Learning Curve Calculator – Four Models in One Tool
The Learning Curve Calculator from MyTimeCalculator brings together four classic models used in operations, psychology, finance and productivity. Instead of memorising formulas, you can plug in a few intuitive inputs and immediately see how time, cost or performance might change as experience grows.
The four tabs correspond to different modelling traditions: Wright’s learning curve for time per unit, the Power Law of Practice for performance, a cost learning curve for unit and cumulative cost, and a simple experience curve for repeated tasks and training.
1. Wright’s Learning Curve (Production Time Model)
Wright’s learning curve dates back to aircraft manufacturing and models how long it takes to produce unit n given a constant percentage improvement each time cumulative production doubles. The key parameters are:
- T(1): Time required for the very first unit.
- Learning rate L: Time per unit after each doubling of units, expressed as a percentage.
- Exponent b: Computed as b = log(L) / log(2).
The time for unit n is then modelled as:
The calculator also sums T(1) to T(n) to estimate total labour time, and reports average time per unit and the improvement of unit n compared with the first unit.
2. Power Law of Practice (Performance over Trials)
In cognitive psychology and skill learning research, performance often follows a power law: large gains early, with smaller marginal improvements later. A simplified model is:
- a is performance at trial 1,
- b is a typically negative exponent (for example −0.3).
Depending on context, P(n) can represent time to complete a task, error rate, points scored, or any other performance metric. The calculator shows P(n), the improvement relative to trial 1, average performance across 1..n and a table suitable for further analysis.
3. Cost Learning Curve (Average and Cumulative Cost)
Many industries use learning curves to forecast how average cost per unit falls as volume grows. One common formulation uses the same log–log relationship as Wright’s curve:
where C(1) is the cost of the first unit and b is derived from the learning rate (for example, 85% learning means the average cost at 2N units is 85% of the average cost at N units). The calculator uses this to estimate:
- Average cost per unit at volume N.
- Cumulative cost for the first N units.
- Approximate cost of each unit in the learning curve table.
- Cost reduction and savings compared with making all units at the first-unit cost.
4. Experience Curve (Repeated Task Efficiency)
For training and productivity work, a simpler assumption is often enough: each time you repeat a task, it takes a fixed percentage less time than before. If r is the improvement per repetition (expressed as a decimal), the model is:
Ttotal = Σ Tk from k = 1 to n
The calculator reports the time at the final repetition, total and average time, and the total time saved compared with repeating the first-repetition time on every attempt.
5. How to Use the Learning Curve Calculator Step by Step
- Choose the tab that matches your context: production time (Wright), individual performance (Power Law), unit cost (Cost) or training reps (Experience Curve).
- Enter the initial time, cost or performance value and the learning rate or exponent, using realistic values from your data or assumptions.
- Specify the unit, trial or repetition number you want to analyse, plus a maximum value for the table.
- Click the calculate button to generate summary metrics and the detailed table for that model.
- Compare the outputs across tabs if you want to see how different models describe the same data or scenario.
6. Limitations and Practical Tips
- Real data are noisy: Learning curves smooth over variation from day to day. Use them as a guide, not a rigid rule.
- No single model fits everything: Wright’s curve suits repeated production, while the Power Law is more common in cognitive and skill learning research. For budgeting and forecasting, the cost learning curve can be more intuitive.
- Check units: Make sure the input units (minutes, hours, currency) are consistent with how you intend to interpret the outputs.
- Revisit parameters with data: As you collect actual times or costs, you can refine the learning rate or exponent to better match your own environment.
Related Tools from MyTimeCalculator
Learning Curve Calculator FAQs
Frequently Asked Questions
Quick answers to common questions about the different learning curve models, when to use each one and how to interpret the results from this calculator.
For manufacturing or repeated production of a similar unit, Wright’s learning curve or the cost learning curve are often most appropriate. For cognitive tasks, training and human performance, the Power Law of Practice or the Experience Curve can be more intuitive. It is common to try more than one model and see which gives the most realistic and stable description of your data over time.
Many studies report learning rates between about 70% and 95%. A 70% rate implies very rapid improvement (time or cost drops quickly as volume grows), while a 95% rate implies a much flatter curve. Values close to 100% mean little improvement. You can start with published benchmarks in your industry and then refine the rate using your own historical data.
In many applications of the Power Law, the performance variable is something you want to decrease with practice, such as time per task or error rate. A negative exponent b makes the function decrease as the trial number n increases, capturing the idea that you get faster or more accurate with repeated practice. If you use a positive b, the model will show performance growing instead of shrinking over time.
No. This tool is designed for quick scenario analysis and educational use. Properly fitting a learning curve to historical data usually involves regression techniques, diagnostics and checking for model assumptions. You can, however, use this calculator to explore reasonable parameter ranges and build intuition before doing more formal modelling in a spreadsheet or statistical package.
Power law and exponential-type curves can change quickly when parameter values are extreme or when you extrapolate to very large unit or trial numbers. In practice, learning eventually saturates and the curve flattens more than a simple mathematical model suggests. If the table outputs become unrealistic at large n, treat them as theoretical extrapolations rather than literal predictions and consider capping your analysis at a smaller range.
No. The calculator provides numerical estimates based on simple learning curve formulas and user inputs. It does not account for all the factors that affect financial outcomes, such as fixed costs, pricing, demand, risk or strategic considerations. Any real-world investment, production or staffing decision should be supported by a broader analysis and, where appropriate, professional advice.