Cohort Retention Calculator
Calculate and visualize retention by cohort from monthly user activity data. Enter values for instant results with step-by-step formulas.
Calculator
Adjust values & calculateRetention Curve
Monthly Breakdown
Projected Retention (Months 13-24)
Formula
Cohort retention tracks the percentage of an original user group that remains active in each subsequent month. LTV is calculated by summing monthly revenue contributions weighted by retention rates. The calculator interpolates between provided data points and projects future retention using exponential decay modeling.
Last reviewed: December 2025
Worked Examples
Example 1: B2B SaaS Monthly Cohort Analysis
Example 2: Comparing Two Acquisition Channel Cohorts
Background & Theory
The Cohort Retention Calculator applies the following established principles and formulas. Large language models process text by breaking it into tokens, sub-word units produced by algorithms such as byte-pair encoding. In English, one token approximates four characters or three-quarters of a word on average, though this ratio varies considerably across languages and code. A 1000-word document typically requires around 1300 to 1500 tokens. Token count drives both context window constraints and inference billing, making accurate estimation essential for budgeting API usage. The capability of a neural network scales primarily with its parameter count. Parameters are the numerical weights adjusted during training via gradient descent. GPT-3 contains 175 billion parameters; larger models in the trillion-parameter range require correspondingly greater compute and memory. Training compute is measured in floating-point operations (FLOPs): the Chinchilla scaling laws derived by Hoffmann et al. in 2022 show that optimal training allocates roughly 20 tokens per parameter, meaning a 70B-parameter model benefits from approximately 1.4 trillion training tokens. Inference latency depends on model size, hardware, and batching strategy. Running a 7B-parameter model in FP16 precision requires roughly 14 GB of GPU VRAM (2 bytes per parameter), while INT8 quantisation halves this to around 7 GB with modest quality loss, and INT4 reduces it to approximately 3.5 GB. This quantisation trade-off between memory, speed, and accuracy is central to deploying models on consumer hardware. Perplexity measures how surprised a language model is by a given text corpus; lower perplexity indicates better predictive accuracy. Embedding dimensions determine the size of the dense vector representations used to encode semantic meaning. Models like OpenAI's text-embedding-ada-002 produce 1536-dimensional vectors, while compact models may use 384 dimensions. Context window size defines the maximum token span a model can attend to in a single forward pass. Extending context windows from 4K to 128K tokens enables document-scale reasoning but substantially increases memory requirements, as the attention mechanism scales quadratically with sequence length without architectural modifications such as flash attention.
History
The history behind the Cohort Retention Calculator traces back through the following developments. The mathematical neuron model published by Warren McCulloch and Walter Pitts in 1943 first proposed that logical functions could be computed by networks of simple threshold units, planting the seed of neural computation. Frank Rosenblatt's Perceptron, introduced in 1957 and implemented in custom hardware by 1960, could learn linear classifiers from examples and generated enormous public excitement before Marvin Minsky and Seymour Papert's 1969 book rigorously analysed its fundamental limitations, demonstrating it could not learn the simple XOR function. The first AI winter, roughly 1974 to 1980, followed as funding agencies in the US and UK grew disillusioned with unrealised promises. A second wave of interest during the 1980s produced rule-based expert systems deployed in medicine and finance, and saw the re-derivation of backpropagation by Rumelhart, Hinton, and Williams in 1986, making it practical to train multi-layer networks on real problems. A second winter from 1987 to 1993 followed as expert systems proved brittle and hardware remained insufficient for genuine deep learning. The deep learning revival crystallised at the ImageNet Large Scale Visual Recognition Challenge in 2012, when Alex Krizhevsky's convolutional network AlexNet slashed the top-5 error rate by nearly 11 percentage points compared to the prior year's winner. This demonstrated that deep networks trained on GPUs with large labelled datasets could achieve human-competitive image recognition. Subsequent years saw rapid advances in recurrent networks, sequence-to-sequence models, and the attention mechanism, culminating in the transformer architecture introduced by Vaswani et al. in 2017. OpenAI released GPT-1 in 2018, demonstrating that unsupervised pre-training on large text corpora followed by task-specific fine-tuning could transfer knowledge broadly across language tasks. GPT-2 in 2019 demonstrated surprisingly fluent long-form text generation. GPT-3 in 2020, with 175 billion parameters, showed that scale alone could unlock few-shot learning. Kaplan et al.'s 2020 scaling laws paper provided the theoretical grounding. ChatGPT launched in November 2022, reaching one million users within five days and igniting mainstream global awareness of large language models.
Frequently Asked Questions
Formula
Monthly Retention Rate = (Active Users in Month N / Original Cohort Size) x 100
Cohort retention tracks the percentage of an original user group that remains active in each subsequent month. LTV is calculated by summing monthly revenue contributions weighted by retention rates. The calculator interpolates between provided data points and projects future retention using exponential decay modeling.
Worked Examples
Example 1: B2B SaaS Monthly Cohort Analysis
Problem: A January cohort of 1,000 users shows: Month 1: 65%, Month 2: 50%, Month 3: 42%, Month 6: 30%, Month 12: 22%. Monthly revenue per user is $50, CAC is $150. Calculate cohort economics.
Solution: 12-month cohort revenue: Sum of monthly users x $50\nMonth 0: 1,000 x $50 = $50,000\nMonth 1: 650 x $50 = $32,500\n...through Month 12: 220 x $50 = $11,000\nTotal 12-month revenue: ~$458,000\nTotal CAC: 1,000 x $150 = $150,000\nROI: ($458K - $150K) / $150K = 205%\nLTV per user: $458,000 / 1,000 = $458\nLTV:CAC = $458 / $150 = 3.1x
Result: 12-month LTV: $458 | LTV:CAC: 3.1x | ROI: 205% | Monthly churn: 12.4% | Curve: Front-loaded churn
Example 2: Comparing Two Acquisition Channel Cohorts
Problem: Organic cohort (500 users): M1 75%, M3 55%, M6 42%, M12 35%. Paid cohort (500 users): M1 55%, M3 32%, M6 18%, M12 10%. Same $50/user/month. Organic CAC $80, Paid CAC $200.
Solution: Organic 12-month revenue: ~$289K, LTV: $578, LTV:CAC = $578/$80 = 7.2x\nPaid 12-month revenue: ~$160K, LTV: $320, LTV:CAC = $320/$200 = 1.6x\nOrganic retains 175 users at M12 vs Paid retains 50\nDifference: Organic delivers 4.5x better unit economics
Result: Organic LTV:CAC 7.2x vs Paid 1.6x | Organic M12 retention 35% vs Paid 10% | Shift budget to organic acquisition
Frequently Asked Questions
What is cohort retention analysis and why is it important?
Cohort retention analysis tracks the behavior of specific groups of users who share a common starting date, measuring what percentage remain active over subsequent time periods. Unlike aggregate retention metrics that mix users from different time periods, cohort analysis isolates the experience of each group to reveal true retention patterns. This is critical because aggregate metrics can mask deteriorating retention when rapid user acquisition obscures increasing churn. For example, a company acquiring 1,000 new users monthly with declining retention might still show growing total active users, hiding the underlying problem. Cohort analysis reveals these trends early, typically 3-6 months before they impact aggregate metrics, giving teams time to intervene before retention problems become revenue crises.
What does a healthy retention curve look like?
A healthy retention curve shows steep initial decline followed by a flattening plateau, resembling a hockey stick on its side. The initial drop in month 1 reflects users who tried the product but found it was not a fit, which is expected and normal. This drop should stabilize by month 3-4, with the curve flattening as remaining users develop habits and integrate the product into their workflows. The ideal pattern shows 60-70% month 1 retention, 40-50% by month 3, and then minimal decline through month 12. A continuously declining curve that never flattens indicates a product retention problem. A curve that flattens early but at a very low percentage like 10-15% suggests the product serves a niche well but lacks broad appeal. The best SaaS products achieve asymptotic retention above 30% after 12 months.
How do I calculate LTV from cohort retention data?
Lifetime value from cohort data is calculated by summing the revenue generated across all periods for the average user in the cohort. For each month, multiply the retention rate by the monthly revenue per user to get the expected revenue contribution. Sum these contributions across the customer lifetime to get cumulative LTV. For a user paying $50 per month with 65% month 1, 50% month 2, and 42% month 3 retention, the 3-month LTV contribution would be ($50 x 1.0) + ($50 x 0.65) + ($50 x 0.50) + ($50 x 0.42) = $128.50. Extend this calculation across 12-36 months using actual and projected retention rates. Compare LTV to customer acquisition cost with a target ratio of at least 3:1. This method is more accurate than formula-based LTV calculations because it uses observed behavior rather than assumed constant churn rates.
What is the difference between user retention and revenue retention?
User retention measures the percentage of customers who remain active, while revenue retention (also called net revenue retention or NRR) measures the percentage of revenue retained from a cohort including expansion revenue. These metrics can diverge significantly because remaining customers may spend more over time through upgrades, additional seats, or premium features. A cohort might show 70% user retention but 110% net revenue retention if the 70% who stay increase their spending by more than the lost revenue from churned users. Companies with net revenue retention above 100% can grow even without acquiring new customers. Track both metrics because strong user retention with weak revenue retention suggests pricing or expansion problems, while weak user retention with strong revenue retention masks a churn problem with short-term expansion revenue.
How does month 1 retention predict long-term outcomes?
Month 1 retention is the single most predictive metric for long-term cohort health because it reflects the quality of the initial user experience and product-market fit. Research across hundreds of SaaS companies shows that month 1 retention below 40% almost always leads to unsustainable economics regardless of later improvements. Each 5 percentage point improvement in month 1 retention typically translates to 3-4 percentage points higher retention at month 12. The first-month experience determines whether users form habits, integrate the product into workflows, and build switching costs. Companies should invest disproportionately in the first 30 days through better onboarding, proactive support, and early value delivery. If month 1 retention is below 50%, focus all product improvement efforts on the first-time user experience before working on later-stage retention.
How do I project future retention from limited historical data?
Projecting retention with limited data requires a combination of mathematical modeling and reasonable assumptions. The simplest approach fits a power law curve to your existing data points because retention curves naturally follow a power law decay pattern. If you have 3 months of data, use the ratio between consecutive months to estimate the decay rate, then project forward while constraining the curve to never drop below a reasonable floor of 5-10%. More sophisticated approaches use logarithmic regression or shifted exponential functions that better model the flattening behavior of mature cohorts. Always present projections with confidence intervals because accuracy decreases significantly beyond 2x your historical data range. Validate projections by comparing predicted retention against actual retention for your earliest cohorts. Update projections monthly as new data arrives.
References
Reviewed by Daniel Agrici, Founder & Lead Developer · Editorial policy