patterns emerge?
One client discovered that enterprise customers who didn’t schedule a QBR meeting in the first 90 days had 8x higher churn in months 4-12. They made QBR scheduling a required onboarding step and cut enterprise churn by 35%.
Another found that customers who hadn’t invited a second team member by day 30 churned at 3x the rate of customers with multiple users. They built prompts encouraging team invitations and reduced single-user churn by half.
These insights only come from analyzing your actual customer data, not from best practices you read in a blog post.
Your aggregate churn number averages together customers with nothing in common. Small businesses behave differently than enterprises. Annual contracts churn differently than monthly. Customers acquired through different channels, sold different products, in different industries, with different use cases all have different retention profiles.
Segment your churn analysis at minimum by:
– Customer size (MRR bands)
– Contract type (monthly vs annual)
– Acquisition channel
– Time since signup (month 1-3 vs month 12+ vs month 36+)
Better if you can add:
– Industry vertical
– Product tier or package
– Implementation type (self-serve vs onboarded)
– Geography
A healthy mid-market SaaS company might see:
– Months 1-3: 5% monthly churn (initial fit sorting)
– Months 4-12: 2% monthly churn (early customer development)
– Months 13-24: 1% monthly churn (mature customers)
– Months 25+: 0.5% monthly churn (sticky long-term accounts)
Annual contracts: 0.8% monthly churn equivalent
Monthly contracts: 3.5% monthly churn
Enterprise segment: 0.4% monthly churn with 25% annual expansion
SMB segment: 4% monthly churn with 5% annual expansion
When you see churn this way, you can forecast it accurately by modeling how customer mix shifts over time. If you’re adding more monthly contracts and fewer annual, churn will increase. If you’re moving upmarket, churn should improve.
Here’s how to build a forecast that’s actually useful:
Start with cohort retention curves. For each historical cohort, calculate their retention rate at each month after signup. Average these to create expected retention curves by segment.
Example: customers acquired through content marketing retain at 95% month 1, 92% month 3, 88% month 6, 85% month 12. That’s your content marketing retention curve.
Apply these curves to your signup forecast. If you’re planning to add 100 customers through content marketing in January, you can forecast that 95 will still be there in February, 92 in April, 88 in July.
Do this for each segment and acquisition channel. Sum them up to get total expected MRR by month.
Include expansion assumptions based on historical net revenue retention. If your enterprise segment expands 25% annually, model that growth on top of your base retention.
Now you have a bottom-up forecast that accounts for how different customer segments actually behave rather than assuming a constant churn rate.
Run sensitivity analysis on your assumptions. What happens if SMB churn increases from 4% to 6%? What if enterprise expansion slows from 25% to 15%? Model these scenarios so you know which metrics matter most and where you need early warning systems.
Involuntary churn (failed payments, expired credit cards) typically accounts for 20-40% of total churn in monthly billing SaaS companies. This is churn from customers who weren’t trying to leave, their payment method just failed.
The good news is involuntary churn is solvable. Implement dunning workflows that automatically retry failed payments, email customers about expiring cards, and offer alternate payment methods. Companies that do this well recover 60-70% of failed payments.
Track involuntary churn separately from voluntary churn. A company showing 3.5% total churn might have 1.5% involuntary and 2% voluntary. If you can cut involuntary churn in half through better payment recovery, your total churn drops to 2.75% without changing anything about product or customer success.
This is one of the highest-ROI improvements you can make because it’s purely operational, not dependent on product development or changing customer behavior.
A well-run SaaS company has a monthly meeting where they review:
– Cohort retention tables showing retention trends
– Health score distribution across the customer base
– Leading indicator metrics (login frequency, feature usage, support patterns)
– Segmented churn by customer type
– Actual churn vs forecast with variance analysis
When actual churn exceeds forecast, they investigate immediately. Is it localized to a segment (one customer type behaving differently)? Is it a cohort issue (recently acquired customers have higher churn)? Is it a product problem (bug or missing feature driving cancellations)?
They forecast churn bottoms-up using segment-specific retention curves and update forecasts monthly as they see leading indicators shift.
They’ve identified the 3-5 metrics that predict churn in their business and they track those religiously. When health scores decline or warning indicators spike, customer success intervenes before customers leave.
Most importantly, they treat churn reduction as equally important to new customer acquisition because they understand that in a subscription business, keeping customers compounds into massive value over time.
Small improvements in churn create surprisingly large valuation changes because of compounding. A SaaS company with 3% monthly churn loses 50% of customers over 24 months. Drop that to 2% monthly churn and you retain 61% instead of 50%, a 22% improvement in 2-year retention.
That 22% retention improvement flows through to LTV, which affects your ability to spend on acquisition, which affects growth rate, which affects valuation. A company worth $50M at 3% churn might be worth $75M at 2% churn, all else equal.
This is why accurate churn forecasting matters. If you can see churn trending up six months before it destroys your growth rate, you can fix it. If you’re surprised when monthly churn suddenly hits 5% because you were only watching aggregate numbers, you’re in crisis mode instead of prevention mode.
—
Q: What’s a “good” churn rate for SaaS companies?
Depends entirely on your segment and business model. SMB monthly billing products typically see 3-5% monthly logo churn, mid-market annual contracts see 10-15% annual churn, enterprise products might see 5-8% annual logo churn. What matters more than the absolute number is the trend (is it improving?) and net revenue retention (are churned customers offset by expansion?). A company with 5% logo churn but 115% net revenue retention is healthier than one with 2% logo churn and 95% NRR.
Q: How far in advance can we accurately forecast churn?
With good leading indicators, you can predict 30-60 day churn risk pretty accurately at the individual customer level. At the aggregate level, you can forecast 6-12 months out based on cohort behavior and customer mix changes, though confidence intervals get wider beyond quarter 2. The forecast accuracy depends on how stable your business is. If you’re changing products, pricing, or target segments, historical patterns matter less and you need to weight recent data more heavily.
Q: Should we include churn reduction in our revenue forecasts?
Only if you have specific initiatives with measurable expected impact. Don’t just assume “we’ll reduce churn by 20%” without explaining how. If you’re implementing a new onboarding process that reduced churn by 30% in a pilot group, you can model that rolling out. If you’re adding a feature that existing customers requested before churning, you can estimate impact based on how many customers cited that as a cancellation reason. But generic “we’ll improve retention” assumptions without concrete plans are wishful thinking.