Sim Loyalty helps you model customer retention and predict long-term growth using a three-segment retention model. Instead of assuming all customers behave the same, we divide them into groups based on how long they stick around.
Customers naturally fall into different retention patterns:
Each segment follows exponential decay: remaining(t) = initial × (1 - churn)^t
When you fit a model to your actual retention data, three segments provide the "sweet spot" for accuracy. Here's why:
Because each segment dominates a different time range, the optimization algorithm can distinguish between them and find unique solutions. This "separation" makes the three-segment model both flexible enough to capture real retention patterns and simple enough to fit reliably.
💡 Practical insight: Adding a 4th or 5th segment rarely improves fit quality because three segments already capture the main shape (fast decay + medium decay + slow/plateau). More segments often lead to overfitting without meaningful gains.
Great question! Here's why modeling adds value:
Bottom line: Actual data = rearview mirror. Model = windshield. You need both to drive forward!
Quick Start: Select a preset from Presets & Saved Sets dropdown (SaaS, E-commerce, Mobile App, or Enterprise B2B) to see example parameters.
Manual Setup:
💡 Tip: Save your parameters with a custom name to quickly switch between scenarios!
The simulator includes several interactive features for fluid exploration:
Pro Tip: The URL autosave means you can bookmark interesting scenarios or share your exact current state just by copying the URL!
If you have actual retention data from your analytics:
What is being fitted: The fitter estimates retention from a single cohort. Acquisition is a separate input used only to simulate total customer-base growth, so changing acquisition does not add a degree of freedom to the retention fit.
🔬 Advanced Multi-Method Optimization: The curve fitting runs and compares 4 different optimization algorithms (Grid Search, Differential Evolution, Random Restart, and Simulated Annealing), then selects the best result. This ensures robust fitting across diverse retention patterns, from simple exponential decay to complex multi-segment behaviors.
Single Cohort Retention: Shows how one group of customers decays over time. The stacked areas show the contribution of each segment (short-term, medium-term, long-term). Blue dots (if present) show your actual data vs. the fitted model (blue line).
Multi-Cohort Growth: Shows your total customer base when adding new cohorts monthly. The stacked areas reveal how much each segment contributes to total growth. The dashed line shows the theoretical ceiling (asymptote).
Your growth ceiling is the customer level you naturally approach when new customers per month and churn stay steady. It's the point where new in = customers out, so the total stops rising.
This is a flow equilibrium, not a motionless customer base. Customers can keep leaving and being replaced even though the total number of active customers no longer changes.
Think of your business as a bucket:
When you're small, the leak is tiny. As you grow, the same percent of a bigger base becomes a bigger absolute leak. Growth naturally slows until the leak equals the tap.
Rule of thumb: Ceiling ≈ monthly acquisition ÷ effective monthly churn.
Example: 1,000 per month and 5% churn → about 20,000 customers.
In this tool, your effective churn comes from the three segments you set (short, medium, long). Early on, the overall percent leaking each month can change as the mix of cohort ages stabilizes; the rule of thumb uses the steady-state effective churn that the model derives from your segments. We calculate and show the ceiling on the Growth chart and in the “Growth Ceiling” KPI.
How to raise the ceiling:
The three-segment exponential model works best for:
It may struggle with:
Check the Avg Error % when fitting. If >10%, visually inspect the blue dots - look for systematic patterns that the model misses.
This model uses steady gross monthly acquisition to isolate retention dynamics and calculate a meaningful growth ceiling. Gross acquisition counts customers entering the business; it is not the same as net customer growth.
A stable market does not contain the same people forever. When people enter and leave the market at balanced rates, its total size can remain stable while its membership changes. A mature business can therefore sustain a steady acquisition flow from new market entrants and replacements for departing customers.
At the growth ceiling, successful acquisition equals customer outflow, so net growth is zero. The business keeps acquiring customers simply to maintain its active customer base.
This assumption is realistic for:
Fixed acquisition lets you answer key strategic questions:
Model boundary: The simulator does not model addressable market size or a market-share cap. Its ceiling is the customer base supported by the selected acquisition and retention assumptions. If a real market cap is lower, successful acquisition would eventually have to fall to the replacement rate as that cap is approached.
For high-growth scenarios (e.g., 1,000 → 1,500 → 2,000/month), run separate scenarios at several fixed acquisition levels. Adding speculative compounding acquisition can hide the retention lesson and removes the fixed growth ceiling.
The simulator combines established ideas from queueing theory, survival analysis, and customer-base analysis:
Interpretation: Using exactly three segments is a practical finite-mixture approximation, not a claim that three distinct customer types objectively exist. A good fit supports the retention curve, but does not prove that the fitted segments are literal groups.
Try fitting these patterns to see model performance:
S-Curve (Poor Fit Expected):
100, 95, 90, 85, 75, 60, 40, 25, 18, 15, 13, 12
Slow initial churn, then rapid - model struggles with this pattern
Smiley/U-Curve (Poor Fit Expected):
100, 70, 50, 40, 35, 38, 42, 47, 52, 56, 60, 63
Retention improves over time (reactivation/network effects) - exponential decay can't capture this
Typical SaaS (Excellent Fit):
100, 82, 73, 65, 58, 52, 47, 43, 39, 36, 33, 31
Classic exponential decay - model excels here
Plateau/Sticky Users (Excellent Fit):
100, 80, 70, 65, 63, 62, 61, 60, 60, 60, 60, 60
Model handles this well with long-term segment at 0% churn