Josh Howard

I studied 10,000+ YouTube Community posts from channels with 200 to 17M subscribers. What makes a post reach a bigger audience? What’s the most engaging type of content? Do community posts even matter? What’s the point? Here’s my findings:

@TheCraigHewitt Heck yeah! Glad to see things moving!

always so interesting what turns out to be winners on YouTube. @RealJoshHoward called it with this one.

@donnfelker Anytime you take a breath for a week or two feels like you get behind so fast 😅

With AI I'm 10 times as productive, but now I also feel 10 times as behind as I did before.

@trq212 An overall session log with routing (generic/light), a specific session log on the chat that stays updated(parent folder log w/expanded info), and this type of documentation are the ultimate combo.

a prompt I've been using a lot recently: implement <SPEC> and while you do, keep a running implementation-notes.html file (or markdown) with decisions you had to make weren't in the spec, things you had to change, tradeoffs you had to make or anything else I should know

@PhilStarkovich Oh my... I got to the first 2 and I thought the meme was "dang, are those real?" like it was a good thing. I indeed did NOT make it to the 3rd! 😶 (bathroom scrolling, sorry lol)

@RealJoshHoward make sure to view all 3 images 😄

Am I doing this right?

@Daniel_Batal 100% with you. I’m deep on the data as well, so I hear you. One of the best things in a while was the audience segment data from last summer. One of my favorite new ones to use for decision making and analysis. I feel like we could have a pretty good chat if it ever presents. 🙂

Interesting take! Viewership is certainly more important that subscriber count but even views don't tell the whole tale. I've watched the first 10 seconds of a YouTube video more times than I can count before deciding it wasn't for me, I bet you have too. But it counted as a view. Take it from a data geek, not all views are created equal and there's a reason YouTube doesn't recommend content to individual viewers solely based on view count. ☮️

One of the best and worst comments I often get is; "Your channel is awesome! I can't believe you don't have a million subscribers!" It really reflects how out of touch people can be when it comes to what real world YouTube success looks like for the vast majority of Creators. 99.94% of all YouTube channels DON'T have one million subscribers. (read that twice to let it sink in) Harvard (or any Ivy League) has an acceptance rate of 3–4% among applicants, meaning getting into an Ivy League college is 50–67 times easier than achieving 1 million subscribers on YouTube. I currently have 570,000 subscribers. With a ton of hard work and dedication, I get an average of 3 million views and 10,000 new subscribers every single month. If I keep grinding and can stay at this pace, that means I'll hit a million subscribers in late 2029. So, 8 years into this channel, I have to work this hard for another 3½ years and don't slip up or it'll take even longer... or maybe never happen. That's f*cking crazy. Listen to your old pal Daniel. Subscribers don’t matter as much as you think. You can be a massively successful YouTube without a million subscribers. Focus on connection with the people you make content for. 👉 Don't get distracted by vanity numbers. ☮️

@Daniel_Batal Yeah, so the way they’re presenting it is “monthly audience” in YouTube advanced mode. Definitely wouldn’t bank on raw views! I’d assume this is based on unique viewers, which I think is the best single stat to use, but could see them configuring this to combat fake views.

@kanekallaway @RobHoffman_ @markbrazil I’ve got an entire setup for podcasts that would probably interest you :)

For a long-form pod, the first thing I'm doing is renting a super sick professional space, with good cameras and a producer for $3-4K and I'm going to shoot 3 pilot episodes. My vision would be about as close to what @markbrazil does with OR. My goal with the first 3 episodes is... 1. To see if I like the medium and could commit to 100 episodes 2. To derisk the execution/production quality excuse so that isn't to blame when the first 3 episodes don't crush I'm taking the rest of the money and... 1. Hiring a goated thumbnail strategiest/designer to go all out on the thumbnail strategy. This is nonnegotiable. Can probably get this done for $300 per episode so $1200 total. Podcast thumbnails don't need to be Mr. Beast level complex, but they need extremely dialed copywriting and psychology. A beginner would overlook this, so this is the highest leverage. 2. For titles, I'm going to 1 of 10 and using their AI to give me a ton of initial title options. I'm also studying what other winning podcasts in my space have done, extracting top 1%, reverse engineering into title formats, and then remixing for variations in Claude/1of10. I'm having my thumbnail strategist gut check the psychology and packaging logic (this will cost ~$50) I'm chopping the footage myself (low effort edit) and shipping. Gritty but easy. No high production intro for these. If we had an extra $1K I'd skip the trailer and pay for shorts (which could do well because of the set) Inevitably, these episodes will not perform super well. I'd expect 500-2K views each. This is because long-form podcasts are extremely tough to grow from scratch given the risk/reward perception dynamic from a cold viewer on a 60+ min video. Only way to win with podcasts is to do a million, outlast, and increase entertainment/educational value overtime. This is why "seeing if you could do 100 episodes" is a critical initial measure. Most people quit because of a) the workload and b) they aren't proud of the product enough to keep feeling bad about it. The visuals help them "get proud faster" to give themselves a chance to keep going. ----------- For medium-form videos, I'd invest completely differently. It's apples and iphones. Entertainment vs education spaces play completely differently. I only focus on the education side, so will breakdown for that. In medium-form the set is a bonus, but definitely not a requirement -- and this is because people are watching with a higher tilt towards learning something specific vs vibes. In this case, I'd probably not invest in a studio on day 1. I'd use my iphone and shoot at 4K. I'd spend $150 on a great sounding mic. 1. I'd take $2-$3K and hire a cracked editor 2. I'd still pay for the thumbnails ($1200) 3. I'd book a 1 hour call with 1 of 10 Strategy (to write the titles/ideas for me) This would get me my first 4 videos. Very possible one hits 10K+ views. The substance covered in these videos is hopefully the inside the creator's brain already and for intros/scripting, just use what I give for free in my videos on my YouTube channel. Completely different strategies because the games are completely different. For someone to click and watch a podcast cold, they have to be willing to risk a lot of time for unclear reward. The visual execution and production quality is the fastest proxy for trust accumulation you could have. The conversation or information shared in the podcast obviously needs to be dialed and compelling, but I'm holding that constant either way, and money likely wouldn't improve this materially. For someone to click on an 18 minute video about Linkedin, storytelling, or peptides, they just need to be interested in the topic, captured by the packaging and then curiosity trapped in the intro. This can happen in a bedroom studio for sure. tl;dr -- I wouldn't start a podcast today unless I knew I had a visual set so premium/sick that it would jump off the feed

Nah, humans are visual first. The game on any algorithmic feeds from now on is visual hooking. Anything that blends in, (e.g. zoom slop) will lose unless king made from external forces/audience. For short-form, you can get away with being hot, interesting looking, or distracting away from poor set. I’d argue premium visuals are now close to mandatory for a new long-form show to have any chance. Premium visuals are a proxy for quality and trust when attention spans can’t be held for long enough to earn trust the slow way. The slow grind 1,000 episode in basement with tarp as background unfortunately won’t work moving forward just because of supply and demand. Set is moat until everyone can one click clone premium fake sets imperceptibly, and then agreed they won’t matter because they no longer cut through. Btw, that a16z show would have had zero chance with a trash visual set-up. That being said, set by itself of course is not enough, so will likely still fail

For anyone doing AI thumbnails, you should know a 1-shot approach is terrible. This is the way it should be done. I think AI-heavy thumbnails are for a younger audience, so not my thing, but if you're creating using AI then perhaps you'll find some great ways to use this.

@Liron_Segev 100% agree on the reverse prompt. Basically anything is possible as long as you know what you want. For me, I think I'm more just worried about the time sink on orchestration & whether it's worth it. There are enough AI distractions and not sure if it would be time well spent😅

I use Hermes and my own Claude both locally and in my VPS. Open a new session in whatever AI you use the most to build and ask: "Based on everything you know about me and the work we have been doing and the apps/ workflows we are building, I want to have an orchestrator but not sure how. Ask me any question so that when we are done, we have a clear plan on what to do, what it does, how it works, the costs and anything else I am not thinking about" See what it says. This is Reverse Prompting and has to be one of the most underrated elements of AI.

It's wild to me that I can be sitting on a plane and log into my dashboard and watch my AI build and deploy a new system for a client 🤯

@glenngabe @OfficialLoganK Same, and I'd love for them to be able to do a lot more with the Workspace automations as well. So much potential with Gemini, but all so fragmented/hidden. Long term Gemini/Google ecosystem is going to win out, but right now it just leaves you wanting more.

@RealJoshHoward @OfficialLoganK I'm in the Gemini early access program and just emailed them about "Scheduled actions" and how I think it's flying under the radar. I think it's pretty powerful... Easy to set up and you get notified by Gemini in several ways. Testing another interesting one today. :)

I think Gemini's "Scheduled actions" are flying under the radar. You can enter any prompt to run daily and Gemini will notify you in several ways (notifications on your phone, in Gemini, etc.) I set a few up recently and they work very well. "Scheduled actions" was announced in June 2025 btw but I don't see many people talking about it. "With scheduled actions, you can streamline routine tasks or receive personalized updates directly from Gemini. In your conversation, simply ask Gemini to perform a task at a specific time, or transform a prompt you're already using into a recurring action. You can manage them anytime within the scheduled actions page within settings." blog.google/products-and-p…

@oligilpin @LouixGilpin C2PA. Not for detection, but for verification. Could potentially help more in other scenarios(Google) to verify or speed it up, especially at your size. YouTube has been doing some interesting things with pattern matching thumbnails, which I assume is what happened here.

@RealJoshHoward @LouixGilpin can you share more?

Maybe you got demonetized because you stole my Atrium channel's thumbnails? YouTube figured that out without anyone reporting it. CC @LouixGilpin who worked mega hard with 10+ thumbnail concepts for each video we do...

@Daniel_Batal So sick! What an absolutely incredible view. (sorry about the lost work though😅)

Losing 15 hours of work because the public BETA of DaVinci Resolve 21 crashed and lost every auto-save… frustrating. But today’s office isn’t the worst 😁

@gregisenberg To that point, I have a system built that take this even further. If you’d like to chat I could share some pretty interesting things that attach to this that takes your idea to the next level 10x and I’ve already built it.

@gregisenberg YouTube has this built in natively now, so if your podcast is there then you’re good. Sneaky thing they added in January that shows you exactly which parts of a video held the best attention with exact timestamps.

We'll soon see the rise of the "Chief Clipping Officers" at companies The person who figures out the 47 second moment inside the 2 hour podcast that gets 10M views Probably will be the highest paid marketing hire of 2027

@geerlingguy Odd you say that. Had observed a video suggesting a client one in Studio, click it, and it’s literally someone playing several minutes and then adding commentary only. No content matching at all. Now I’m more curious.

The channel was removed! Thanks YouTube. Also nice to see shorts able to be (sorta) disabled on Mobile now. Want it in browser / desktop next! x.com/teamyoutube/st…