SaaS Cohort Retention Curve: The Decay Pattern

The decay pattern is the quieter failure mode in SaaS retention. There is no dramatic cliff, no month-1 crisis to mobilize around, no sharp alarm in the retention matrix. Customers simply leave, steadily and continuously, month after month, without acceleration and without remission. The cohort curve descends at a constant angle and shows no sign of finding a floor.

Decay is dangerous precisely because it is easy to rationalize. The headline churn rate looks manageable. Month-1 retention looks healthy. The product team points to smooth onboarding and reasonable activation rates. But underneath, the cohort base is eroding — not with a bang but with a slow bleed that compounds over time. A product with a 3% monthly decay rate loses 30% of any cohort annually and roughly half within 18 months. Sustained across five cohorts, that is a business permanently running to stand still.

Summary:

• Decay is a gradual, continuous churn with no single large drop — it looks normal month-to-month but compounds severely.
• Causes: habituation, competitive displacement, and value dilution — not onboarding failure.
• The half-life model gives a single number for decay severity: below 24 months is a serious problem.
• Decay without a floor indicates absent PMF; decay with a floor approaching asymptote indicates partial PMF.
• The correct intervention is product deepening and switching cost construction, not onboarding optimization.

What Decay Looks Like in the Retention Matrix

Decay is identified by its uniformity. In a retention matrix, the period-over-period drops are roughly equal across all time windows — no single month shows the disproportionate loss that would indicate a cliff, and no period shows a leveling off that would indicate movement toward an asymptote.

Period-over-period drops: M0→M1: 8pp. M1→M2: 7pp. M2→M3: 7pp. M3→M6: 15pp (5pp/month). M6→M9: 12pp (4pp/month). M9→M12: 9pp (3pp/month). The drop rate is decreasing slightly with cohort age, but there is no inflection point — no sign of stabilization. The curve is descending at a predictable rate without a floor in sight.

Compare this to a cliff pattern, where M0→M1 would be 20–30pp and subsequent periods would show 2–5pp monthly drops. In decay, the early months look relatively healthy and the problem only becomes apparent when you watch the full multi-year trajectory.

To quantify decay rather than just describing it visually, use the monthly decay rate calculation: decay_rate = 1 - (R_n / R_m)^(1/(n-m)) where R is retention percentage and n, m are month indices. For the January cohort above: 1 - (42/92)^(1/11) = 1 - 0.9254 = 7.5% monthly decay. That is extreme — it implies a retention half-life of about 9 months.

The Three Drivers of Decay

Driver 1: Habituation

Habituation occurs when the product delivers genuine value initially but gradually becomes background infrastructure — used occasionally, never central to the user's workflow, never generating the kind of regular engagement that makes cancellation feel costly.

This is common in reporting tools, compliance software, and products purchased for periodic use cases (quarterly planning, annual budgeting). The customer logs in 3–4 times, extracts the value they need, and then goes quiet. Each month of quiet login activity inches closer to a cancellation that eventually happens not because the product failed, but because the buyer can no longer justify the subscription cost against infrequent use.

Habituation decay is product-solvable: the product needs to engineer reasons for higher-frequency engagement, or needs to reposition toward a use case that generates daily or weekly touchpoints. But it is not an onboarding problem. Customers who experience habituation decay have already been onboarded successfully — the issue is that the product's core use case does not generate the behavioral frequency that creates switching cost.

Driver 2: Competitive Displacement

Competitive displacement is the most intellectually honest decay driver. The product is being replaced — either by a direct competitor with a better feature set, a different tool that absorbs the job-to-be-done through an adjacent workflow, or a platform that bundles the functionality at lower marginal cost.

Displacement decay is identifiable through exit survey data: churned customers who cite "switching to [competitor]" rather than "cost" or "not using it enough." It tends to manifest as decay rather than a cliff because displacement is a gradual decision — buyers evaluate alternatives for weeks before migrating, continue using the product until migration is complete, and churn at the subscription renewal rather than immediately.

According to (ProfitWell's Retention Research, 2023), competitive displacement accounts for approximately 27% of voluntary SaaS churn on average, with higher rates in crowded categories like project management, CRM, and marketing automation. The intervention is competitive moat construction — features, integrations, and network effects that raise the switching cost to a level that makes displacement irrational.

Driver 3: Value Dilution

Value dilution is a company-growth problem more than a product problem. As the customer's business grows, their needs evolve faster than the product does. The product that was the perfect fit at $2M ARR may be an increasingly poor fit at $10M ARR because it does not scale with the customer's complexity, team size, or workflow sophistication.

This creates a specific decay pattern: cohorts acquired during earlier periods of the vendor's development show higher decay rates than recent cohorts, because those early customers have had longer to grow beyond the product. Vintage cohort analysis reveals this pattern — year-over-year deterioration in early cohort retention as the vendor-customer fit erodes.

The intervention is product roadmap re-alignment with customer growth trajectories — specifically, building the mid-market or enterprise tier that allows customers to "graduate" within the vendor relationship rather than churning out of it.

The Half-Life Model for Decay Benchmarking

The half-life model borrows from physics to give a single intuitive number for decay rate severity. Retention half-life is defined as the number of months it takes for a cohort to lose half of its members from the post-cliff base.

Formula: half-life = ln(0.5) / ln(1 - monthly_decay_rate)

Or equivalently: half-life = -0.693 / ln(1 - decay_rate)

(SaaS Capital's 2024 B2B SaaS Benchmarking Study) places top-quartile annual logo churn at 6–8%, which corresponds to monthly decay rates of 0.5–0.7% — half-lives well above 100 months. Bottom-quartile annual logo churn above 20% corresponds to monthly decay rates above 1.8% and half-lives below 38 months.

The half-life number is useful in investor conversations and board reporting because it is intuitive — a business with a 14-month retention half-life is measurably in more danger than a business with a 46-month half-life, and the number travels well across contexts that non-technical stakeholders can reason about.

Calculating the Expected Floor From Decay Cohorts

Pure decay cohorts never reach a floor — they decay toward zero. But most real-world decay cohorts are not pure: they have a segment of highly-committed customers who are functionally immune to the decay forces affecting the broader base. These customers form the floor, and the floor is observable in the cohort data.

The floor becomes visible when the rate of decay begins to slow in later cohort periods. If months 12–18 show a smaller percentage point drop than months 6–12, the cohort is approaching its floor. The expected asymptote floor can be estimated by fitting an exponential decay model to the cohort data and extrapolating to where the curve flattens.

A simple approximation: the floor ≈ R_12 - (R_6 - R_12) × (R_12 / R_6). For a cohort at 62% at month 6 and 45% at month 12: floor ≈ 45 - (62-45) × (45/62) = 45 - 12.3 = 32.7%.

A floor below 30% for SMB SaaS indicates very weak product-market fit — fewer than 1 in 3 customers finds the product durable. A floor above 60% suggests strong PMF among a core segment even if the broader population churns. For the relationship between the asymptote floor and PMF signals, see the asymptote pattern analysis.

Decay Rate as a Product-Market Fit Signal

The decay pattern provides one of the clearest signals about product-market fit quality available in SaaS analytics. The relationship is direct: strong PMF produces low decay rates; weak PMF produces high decay rates.

This is because true product-market fit — defined operationally as the condition where customers would be deeply disappointed to lose the product — creates behavioral lock-in that manifests as low churn. Users who would be deeply disappointed to cancel integrate the product into their workflow more deeply, invite colleagues, build automations and integrations, and generate expansion revenue. Each of these behaviors reduces the monthly probability of cancellation.

The Sean Ellis PMF benchmark (40% of users would be "very disappointed" to lose the product) predicts decay rates with reasonable accuracy. Products above the 40% threshold typically show monthly decay rates below 2%. Products below the 40% threshold show decay rates above 3%. This correlation is not coincidental — both metrics are measuring the same underlying variable: how deeply users depend on the product.

Connecting decay rate to PMF also clarifies what decay improvement requires. It is not a process problem (better onboarding, better customer success), though those can help at the margins. It is a fundamental product problem: the product must become more difficult to replace, more deeply embedded in customer workflows, and more aligned with the use cases that create genuine dependency. As covered in the churn root cause taxonomy, decay-driven churn traces back to product-category PMF issues in the majority of cases — not to service failures or pricing sensitivity.

Distinguishing Decay From the Cliff: The Intervention Imperative

The cliff pattern and the decay pattern are often confused because both result in high overall churn. The distinction matters because the interventions are orthogonal.

The failure mode to avoid: deploying a high-effort onboarding redesign in response to a decay problem. Customers who have been retained through month 6 have already survived onboarding — their churn is not an onboarding failure. Applying an onboarding fix to decay churn treats the wrong disease with the wrong medicine and burns 6–12 months of product cycle without moving the decay rate.

The diagnostic is simple: calculate the month-over-month drop for each period and test whether the distribution is front-loaded (cliff) or uniform (decay). If the M0→M1 drop is within 2 percentage points of the M6→M9 drop, the pattern is decay-dominant and the intervention strategy should be product and competitive positioning, not onboarding.

Frequently Asked Questions

See the FAQ section in the frontmatter above for detailed answers to the most common questions about the decay pattern.

The decay pattern is a long-game problem that demands a long-game solution. It cannot be fixed in a sprint, cannot be addressed through onboarding optimization, and cannot be outgrown through acquisition speed. The only sustainable path to reducing decay is building the kind of product that customers find genuinely difficult to replace — a product embedded in critical workflows, expanding to meet customer growth, and continuously widening the competitive moat that makes displacement irrational. Cohort decay rate is the honest score on that project. Watch it closely, benchmark it against the half-life model, and invest in product depth accordingly.