Cohort 25, compared with Cohorts 24 and 23. Only what changed.
The bar moved from getting views to holding them: a 15-second average watch time is now where the money is made.
How this was checked
Compared the kickoff (Zoom 1) and the coaches' teaching session (Zoom 2) of each cohort, topic by topic, against the full transcripts. A claim is shown only if a verbatim quote exists on both sides. "Dropped" means not said in S25 Zoom 1 or 2; it may still live in the curriculum.
Stated as policy7
A rule, number or tool stated as policy.
NewNumbers and cadence Average watch time now has benchmarks: 8 seconds is typical, 15 is where the money is S23 and S24 named no numbers. S25 gives three.
Cohort 25
The majority of videos get an eight second average watch time. ... If you can get to 12 seconds, you're going to be in a good spot. If you can get to 15, like average, uh, 15 second average watch time, that's where the money is made on social.
taking an average watch time from eight seconds to 15 seconds and then 15 seconds to 20 seconds
Before
watch time is such a big piece of this equation
S24 names watch time as important but gives no number. Neither cohort's Zooms 1 and 2 contain one.
ChangedAI and tools The coaches now edit with AI. AI scripts are still banned. S24: AI "can't edit your video." S25: the coaches edit with ChatGPT inside DaVinci and promise a tutorial.
Cohort 25
It was completely enabled by the powers provided by ChatGPT and Astra. ... the DaVinci Resolve ChatGPT integration has allowed us to rapid fire. We made the decision like, I don't know, it was like 10 days ago that we're going to update all this curriculum for the next session, and we didn't have to edit any of it.
Uncertain: Astra
Colin and I have been working a bunch on using ChatGPT with Astra with DaVinci to edit. We don't cover that inside here. We will make a tutorial for it as soon as it's good enough that we feel like everyone can use.
Uncertain: Astra
We would not be using AI for any of your scripts. If you post AI scripts inside here and they're noticeably AI, we're going to talk shit on you because it means your content is going to be bad.
Before
You start posting AI scripts in here, they're going to get torn apart, right? ... Claw can edit your video. Claw can't edit your video. It's trash.
Uncertain: Claw
If you jump into this and have ChatGPT write your script, every time we've seen that, first of all, we can tell.
"Astra" is almost certainly a mangled tool name; the transcript never defines it. "Claw" in S24 is likely Claude or another tool name. Check the recordings before repeating either.
ChangedFormats and homework The whole course was rebuilt, not patched S24 flagged two new sections. S25 says every homework and lesson is new, because the algorithms changed.
Cohort 25
this is all new. Every single piece of homework and curriculum here is new based on kind of recent things that have changed in the algorithms, how content performs.
Before
if you are a repeat cut 30, the vlog section is new and the green screen section is new because those are kind of actively changing in the zeitgeist right now.
ChangedFormats and homework The 10-shot moved to week 1, and green screen joins graphic yap in week 2 S24 week 1: one-shot, yap, copywork. Week 2: 10-shot plus graphic yap. S25 week 1: one-shot, prop yap, 10-shot. Week 2: graphic yap plus green screen.
Cohort 25
The third video is the 10 shot. 10 shot is exactly what it sounds like. It's an expansion of the one shot, typically like five to 10 up to 20 short clips with really solid text copy over it.
two of them are going to be very specific homework assignments. One is going to be a graphic app, which we'll look at in a sec. One is going to be a green screen.
Uncertain: graphic app
Before
The third video for week one is to copy work a winning video.
make three videos, no uploads, only posting published links to one of those is going to be a 10 shot. One of those is going to be a graphic gap
"Graphic app" and "graphic gap" are Whisper's rendering of "graphic yap".
DroppedFormats and homework Picking a track is no longer the first homework S24 opened with "pick a track channel" and sent bios there. S25 never mentions tracks and sends bios to Idea Workshopping.
Cohort 25
homework is workshop your bio, update your bio, post before and after an idea of workshopping for feedback.
Before
But first and foremost, you are going to pick a track channel.
our homework for this next week is rewrite your bio, share that in your track channel for feedback
part of your homework is going to be to pick a track
ChangedPlatform and algorithm Collab posts now need 5,000 followers, up from 2,000 to 3,000 Same rule, a higher bar. Both warn that small accounts collabbing hurts reach.
Cohort 25
Do not collab post unless there is a literal reason to establish creators, 5,000 followers or more who are collaborating with each other ... If you are too small accounts collaborating, hurts in the algorithm more than it helps.
Before
it has to be from an established account, let's say even over 2,000, 3,000 followers.
The collabs tool will almost always delete your reach.
S23 gave no number.
ChangedPlatform and algorithm All three hooks now have to stay on screen for three seconds S24 said you have the first frame and first second. S25 asks for the visual, verbal and text hooks to hold the full three seconds.
Cohort 25
the title hook, which is the text on screen, those first three seconds, right? Need to have all three elements and all need to stay on the screen for those three seconds. Like that is absolutely key.
Before
back in the day, it'd be like, oh, you have three to five seconds to get someone's attention. It's like, no, like you really have like the first frame and like the first second to really get someone's attention.
think about in your first three seconds and in reality like the first second the first frame
S23 also said three seconds, so S24 was the outlier. What is new in S25 is the explicit hold.
Repeated in the session3
Repeated more than once in the session.
ChangedFormats and homework A master-level graphic yap runs an image every second; starters stay at two to three The starter rule is unchanged. The one-second ceiling is new.
Cohort 25
a master level graphic yap, if you will, you look at one of Oren's graphic yaps, it's roughly an image every second. Do not try and do that on your first one.
this week we're going to probably have you change it up more so every two to three seconds versus one second
Before
A good rule of thumb for this is maybe one graphic for roughly every two to three seconds of the video.
DroppedAccount setup Font and brand-kit advice is gone from the Zooms S23 and S24 both told brands to skip their brand fonts. S25 Zooms 1 and 2 never mention fonts.
Cohort 25
Not said in Zoom 1 or 2 (searched the full text).
Before
as far as we would generally recommend going is what you saw in the video Alex shared where they're using instrument serif in edits but it's like recognizable fonts mean people are going to be able to read it quicker and faster
We're going to tell you what fonts to use. You are not going to use your brand's fonts if you're a brand or a business account.
Absence in two Zooms does not prove the lesson was cut.
Smaller signals (4). Anecdotes and answers to chat questions. Collapsed on purpose.
ChangedPlatform and algorithm Trial reels: still off limits, and the unlock no longer has a number S23 gave 20,000 followers. S24 said don't think about it. S25 bans it "this entire time" and says to add it later.
Cohort 25
We do not want you to touch trial reels this entire time. ... you get good at it, and then you introduce stuff like trial reels.
Before
If you are an experienced creator and you're doing really well, we talk about things like trial and stuff. We could talk about it in office hours. But for anyone who is new, don't think about it for a second.
Once you get 20,000 followers and have a nuanced social strategy where you need to be A-B testing hooks, then you can mess around with trials.
Answered from the chat in all three cohorts.
UnchangedPlatform and algorithm Schedulers: the tone varies by Zoom, the answer does not Post manually in all three cohorts. S25 Zoom 1 is softer than S24. S25 Zoom 2 is harsher.
Cohort 25
They may not stop you from going viral. They may not be a complete negative, but including the built-in schedulers in these apps tend to underperform if you post it organically.
never, ever, ever, ever, ever post through meta business or any third party scheduling. Don't do it. For like, for several reasons, one of which being you can't add location.
Before
We do not recommend using any scheduling software, including the native schedulers. We have pretty thoroughly tested this. It won't ruin your account or anything, but it definitely does not. It definitely hurts.
Not a real shift. The new element is the location reason.
DroppedPlatform and algorithm "TikTok for first views, Instagram for the long run" is not repeated S25 calls both platforms mandatory and says traction can come from either.
Cohort 25
Instagram and TikTok, I would consider mandatory. Post every one of the videos you make here on Instagram and on TikTok.
Before
what we typically find is that people will more easily get their first views and first successful videos on TikTok, and they'll end up growing better and having a better time over time on Instagram.
DroppedPlatform and algorithm "Local content is a superpower" is gone; the small-country advantage and local-business advice stay Partly dropped. S25 Zoom 2 still says location matters for local businesses.
Cohort 25
If you're a local business, like location, location, location, all day long.
Before
if you make content locally about the places you are, that is a superpower. ... Algorithms love local content and content around places.
Stable core
Still said every time: the content-market-fit Venn, a three-part hook, the bio as a landing page, post vitals, hashtags don't matter, cut 30 to 40 percent of your words, don't batch early, post natively.
Quotes from both cohorts
- Hashtags don't matterCohort 25 · Zoom 2: "hashtags do not matter, right?"Cohort 24 · Zoom 2: "I think it's reasonable to say hashtags do not matter."
- Cut the wordsCohort 25 · Zoom 2: "delete 20, 30, 40% of the words and have the exact same meaning"Cohort 24 · Zoom 2: "delete 30 to 40% of the words and have the exact same video"
- Don't batchCohort 25 · Zoom 2: "highly recommend not batching for now"Cohort 24 · Zoom 1: "Do not batch in the early days."
Checked and cut
First-pass claims that did not hold up against the transcripts, or that held in a smaller form.
- Live critique of member videos is now Q&A onlyNot supported. S25 Zoom 2 still opens with workshopping "direct from video sharing". Only the S24 invitation to bring your own clips is missing.
- Hook window widened from the first frameKept, but reframed. S23 already said three seconds, so S24 was the outlier.
- Schedulers softened and hardenedKept as low and stable. All three cohorts say post manually.
- Graphic-yap density doubledReframed. The starter rule of two to three seconds still applies in S25.
Private to Juniper Island. Quotes are Cut30's, from Whisper transcripts; spellings marked uncertain may be wrong.