Logarithmic projection
Model growth that is fast early and gradually levels off over time.
The logarithmic method fits a curve of the form y = a + b × ln(x) to your historical data, where growth slows as time progresses.
How it works
Using least-squares regression on log-transformed time values, TotalKPI finds the constants a and b that best describe your data's curve. The resulting formula is then evaluated at future time points to generate the projection.
The natural log transformation compresses the time axis, so each additional time period contributes less to the predicted value than the last - capturing the characteristic "fast then slow" shape.
When to use it
- Your metric grew quickly early on and is now decelerating toward a plateau.
- The growth curve looks like it is flattening when you eyeball it on the chart.
- Examples: cumulative user signups after a launch spike, organic search traffic maturing, adoption of a new feature.
Limitations
- Only models decelerating growth - it cannot represent acceleration or decline. For a steady downward trend use Linear trend.
- Requires positive, non-zero time values and is fit on the historical data as-is; very short histories may produce an unreliable fit.
- The model asymptotically approaches a ceiling but never truly flattens - very long-horizon forecasts will continue creeping upward rather than plateauing completely.