How do my Shorts and long-form videos compare?

Why one metric is not enough

A Short with a 10% like rate can earn a fraction of the revenue and a tenth of the watch time of a long-form video with a 5% like rate. “Better engagement” is not “better for the channel”. Odinstat reports each metric’s winner separately and states the trade-off plainly: for example, Shorts win on engagement while long-form wins on watch time, subscriber growth, and revenue.

How the comparison is run

For each metric, Odinstat resamples your Shorts and your long-form videos many times (a seeded bootstrap, so the same data always gives the same answer) and records how often one format comes out ahead. A format is declared the winner on a metric only when it wins in at least 80% of resamples; below that the metric is reported as too close to call. True proportions such as like rate and subscriber conversion use a Beta-Binomial model; per-video amounts such as views and watch time use the bootstrap directly.

What this is not

You did not randomise which ideas became Shorts and which became long-form, so this is not an A/B test and Odinstat does not call it one. Topic, publish date, and promotion all differ between the two groups. The result tells you how the two formats have actually performed for you, which is the question most creators are asking, and it is labelled observational everywhere it appears.

What this is not

• Deciding a format for a channel that has not published both. There is nothing to compare yet.
• Attributing the difference to the format alone. Topic and timing are not controlled.