Mastering SaaS Due Diligence: How to Decode Cohort Analysis and Retention Curves

Deal Alert AI is reader-supported. We earn commissions from affiliate links at no cost to you.

This post is based on a video from our Deal Alert AI YouTube channel. Watch the original or read the full breakdown below.

Buying a SaaS business is fundamentally different from buying an e-commerce store or a content site. In e-commerce, you are often buying inventory and marketing channels. In SaaS, you are buying a churn rate and a growth curve. If you cannot distinguish between a healthy, compounding user base and one that is dying by inches, you will overpay for a liability. I see this mistake constantly in the marketplace. Buyers are thrilled to see $50,000 in Monthly Recurring Revenue (MRR), but they fail to ask the right questions about where that revenue is coming from and how fast it is leaving.

When I review deals for my clients or evaluate opportunities on platforms like Empire Flippers or Flippa, the first spreadsheet I pull up is not the P&L statement. It is the cohort analysis. This data structure allows you to see the behavior of specific groups of users who signed up within a defined time frame. Without this, you are flying blind. You are judging a car by its paint job while ignoring the engine diagnostics.

At Deal Alert AI, we build algorithms to parse this type of data precisely because it is so critical and so often misunderstood. Today, we are going to dive deep into the mechanics of cohort analysis and retention curves. We are going to strip away the jargon and look at the raw numbers that determine whether a SaaS business is worth 4x its annual profit or 10x. This is technical, but it is also the most valuable skill you can learn as a buyer. If you understand how to read these curves, you will never be the sucker on the end of a deal again.

Understanding the Basics of Cohort Analysis

A cohort is simply a group of people who share a common characteristic or experience within a defined period. In the context of SaaS due diligence, the standard cohort is based on the month (or week) a customer first signed up. For example, "Cohort A" consists of all customers who purchased the software in January 2023. "Cohort B" consists of all customers who purchased in February 2023. By tracking these groups over time, you can isolate the performance of each vintage of customers from one another.

This isolation is critical because it prevents distortion from new sign-ups and cancellations. If you only look at the "current" active user base, you are mixing old, loyal users with brand new, unstable ones and a subset of churned users. This creates a noisy data set. Cohort analysis removes the noise. It shows you the survival rate of a specific group as it ages. If January users have a 90% survival rate at month 3, but February users only have a 50% survival rate at month 3, you have a major signal. The product, the marketing, or the onboarding process changed significantly between those two months.

When you request data during due diligence, you must ask for a raw cohort export, not just the dashboard view. Dashboards can be tweaked to hide bad data. Raw exports in CSV format cannot lie. You need to see every customer ID, their first purchase date, and their last active date or cancellation date. From this raw data, you can build your own retention matrix. This ensures the seller has not manipulated the definition of "active user" to inflate their numbers. It is the bedrock of honest due diligence.

Reading the Retention Curve: What It Looks Like

Once you have calculated the retention rate for each cohort, you plot them on a retention curve. The X-axis represents the time since the customer signed up (Month 1, Month 2, Month 3, etc.), and the Y-axis represents the percentage of original customers still active. A perfect retention curve is flat, meaning 100% of customers stay forever. This is rare. Most SaaS curves slope downward, indicating churn. The question is not "does it slope down?" but rather "how fast does it slope down, and does it plateau?"

There are two distinct types of declining curves that every buyer must recognize. The first is the "cliff" curve. In this scenario, retention drops dramatically in the first one or two months, and then it stabilizes at a lower level. For example, a cohort might retain 60% of users in Month 2 and 55% in Month 3. This suggests that the product has high initial confusion or poor onboarding, but users who survive the initial learning curve become loyal. This is manageable. You can fix onboarding. You can improve the first user experience.

The second type is the "decay" curve. Here, retention drops consistently every single month. Month 1: 100%. Month 2: 90%. Month 3: 81%. Month 4: 72%. This indicates that users are not forming a habit. They use the product for a short burst and then leave. In this case, the product is likely not solving a core pain point, or the value proposition evaporates after the initial trial. Fixing this is extremely difficult. It often requires rebuilding the core product. As a buyer, you are not a product team. You are a cash flow investor. You want a flat line, not a steep slide.

Identifying Churn Leaks and Timing Patterns

Retention is a lagging indicator. You see the slope, but you do not immediately see the cause. To understand the "why," you must look at the timing of cancellations. In my experience analyzing hundreds of SaaS businesses, churn usually happens in specific windows. If you see a spike in cancellations at Day 14, it often correlates with the end of a 14-day free trial. If you see a spike at Month 3, it might correlate with the end of a discounted introductory period. Identifying these windows allows you to pinpoint the specific friction points in the customer journey.

Consider a project management tool for small businesses. The data might show that 80% of users are active at Day 7, but only 40% are active at Day 30. This suggests that users sign up, create a few tasks, and then never return. The product may be brilliant, but the "Aha!" moment is not occurring fast enough. The user needs to see value immediately. If the value proposition requires setup, configuration, or team adoption, and the small business owner finds that too much work, they churn. This is a behavioral economic issue. It is not a code bug. It is a value delivery failure.

You must also look for "revocation" vs. "cancellation." In corporate sales, we often see negotiated decreases in seats. This is not true churn; it is contract change. However, in B2C or SMB SaaS, cancellation is permanent. You need to separate these. If a seller is averaging out "contract adjustments" with "user cancellations," they are masking the true churn rate. A business where 50% of customers cancel every year is fundamentally different from one where 10% cancel and 40% downgrade. The former is a consumption product. The latter is a subscription product. The valuation methodology changes entirely based on this distinction.

Calculating Lifetime Value (LTV) from Cohort Data

The ultimate goal of cohort analysis is to calculate the Lifetime Value (LTV) of a customer. LTV is the total revenue a customer generates over their entire relationship with the company. In due diligence, you cannot wait 5 years to know the LTV. You have to estimate it based on the retention curve. The most common method is to look at the retention rate at the longest observed cohort and assume it stabilizes from there. If your oldest cohort is 12 months old and shows a retention rate of 20%, and Month 10 was 25% and Month 11 was 22%, you can model that the retention will settle around 20-21% in infinite time.

Once you have that stabilized retention rate, the formula for LTV is straightforward: LTV = (Average Monthly Revenue per User) / (Churn Rate). Let’s say the Average Monthly Revenue (ARMR) is $50. If the monthly churn rate is 5%, the LTV is $50 / 0.05 = $1,000. This means that for every customer you acquire, you will earn $1,000 in total over their lifetime. If it costs you $100 in marketing to acquire that customer (Customer Acquisition Cost or CAC), your LTV:CAC ratio is 10:1. This is a heaven-sent ratio. It means the business is immensely profitable. However, if the churn rate is 15%, the LTV drops to $333. Your 10:1 ratio becomes 3.3:1, and the margins thin significantly. The sensitivity of LTV to churn is exponential.

Here is where the trap lies. Sellers often calculate LTV based on the average churn rate of all customers, including the long-tail loyal users who have been there for 5 years. This inflates the LTV. You must calculate LTV based on the *new* cohort behavior. If new customers are churning faster than old ones, the future LTV will be lower than the historical LTV. This is why recent cohort data is more valuable than historical data. A business that had great retention in 2018 but is losing new users in 2024 is declining, even if the overall database still looks healthy. You are buying the future, not the past.

The Logic Churn vs. Process Churn Dilemma

Not all churn is created equal. When analyzing retention curves, you must determine if the drop-off is due to "logic churn" or "process churn." Logic churn occurs when the customer decides the product is no longer useful, too expensive, or a better alternative exists. This is a market-driven phenomenon. It is the reason customers leave. Process churn, on the other hand, occurs due to friction in the billing, login, or usage experience. It is avoidable. It is a failure of operations.

If you see a regular, steady drop in retention that does not correlate with specific spikes in support tickets or billing failures, it is likely logic churn. The market has moved on. The product has become a commodity. In this case, the retention curve is a reflection of the macro environment. You cannot fix logic churn with a patch. You can only fight it with price or added features. However, if you see erratic drops that correlate with server outages, email delivery failures, or credit card declines, that is process churn. Process churn is a debt. It is fixable. If you buy a business with high process churn, your first task is not to launch a new feature. It is to fix the infrastructure and the billing pipeline. This can dramatically improve retention without changing the product at all.

Distinguishing between these two is vital for pricing. If the churn is primarily logical, the risk belongs to the buyer. You are betting on the product's staying power. If the churn is primarily process-related, the risk is operational. You can de-risk this by hiring a CTO or an operations manager to fix the leaks. In many deals I have seen on Flippa, the seller claims the product is bad, but the data shows they just have terrible email deliverability. Fixing the email server can boost retention by 10-15% instantly. That is free money for the buyer. You just have to find it in the data.

Visualizing Expansion and Net Revenue Retention

Retention is not the whole story. In SaaS, you also need to look at Net Revenue Retention (NRR), formerly known as Gross Retention. NRR measures whether existing customers are spending more or less over time. If a customer stays active but downgrades from a $100/month plan to a $50/month plan, the cohort retention percentage might look stable, but the revenue retention has dropped. This is a critical distinction. A 100% user retention rate with 50% revenue retention is a dying business.

You need to build a revenue cohort matrix, not just a user cohort matrix. For each month, track the total revenue generated by the customers who signed up in that month. Then, track that revenue in subsequent months. You will see a curve similar to user retention, but with a crucial twist. In healthy SaaS businesses, this curve can actually go *up* for the first 6-12 months. This is due to upsells, cross-sells, and seat expansions within existing teams. If the revenue per user in Month 6 is higher than in Month 1, the business is compounding. This is the "flywheel" effect.

If the revenue per user stays flat or declines over time, even if user count remains stable, you have a problem. It means the customers are not finding reasons to spend more. They are using the product minimally. This caps your growth potential. You can buy more users, but their value will never increase. When evaluating a deal, I look for NRR greater than 100%. This means the existing base is growing in revenue even without new sales. If NRR is below 100%, the business must constantly acquire new customers just to stay flat. If NRR is above 110-120%, the business can grow exponentially with minimal marketing spend. This is the difference between a sprint and a marathon.

Building Your Due Diligence Checklist for SaaS

Knowing the theory is one thing, but executing the analysis under pressure during a due diligence period is another. You need a systematic approach to ensure you do not miss critical red flags. I have compiled a checklist that I use for every single SaaS acquisition. This is the minimum standard for data review. If you cannot get this data, you should not proceed with the offer. Sellers who hesitate to provide this information are hiding something.

1. Raw Customer Database: Request a CSV file with User IDs, Signup Dates, Current Plan, MRR, and Last Login Date. Do not accept screenshots or PDF reports.
2. Churn Log: A record of all cancellations with dates and reasons if available. Look for patterns in the "Reason" field or timing relative to signup.
3. Upgrade/Downgrade History: Data on all plan changes. This allows you to calculate true Net Revenue Retention vs Gross Retention.
4. Marketing Spend by Channel: To correlate CAC with cohort retention. High-spending channels with low retention are bleeding cash that isn't visible in the P&L.
5. Support Ticket Volume: Monthly ticket count per 100 active users. Spikes in tickets often precede spikes in churn.
6. Pistachio/Billing Provider Logs: Verify that the MRR reported matches the actual billing cycles. Look for failed payment attempts that are still counted as active.
7. New vs. Returning User Revenue: Split the total MRR into revenue from new sign-ups in the last 30 days vs. existing users. This highlights dependency on acquisition.
8. Free Trial Conversion Rate: If there is a trial, track how many convert to paid. A drop in conversion rate often predicts a future drop in MRR.
9. Cancellation Refund Policy: Understand if refunds are automatically refunded or held as credit. This affects the cash flow impact of churn.
10. API/Usage Metrics: If applicable, check for a drop in API calls or feature usage. This is the leading indicator of disengagement before a user actually cancels.

Valuation Impact: How Retention Changes the Price

Finally, let us talk about money. How does this data affect the price you pay? In the SaaS world, the most common valuation metric is a multiple of Annual Recurring Revenue (ARR) or Annual Profit. However, a business with 10% monthly churn and a business with 2% monthly churn, even if they have the same ARR today, have radically different values. The low-churn business is worth a premium because its future cash flows are predictable and growing. The high-churn business is worth a discount because its future cash flows are unstable and shrinking.

Imagine two businesses, both with $1 million ARR. Business A has a monthly churn rate of 5%. This means it loses 40% of its customers every year. To maintain $1 million ARR, it must acquire $400,000 in new ARR every year just to replace what it lost. It must keep getting better at sales or lower prices to hit growth targets. This is expensive. Business B has a monthly churn rate of 1%. It only loses 11% of customers annually. It acquires $110,000 in new ARR to maintain its size. It can keep pricing power. It can afford to grow slowly. Business B is a fortress. Business A is a sieve. In the marketplace, Business B might sell for 8x ARR. Business A might sell for 4x ARR. The difference in value is $4 million, purely based on customer behavior.

When I advise clients, I tell them to calculate the "Churn Discount." If you see a retention curve that is flattening at a low level, apply a significant discount to the price. You are buying a business that requires active management of the "leak." You are buying operational risk. If the curve is flat and high, you are buying an annuity. The market prices annuities higher. Use the data to justify your offer. If a seller asks for $5 million, and your cohort analysis shows their LTV is declining due to poor re-onboarding, you offer $3.5 million. You are not haggling arbitrarily. You are correcting for the risk of decay. This is how you win as a buyer. You do more work, you look deeper, and you pay less for the same headline numbers. That is the edge that platforms like Deal Alert AI provide. We make the data visible so you can make the call.

Due diligence is not about finding a perfect business. Perfect businesses do not exist at reasonable prices. Due diligence is about understanding the flaws. Cohort analysis and retention curves are your X-ray machine. They show you the skeletal structure of the business. If the skeleton is broken, no amount of revenue skin will save it. Master these metrics. Practice them on public data if you can, or on mock scenarios. When you do find a real deal, you will be the one who sees the hidden risks and the hidden rewards. Go analyze the data. The numbers are waiting.