What YouTube’s Biggest Changes Mean for Your Channel - Made On YouTube 2026 (Video)

Every time I publish a video, I make a bet I can't check. Should my iOS 27 video open with a quick overview, or should feature number one start the second you hit play? At Made on YouTube 2026, YouTube announced a way to test that, and I sat down with creator liaison Rene Ritchie and TechMe0ut to go through it and everything else on stage.

Video A/B testing brings the thumbnail and title tests already in Studio to the video itself. You upload two or three versions, YouTube splits the audience, and you get a separate retention graph for each, plus about a week to choose. You can see whether a hook holds people or crashes right after, try a different story structure, or move a sponsor read. TechMe0ut wants to know if her opening montages run too long. YouTube picks a winner, but the creator gets the final say.

Smaller channels should temper expectations, because the test needs statistical significance. Rene's analogy stuck with me: flip a coin five times, get three heads, and heads didn't win. You ran out of flips. So test big differences (face or no face, text or no text), and don't stop early.

Dynamic thumbnails confused TechMe0ut and me at first. You pick three, and instead of crowning a winner, YouTube keeps all three in rotation and shows each to the audience most likely to click. It starts with regulars and newcomers, so regulars might see your face while new viewers see the device. TechMe0ut drew the line clearly. Dynamic thumbnails are for going in with intention, and A/B tests are trial and error. My feature request is a mix of both, where YouTube tells me whether my three should stay dynamic or one is the clear winner.

Ask YouTube is where my viewer brain and creator brain disagree. In YouTube's demo, someone hunting for an affordable retro Kodak got sent to the 6:15 mark of a video, watched until 7:04, bought through YouTube Shopping, and left. As a viewer, I love that. As a creator, I worry about watch time, and about the problem publishers hit when AI search answers mean nobody clicks through. Rene says traffic sources are judged mostly on their own performance, so a viewer who grabs a quick answer and leaves doesn't change how the video gets recommended on the homepage or in suggested videos. He also argues the views are additive, since a phone review shot in Bushwick could reach someone researching Bushwick who would never click a phone review. That example sold me.

Custom feeds go after my biggest YouTube annoyance. About 90 to 95 percent of my TV viewing is YouTube, and on the TV app, scrolling down one row can show me four of the six thumbnails I just saw. Now you describe what you're in the mood for and keep that feed, up to three of them. Rene's are latte art and cozy woodworking. Mine starts with golden era pro wrestling documentaries, because I wrestled from 1998 to 2001.

Then I asked when any of this ships. Some features announced last year still haven't reached my account. Rene called them "out-ish" and walked through the usual path from alpha to beta to staged rollouts at 1, 10, and 50 percent. Custom feeds are live now for YouTube Premium subscribers, and Ask YouTube already works from the search bar, so if you mostly watch on a TV, make a feed tonight. Rene didn't give a date for video A/B testing, so my iOS 27 intro stays a guess a little longer.