Workers and processes get faster with repetition, and Wright’s law describes how the time for each unit falls as total output grows. This learning curve calculator uses the first unit time and the learning rate to give the time for any unit, the total and average time for the first n units and the labor cost.
How the calculation works
- Learning exponent b = ln(learning rate) ÷ ln(2), a negative number for rates below 100 %.
- Time for unit n = time for the first unit × n^b.
- Total time ≈ time for the first unit × n^(b + 1) ÷ (b + 1), the integral of the curve.
- Average time = total time ÷ n.
- Labor cost = total time × labor cost per hour.
Example
A first unit takes 100 hours and the process has an 80 % learning rate, so each doubling of output cuts the unit time by 20 %. The exponent is −0.322, and unit 50 takes about 28.4 hours, a 72 % reduction. All 50 units take about 2,090 hours, or 41.9 hours on average, and cost about $73,200 at $35 an hour.
Why it matters
Bids and budgets that ignore the learning curve overestimate the labor for long runs, while those that assume the curve continues forever underestimate it. The model gives a defensible way to price a contract, plan hiring for a ramp-up, or compare the cost of a first batch with a later batch from the same process.
Good to know
The learning rate is estimated from your own history: divide the time of a later unit by the time of a unit with half the cumulative output. The cumulative average version used by some industries gives slightly different totals, so be consistent in how you measure it. Learning slows when workers change, tools change or the product design changes, because the curve restarts for the new elements. Include setup, rework and idle time only if they are part of the data that you used to estimate the rate.
Related tools
The labor productivity calculator measures output per hour, the PERT estimate calculator handles uncertain task durations, and the team capacity calculator plans sprint hours.