One month of automated shorts: our numbers, unfiltered
47,800 views, 32 subscribers, and a posting-rate mistake that cost us. What we measured cutting podcasts into shorts every day for a month.
Most articles about short-form recite advice. This one reports what we measured on our own channel, with the exact numbers, including the ones that do not flatter us.
The setup
A YouTube channel about finance, fed only with excerpts from long podcasts. Nothing filmed. We take an hour-long episode, the pipeline finds the passages that stand on their own, crops vertically onto whoever is speaking and burns in the captions.
Pace: three shorts a day, at noon, late afternoon and evening. Clips run from twenty seconds to a minute. Titles quote the speaker word for word, in quotation marks when it is an opinion rather than a fact.
The raw numbers
Over twenty-eight days: 47,800 views and 166 hours of watch time. The channel went from 8 to 32 subscribers. The best short reached 2,300 views, the weakest around 350.
A short that works starts fast. The best one was at 900 views after two hours, another at 480 after one. When the first hours are slow, the video does not catch up later.
The mistake that cost the most
On day one we posted ten shorts back to back. YouTube eventually refused the ninth upload with a daily limit message. That day's videos competed with each other and the last ones sat at zero views for hours, which looks a lot like a penalty without being one.
YouTube's technical ceiling sits well above the useful limit. Three a day, hours apart, beat ten at once. It is the one rule this month actually settled.
What retention tells you
Average percentage viewed is the number that best predicts what happens next. Our best shorts sit between 67 and 84 percent, the weakest around 60. Past 80 percent, the video keeps being pushed for days instead of dying the next morning.
Two things raise it. Cut exactly where the idea ends, with no trailing silence and no sentence left hanging. And open on a line that makes sense without context, never mid-answer.
What these numbers do not prove
One channel, one month, one niche. Nothing here guarantees the same result elsewhere, and it is not a clean test: the posting rate changed at the same time as the editing quality, so the two effects cannot be separated.
More importantly, views are not customers. Those 47,800 views on finance excerpts produced zero sign-ups for the tool, because the audience came for the content, not for the software behind it. An audience only helps when it overlaps the product's own.
Try it on your own video
Paste an episode link and see what it gives on your content rather than on ours.
Try SignalHunter