Mining Ideas From Competitor Catalogs

You've got your model channels. Now you go through their catalogues and take the ideas that already worked.

That's the method stated plainly, because people get squeamish about it. You open a reference channel, pull its best-performing videos into a spreadsheet, and rebuild them as your own with AI. This whole business is pattern recognition and mimicry. Somebody already proved a topic pulls views in your niche, so you're not gambling on whether the idea works. You're producing your version of a result that exists. The originality goes into the execution, not into demanding a topic nobody has tried.

Two sorts do most of the work on a channel page. Most popular shows what the audience rewarded across the channel's whole life. Most recent shows what's landing now. The videos near the top of the popular list that also went up in the last year are the strongest ideas that channel has.

Recency matters more than people expect. Weight the last twelve months heavily and only reach further back when the recent shelf is thin. A hit from eight months ago tells you the topic works now. A hit from five years ago tells you it worked in a different era. So when you sort by most popular, filter by recency on top of it rather than taking whatever sits at number one.

That rule bends by type of video, and this part is worth getting right. Evergreen topics and top-ten style videos don't expire. If a competitor made a great one seven years ago and it stopped pulling views long ago, that idea is still entirely fair game. The demand never went anywhere, the video just aged out of the recommendations, and there's an audience now that never saw it. Trending topics are the opposite. Anything tied to news or a current moment needs a source video days old, not months. Copying a trend that peaked in March means making a video for people who moved on in April.

When you're working an evergreen channel, start at the top of the most-viewed list and go down it in order. Don't wander into the middle of the catalogue hunting for something you personally find more interesting. That ranking is the audience's verdict, and your first uploads should spend their effort on the strongest ideas available. On recent uploads, look for real engagement behind them instead. Views that are high for that channel's size, an active comment section, people arguing in it.

Then adapt. Never remake somebody's video one to one, because that audience has already watched it and your version earns no clicks. But don't confuse that with being late. Being late to a proven idea costs you nothing. The same concept supports thousands of uploads, as long as yours carries a distinct angle or style.

Read the comments under their top videos too. That's where viewers tell you the questions they keep asking and the specific things they liked. Both go straight into your topics and your format.

Now use AI properly, because this is where most people waste it. Never ask for video ideas from nothing. Feed it the actual performance data. A screenshot of the channel page, a screenshot of their most popular videos, or the breakdown you just ran. Then ask it to analyse the strategy behind those numbers and hand back a ranked list of new ideas, including remixes that fuse two of their hits into one concept. If you've run a competitor breakdown inside Claude, ask for their five best-performing topics with the original links and a full summary of each. Proven ideas with evidence attached, rather than guesses dressed up as suggestions.

One caution. Ultra-long videos are only worth copying if other channels in the niche already make them, because that's what tells you the audience sits through it. Otherwise test small.

Their best videos are a menu. Order from the top.