Logarithmic Extrapolation for Diminishing Returns

Not all growth accelerates. In many real-world scenarios, gains diminish over time — each additional unit of effort produces less and less return. This is where logarithmic extrapolation becomes essential.

The Logarithmic Model

The logarithmic function is expressed as:

y = a + b · ln(x)

Key characteristics:

• y increases with x, but the rate of increase slows down
• The curve is concave down (flattening)
• Works only for x > 0

When to Use Logarithmic Extrapolation

• Skill acquisition: Learning curves where progress slows
• Market saturation: Growth that decelerates as a market matures
• Physical phenomena: Many natural processes follow log patterns
• Effort-output relationships: Where doubling effort doesn’t double output

Comparison with Exponential

While exponential models accelerating growth (curves upward), logarithmic models decelerating growth (curves downward). Using an exponential model when the true pattern is logarithmic will lead to wildly overestimated predictions.

Practical Example

Consider monthly users for a mature product:

• Month 1: 1,000
• Month 6: 3,500
• Month 12: 4,800
• Month 24: 5,900

A linear model would predict steady growth indefinitely. An exponential model would predict explosive growth. But a logarithmic model captures the reality: growth is real but slowing.

Tips

• Always ensure x values are positive (logarithm requires x > 0)
• If your R² for logarithmic is significantly better than linear, the diminishing-returns pattern is real
• Logarithmic extrapolation is more conservative than exponential, making it safer for long-range predictions