Shelf life
How long do my videos stay relevant?
Odinstat follows each video’s daily views across its first 90 days, fits both an exponential and a power-law decay curve, keeps the better fit, and reports the half-life: the number of days it took daily views to fall to half their peak.
A channel of long half-lives makes evergreen content; a channel of short ones lives on launch day. Both are valid strategies, and knowing which you have changes what you publish next.
Two curves, one winner
Exponential decay describes videos that fade at a constant rate: strong launch, steady drop. Power-law decay describes a fast early fall followed by a long tail, the signature of search and suggested traffic that keeps arriving months later. For every video Odinstat fits both, compares the goodness of fit, and reports the model that describes the data better along with its half-life.
Why the gates are strict
Half-life is only meaningful for a video tracked from launch. A video that Odinstat started tracking mid-life has no peak to measure from, so it is skipped rather than estimated. A video needs at least 30 days of daily views inside the 90-day window before either curve is fitted; with less, the tail that separates exponential from power-law is not visible yet.
A fit whose r² is 0.3 or lower is dropped. That video’s views did not follow a decay pattern at all (a re-share, a later spike, a playlist pickup) and reporting a half-life for it would be inventing a number.
Reading the result
Odinstat reports the longest- and shortest-lived videos with their half-lives and fit quality, and the channel’s typical half-life. Compare the long-lived videos’ topics and traffic sources with the short-lived ones and the evergreen pattern on your channel usually becomes obvious.
What you need
The question unlocks at 3 videos, but a video only receives a half-life when Odinstat has tracked it from launch with at least 30 days of daily views inside the 90-day window and the better decay fit has an r² above 0.3. Videos that miss either condition are listed as skipped, not estimated.
What this is not
- Forecasting lifetime views. Odinstat has a separate trend projection for upcoming videos; shelf life describes the past.
- Videos published before you connected, unless your daily history for them starts at launch.
Odinstat runs this on your own channel, read-only, with a 14-day free trial from $10/month.