Finding Success With a Company YouTube Channel — Alex Kretzschmar ![rw-book-cover](https://d34adp677peecb.cloudfront.net/static/images/article2.74d541386bbf.png) 🔗 **Source:** [Finding Success With a Company YouTube Channel — Alex Kretzschmar](https://www.youtube.com/watch?v=tiEefGcH-OY) > [!note]+ Document Notes > tags: [developer-relations, company-youtube-channel, product-led-growth, content-workflow, audience-trust] > author: Alex Kretzschmar > source: Dev Rel (YouTube) > reference: https://www.youtube.com/watch?v=tiEefGcH-OY > date created: 2026-10-15 17:43 > > > [!info] Generated analysis — not the author's words. Sections below are machine-written > > from the captured source. Only text inside `> [!quote]` blocks is verbatim from the author. > > Do not quote or attribute any other line to the named author. > > # 🏷️ Title: Finding success with a company YouTube channel — Alex Kretzschmar > <!-- deep-read:sections start --> ## 🧭 Table of contents - [📝 Summary](#-summary) - [📖 Main content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning styles assessment](#-learning-styles-assessment) - [📊 Accessibility assessment](#-accessibility-assessment) - [🌍 DEIB assessment](#-deib-assessment) - [✅ Logical validity](#-logical-validity) - [🛡️ DARVO analysis](#️-darvo-analysis) - [🎯 Conclusion](#-conclusion) - [🔄 Refactor appendix](#-refactor-appendix--finding-success-with-a-company-youtube-channel--alex-kretzschmar) ## 📝 Summary - **What happened.** The Dev Rel channel published a 32:38 conference talk (uploaded 2026-10-14) from DevRelCon 2026 in NYC. Alex Kretzschmar, who the transcript says leads DevRel at Tailscale, explains how the Tailscale YouTube channel grew after Alex took it over on August 1, 2023, then answers four audience questions from 00:25:12 on. - **The claim.** Weekly, useful, honest videos that meet viewers where they are will grow a company channel and move business metrics. Alex says the channel started at 1,600 subscribers and 23,000 annual views, and reached over 75,000 subscribers and over 6 million annual views in 2025 after 1 million in 2024. - **The mechanism.** Four moves recur: publish on a fixed cadence (the switch from biweekly to weekly is credited for the curve at 00:02:17), pick topics from real customer problems, remove production friction (fixed room, window presets, a bank of 50 pre-shot thumbnail faces), and cut videos to what the target viewer needs (32 minutes down to 14). - **Why it matters.** Alex says YouTube was the number one discovery source for the whole Tailscale product until about two months before the talk (00:05:09). For a product-led growth company, that makes video the top of the funnel, not a side project. - **Key stakeholders.** Tailscale's DevRel team (now five or six people, 00:13:44), viewers at different skill levels, sales and marketing teams who consume the attribution, enterprise buyers such as a CISO, and the DevRelCon audience. - **Risks and tradeoffs.** Every result is a single-operator case study with no controls. The description gives different numbers (66k subscribers, 3 million views in 2025) than the speech (75,000 and 6 million). Alex also admits the attribution is soft ("vibe merchants," 00:18:02) and that the cadence is a long-haul load on one person. - **Bottom line.** The method is transferable and specific: cadence, audience-level targeting, friction removal, and cutting. The growth figures are unverified here, and the description and speech disagree on them. ## 📖 Main content **The core argument.** A company channel earns trust the same way any publisher does: by being useful repeatedly. Alex frames each video as a commercial that has to beat every other use of a viewer's evening (00:08:42 to 00:09:26), so the only durable edge is accumulated usefulness. Alex's own summary at 00:24:30 is four verbs: be useful, be intentional, don't rush, be sustainable. **Why YouTube fits a product-led growth company.** Tailscale's free tier lets people try the product at home with no sales conversation (00:05:09). Those people then bring the tool to work. DevRel's job in that motion is top-of-funnel awareness tied to real problems (00:05:52). The talk's cycle is discover, understand, succeed, then grow into the product (00:16:35). **Four video purposes.** The talk groups videos by what they do for the viewer and the business: - **Brand awareness.** Put the company logo next to a product the audience already searches for. The example is a self-hosted photo app, which the transcript spells "image" (almost certainly Immich, an open source Google Photos alternative). Nobody wakes up wanting a VPN, but people do want to stop paying for cloud photo storage (00:13:01). - **Education.** A numbered curriculum of 10 to 15 core-feature videos under one logo and playlist, called Tailscale Explain (00:14:26). - **Adoption of specific features.** A Talos Linux end-to-end guide, and a Kubernetes operator video made after the team came back from KubeCon saying nobody knew the operator existed (00:15:09). - **Product updates.** Alex says these podcasts "don't perform very well" (00:15:54) and uses that to argue against a views-at-all-costs mindset. **The audience-level problem.** The hardest part of a video is finding the start line (00:11:34). Alex uses a 2008 iPhone jailbreak guide that said to "just" SSH in as the example of expert blindness. A subnet-router video can skip networking basics, but a what-is-Tailscale video must start earlier. Alex calls the layering tactic "the Pixar thing": something for the kids and something for the grownups in the same video (00:12:17). **Cutting as a growth tactic.** A 32-minute video from month four at Tailscale was redone six months later at 14 minutes after reading comments and asking the product team for feedback. Alex says the shorter one now has more views and is embedded in documentation (00:19:28). The long one stays up for viewers who want kernel-level Docker networking. **Process.** Alex stopped recording first and editing later, because unscripted takes cost hours in the edit suite (00:20:11). For complex ideas, Alex records a walk-and-talk voice memo, feeds the transcript to a chat model to extract points, repeats up to five or six times, then deliberately does not read the model's output on camera (00:21:36 to 00:22:20). A fixed attic desk, a dedicated logged-in laptop, window-layout presets and a Final Cut headshot preset remove setup friction (00:23:02 to 00:23:46). **Q&A (from 00:25:12).** Four topics: refreshing videos that age (2.5-year-old Home Assistant video refreshed after a colleague noticed the matching blog post ranked second on tailscale.com), thumbnails (an hour-long session producing about 50 face variants), idea sourcing (an Obsidian note with hundreds of ideas), and Shorts (under two minutes, face at the bottom, UI at the top, Alex concedes it isn't planned up front). A final question asks how to prove video influence to sales. Alex says a call-transcript search tool returns thousands of hits for the name. **Contested or unverifiable in this source.** - **Subscriber and view counts.** The speech says over 75,000 subscribers and over 6 million 2025 views (00:03:00 to 00:03:42). The description says 1,600 to 66k subscribers in 2 years and over 3 million views in 2025. The description may be older copy, or the talk may be rounding up. Neither is checked against YouTube data here. - **Attribution.** "YouTube number one discovery" (00:05:09), "60, 80% or so" of traffic from YouTube (00:24:30), and Kubernetes operator adoption rising "as a result" of one video (00:15:09) are stated without a figure, window, or method. - **Platform advice.** The Mr. Beast algorithm-equals-audience claim (00:02:17) is attributed to a video Alex does not cite. ## 🗒️ Notes #### ⏱️ [00:00:08](https://www.youtube.com/watch?v=tiEefGcH-OY&t=0) Introduction and background - Alex was born in Huntington (about 15 miles from the venue), left at three for England, and now leads DevRel at Tailscale. - Side projects: the Self-Hosted podcast (six or seven years), a personal channel the transcript spells "KTZZ Systems," and the Tailscale channel. The dog is named Archie after Arch Linux. - Three-year anniversary at Tailscale is "next week" at 00:00:52. #### ⏱️ [00:01:33](https://www.youtube.com/watch?v=tiEefGcH-OY&t=93) Channel baseline and growth curve - Starting point on August 1, 2023: 1,600 subscribers, 23,000 annual views, no program, a few scattered efforts. Alex was hired as developer advocate with the remit to make videos. - A bot queries the Google API and logs view stats to a spreadsheet; the chart shows the curve bending after the move from biweekly to weekly, roughly March of the prior year. > [!quote] [00:01:33] > I picked up the channel on August 1st, 2023 and it at that point had 1,600 subscribers and it had 23,000 annual views. #### ⏱️ [00:02:17](https://www.youtube.com/watch?v=tiEefGcH-OY&t=137) Cadence trains the audience and the algorithm - Regular releases train both the audience and the recommendation system; Alex treats the two words as interchangeable. - Mr. Beast is cited (no link) for the idea of replacing "algorithm" with "audience" to humanize the work. - Current stats: over 75,000 subscribers, over 6 million annual views in 2025 after a million in 2024 (00:03:00 to 00:03:42). #### ⏱️ [00:03:42](https://www.youtube.com/watch?v=tiEefGcH-OY&t=222) What Tailscale is and product-led growth - Tailscale connects devices and services directly, without a central choke point. The example is a phone reaching a Home Assistant box on a Raspberry Pi. - Product-led growth means the product's usefulness is how the product grows. Only a couple of hands go up when Alex asks who works at a PLG company. - AI agent traffic has risen "in the last 6 months" (00:04:26 to 00:05:09), referenced from the previous talk. #### ⏱️ [00:05:09](https://www.youtube.com/watch?v=tiEefGcH-OY&t=309) YouTube as number one discovery channel - The free tier is the strongest marketing asset: no salesperson, home use first, then work adoption. - DevRel's role is top-of-funnel: brand awareness tied to real problems (00:05:52). > [!quote] [00:05:09] > YouTube was until about 2 months ago the number one discovery mechanism for Tailscale, the whole product. #### ⏱️ [00:06:35](https://www.youtube.com/watch?v=tiEefGcH-OY&t=395) "Can you just make me a video?" - The request sounds small but hides diagrams, animation, editing style, thumbnail, and description. - Alex says a good video must be entertaining, accurate, engaging, and honest (00:06:35). - Scope creep example at 00:07:17: one video becomes five videos in a three-day campaign. Alex calls that half flattering and half disrespectful of the process, while assuming the requester means well. - Retention-style editing with a cut every 7 seconds works but is "exhausting to watch." #### ⏱️ [00:08:42](https://www.youtube.com/watch?v=tiEefGcH-OY&t=522) A company video is a commercial - The agentic AI workflows thumbnail is Alex's favorite. The video distills what a harness and an agent are. - Publishing is the start of the life cycle, not the end, and the video sits next to everything else uploaded that day. > [!quote] [00:08:42] > A video for a company on a company channel whether you like it or not, it's a commercial. #### ⏱️ [00:10:10](https://www.youtube.com/watch?v=tiEefGcH-OY&t=610) Be useful, and know the start line - Questions to ask before any video: who is the target customer, and what is the product fit? For Tailscale: self-hosting, digital sovereignty, leaving OpenVPN, connecting things. - Expert blindness: after 15 years in IT, Alex can't remember being new. The 2008 iPhone jailbreak guide that said to "just SSH" is the example (00:10:52). > [!quote] [00:10:10] > And so I I strive to make videos which exist to be seen, not videos that are merely seen to exist. #### ⏱️ [00:12:17](https://www.youtube.com/watch?v=tiEefGcH-OY&t=737) The Pixar approach and brand awareness - Layer content: Vim key bindings for the experienced alongside SSH basics for beginners. - Brand awareness means putting the logo next to a product the audience already seeks. The photo-app example surfaces at 00:13:01 (transcript spelling "image," probably Immich). #### ⏱️ [00:14:26](https://www.youtube.com/watch?v=tiEefGcH-OY&t=866) Education curriculum and feature adoption - Team went from a solo effort to five or six people; limited calories mean picking carefully when it's one or two people. - A 10 to 15 video playlist under one logo teaches core features. Talos Linux guide filled a gap Alex found: no end-to-end getting-started video. - After KubeCon, colleagues reported nobody knew about the Kubernetes operator. Alex says that video increased operator adoption (no number given). #### ⏱️ [00:15:54](https://www.youtube.com/watch?v=tiEefGcH-OY&t=954) Beyond the zero-sum views mindset - Product-update podcasts perform poorly, which Alex uses to argue that more views is not the only goal. - One useful video reaching one CISO can drive a top-down adoption, which helps sales and marketing too. DevRel is "a cog in a wider machine." - The four video types map to discovery, understanding, success, and growth (00:16:35). #### ⏱️ [00:17:20](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1040) Authenticity and business metrics - A two-part series passed half a million views (no dates). - Alex says finding a company whose product matches personal interest is hard; passion comes through when real. - Metrics named: sign-ups. Alex quotes someone from last year's conference calling DevRel "vibe merchants" and cannot recall who said it. #### ⏱️ [00:18:45](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1125) Respect time: 32 minutes to 14 - Viewer frustration at "get to the point" is the test for cutting. - The 32-minute video was made four months into the job; the 14-minute rework came six months later after reading comments and asking the product team for feedback. - The shorter version has more views and is embedded in documentation. Alex keeps the long one for viewers wanting kernel-level depth. > [!quote] [00:18:45] > If you are respectful of the audience's time, again, they will choose to watch you next time because you're building that trust #### ⏱️ [00:20:53](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1253) Preparation and using AI for scripting - Alex used to turn the camera on first, then burned hours in the edit suite because the problem wasn't understood. - AI use is limited to scripting: an hour-long walk-and-talk transcript goes to chat for a point list, repeated five or six times for dull topics. > [!quote] [00:20:53] > We all know that you can't teach something you don't understand, at least not properly. #### ⏱️ [00:22:20](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1340) Do not read the model output on camera - Reading generated text means presenting the statistical mean of the English language. Viewers tune in for Alex's opinions and unplanned jokes. - Friction list at 00:22:20 to 00:23:46: lights, camera, recording monitor, logged-in accounts, do-not-disturb. The fix is a fixed attic desk, a dedicated laptop, a window-layout app (the transcript spells it "Bento Box"), and a Final Cut headshot preset. #### ⏱️ [00:24:30](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1470) Discovery data and the four principles - Chart shown: YouTube search and browse. Alex says 60 to 80% of traffic comes from YouTube itself, so "YouTube recommends YouTube." - Principles: be useful, be intentional, don't rush, be sustainable. A video a week for 6 months or a year before judging. > [!quote] [00:24:30] > DevRel is a long-term investment. #### ⏱️ [00:25:12](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1512) Q&A: refreshing aging videos and thumbnails - Alex never says "like, comment, and subscribe" in videos; useful content earns the next click. - Evergreen answer: videos have an "entropy." A Home Assistant video, 2.5 years old, was refreshed after a team monthly business review showed the related blog post was number two on tailscale.com. The plugin UI had changed. "Fresh sauce" in the air fryer, not the microwave. - Thumbnails: about three years ago Alex shot 15 shirts and an hour of faces on a time lapse, leaving about 50 headshot variants. Alex refuses silly faces to keep a calm, trustworthy brand. #### ⏱️ [00:28:06](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1686) Q&A: where ideas come from - An audience member calls recycling content "reheating our own nachos." - Alex files weekend project ideas as "content brain" entries in an Obsidian note holding hundreds, half never made. - Topic selection follows customer problems that are adjacent to the product (Google Photos has nothing to do with Tailscale; connecting things does). Finding a company that speaks to you makes ideas come naturally, which Alex calls "really bad advice." #### ⏱️ [00:29:30](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1770) Q&A: Shorts and repurposing - Shorts can run 3 minutes; Alex aims for under 2, ideally 90 seconds. - Layout: face at the bottom, UI at the top, giant font. Alex has "no better answer" for showing UI on mobile. - Alex does not plan long-to-short repurposing, says the social media manager will object, and calls it a company gap. #### ⏱️ [00:30:57](https://www.youtube.com/watch?v=tiEefGcH-OY&t=1857) Q&A: proving influence to leadership - A call-transcript tool lets Alex search the name and see thousands of results from sales calls. - Conference attendees report watching the videos. Some enterprise prospects shared videos with their C-suite, and a former developer turned CISO gets closer to the technology. - Closing line at 00:32:23: "meeting people where they are" is the big thing. ## 📚 References The transcript itself contains no links; the following come from the video description or are primary sources for named entities. - [Finding success with a company YouTube channel. Alex Kretzschmar](https://www.youtube.com/watch?v=tiEefGcH-OY). the source video, 32:38, uploaded 2026-10-14 to the Dev Rel channel. - [DevRelCon NYC 2026](https://nyc.devrelcon.dev/). the conference where the talk was presented, per the description. - [Tailscale](https://tailscale.com). the product and employer; its YouTube channel is the case study. - [Immich](https://immich.app). likely the self-hosted Google Photos alternative the transcript spells "image." Verify against the video's on-screen logo before citing. - [Talos Linux](https://www.talos.dev). the Kubernetes distribution Alex made an end-to-end getting-started video for. - [Obsidian](https://obsidian.md). where Alex keeps the idea backlog. - Mr. Beast video on algorithm versus audience, named at 00:02:17, no title or link given. - Alex's own channels (Self-Hosted podcast, "KTZZ Systems," the Tailscale channel), named at 00:00:08 to 00:00:52 without URLs. ## 🗣️ Paraphrase Alex took a company channel with 1,600 subscribers and no strategy in August 2023 and says it passed 75,000 by October 2026, with the steepest growth after switching from biweekly to weekly releases. The point of the talk is that a company video is a commercial competing with everything else on YouTube, and the only way to win that competition repeatedly is to be useful repeatedly. Usefulness, as Alex describes it, comes from knowing where the viewer starts. A subnet router video can skip networking basics. A what-is-Tailscale video can't. Alex layers material so beginners and experts both get something, and tests ideas against real customer problems, such as moving off paid photo storage, rather than product features. Most of the practical advice is about effort. Cut videos down (32 minutes became 14 and gained views). Script by talking through ideas into a voice memo, but don't read the model's summary on camera. Build a recording corner that needs no setup and shoot 50 thumbnail faces in one sitting. The Q&A adds that old videos should be refreshed when the product UI changes, that Shorts are an unsolved layout problem for UI demos, and that sales-call transcripts are where Alex finds proof the videos reach buyers. ## 📰 Newsletter **One DevRel hour a week can build an audience, if you cut the right things.** At DevRelCon NYC, Tailscale's Alex Kretzschmar walked through how a company YouTube channel went from 1,600 subscribers in August 2023 to a reported 75,000-plus. The most reusable ideas are mechanical. Ship on a fixed cadence (the weekly switch is credited for the curve). Pick topics from customer problems adjacent to the product. Remove setup friction with a fixed recording room and presets. Cut videos to the viewer's start line: a 32-minute explainer became a 14-minute one that now sits in the docs and has more views. For WordPress teams, the closest parallels are the Training and Documentation teams. A numbered curriculum of 10 to 15 videos under one playlist (Alex's "Tailscale Explain") maps to Learn WordPress lesson paths, and refreshing a video when the admin UI changes maps to the screenshot-drift problem every handbook has. One caution: the description says 66k subscribers and 3 million views in 2025, while the speech says 75,000 and 6 million. Use the range, not a single figure. Links: [the talk](https://www.youtube.com/watch?v=tiEefGcH-OY), [DevRelCon NYC](https://nyc.devrelcon.dev/), [Tailscale](https://tailscale.com). ## ✅ Action items #### DevRel - Developer advocacy: pick one weekly or biweekly slot and publish for six months before judging results. Alex's own rule is a video a week for 6 months to a year (00:24:30). - Education: build one numbered 10 to 15 video playlist on core features under a single named series, as the Tailscale Explain curriculum does (00:14:26). - Developer experience: write a start-line statement (what the viewer already knows) at the top of every video brief; skip basics for advanced topics, start earlier for introductions. - Developer marketing: pair a brand-awareness video with an adjacent product the audience already searches for, and track the logo-adjacent search terms. - Community management: read comments on long videos and feed them to the product team before cutting a shorter version, as the 32-to-14 minute rework did (00:19:28). - Internal/external/cross-project: when someone asks "can you just make me a video," reply with the checklist at 00:06:35 (diagrams, animation, edit, thumbnail, description) so scope is visible before agreeing. #### OSPO - Policy and governance: no direct action from this source; nearest relevance: a short disclosure rule for company-channel videos, since Alex states each one is a commercial (00:08:42). - License compliance: no direct action; nearest relevance: confirm third-party logos and open source app screenshots (the photo-app example) are used within their trademark terms. - Security management: no direct action from this source. - Software inventory: no direct action from this source. - Developer support: no direct action; nearest relevance: videos for open source tools double as support deflection, so link the video from the matching docs page, as Alex does with the 14-minute Docker video. - Community engagement: no direct action from this source. - Strategic alignment: define which two business metrics a channel is expected to move (sign-ups at minimum, per 00:18:02) and who attributes them, since Alex concedes DevRel is "vibe merchants" without them. #### Community - Maintainers: no direct action; nearest relevance: a maintainer walkthrough video for the top three setup questions. - Contributors: record a start-here video for new contributors that skips what a first-time contributor already knows. - Sponsors: no direct action from this source. - Ecosystem: no direct action; nearest relevance: cross-project collaboration videos (Alex's Talos example). #### WordPress - Core: no direct action from this source. - Core Performance: no direct action; nearest relevance: short explainer videos for each performance feature in a release. - Core AI: no direct action; nearest relevance: Alex's rule of not reading model-generated script on camera (00:21:36 to 00:22:20) applies to any AI-assisted explainer. - **Core Program Team:** no direct action from this source. - Mobile: no direct action; nearest relevance: Alex's Shorts layout note (face at the bottom, UI at the top, giant font) for app demos. - Playground: pair each Playground blueprint with a two-minute video, since UI demos are hard in vertical formats. - Community: no direct action; nearest relevance: WordCamp session recordings re-cut to 14 minutes with chapters. - Training: build a numbered playlist per lesson path and refresh the lesson video when admin UI changes (00:25:56 to 00:26:40). - Marketing: audit existing video against the four purposes (awareness, education, feature adoption, updates) and note which are missing. - Support: link a short video from the top ten support replies, and measure whether tickets drop. - Design: no direct action from this source. - Documentation: embed the shorter video in the matching handbook page, as Alex did with the 14-minute video. - Photos: no direct action from this source. - TV: no direct action; nearest relevance: submission guidelines for length (14 minutes versus 32) and for refreshing aged workshops. - Polyglots: no direct action from this source. - Accessibility: no direct action; nearest relevance: human-corrected captions on every published tutorial. - Meta: no direct action from this source. - CLI: no direct action; nearest relevance: a short terminal-demo video series using fixed window presets. - Hosting: no direct action from this source. - Tide: no direct action from this source. - Openverse: no direct action from this source. - Themes: no direct action; nearest relevance: a block theme tour video for the top ten features. - Plugins: no direct action; nearest relevance: plugin-author video guidelines (start line, one purpose, cut to the point). #### The WPOCC - Programs: no direct action; nearest relevance: if The WP Open Community Collective funds contributor media, require a stated target viewer and start line per grant. - Projects: no direct action from this source. - Funding experiments: no direct action; nearest relevance: fund a six-month cadence rather than one-off videos, since Alex says one-off attempts can't be judged. ### FAIR - Technical independence: no direct action from this source. - Policies and covenants: no direct action; nearest relevance: disclosure of commercial relationships in published explainers. - Productized governance artifacts: no direct action from this source. ## 🧪 Hypotheses 1. A fixed weekly upload cadence raises 12-month views per video more than a biweekly cadence with the same total production hours (supported by the curve change at 00:02:17, with no control). 2. Re-cutting a long tutorial to about 40 to 50% of its length, using comment data, raises views and docs embed rate (the 32-to-14 minute example at 00:19:28). 3. Videos aimed at a stated viewer start line retain viewers longer past the 30% mark than videos that open with general product background. 4. A fixed recording setup with scripted presets shortens time from idea to publish by at least 30%, compared with building the setup per video (00:22:20 to 00:23:46). 5. A thumbnail bank of pre-shot faces reduces thumbnail time per video to under 10 minutes without lowering click-through rate (00:26:40 to 00:27:22). 6. Company-channel videos produce a countable footprint in sales-call transcripts, and keyword mentions correlate with enterprise pipeline entry within 90 days (00:30:57). ## 📊 Methodology - **H1, cadence.** Use channel analytics for 12 months before and after a cadence change. Compare views per video at days 7 and 28, normalizing for topic, using at least 20 videos per arm. Where possible, run staggered cadence across two comparable playlists. - **H2, re-cuts.** Pick 10 long videos with comments. Re-cut half using comment and product-team feedback, and leave half as controls. Compare average view duration, relative retention at the median point, and docs-embed click-through after 60 days. - **H3, start line.** Tag videos with a stated audience level in the brief. Compare retention curves for beginner and advanced videos with and without an explicit start-line opener, using YouTube retention graphs. - **H4, setup.** Time-log three production stages (setup, record, edit) for 15 videos before and 15 after a fixed-room process, and compare medians. - **H5, thumbnails.** A/B test thumbnails using YouTube's test-and-compare feature for 20 videos. Record minutes per thumbnail from a time log. - **H6, attribution.** Export sales-call transcripts for a fixed quarter and count named references to the channel or presenter. Match call dates to the account's first sign-up and first YouTube view where analytics allow. Pre-register the 90-day window. - **Controls throughout.** Record topic category, length, publish day, and external events such as launches so cadence effects aren't confused with a viral hit. ## ❓ Questions - Which figures are right for subscribers and 2025 views, the speech (75,000 and 6 million) or the description (66k and 3 million)? - What was the "number one discovery mechanism" measured against, and over what window (00:05:09)? - How many views did the 14-minute video get compared with the 32-minute original, and in what time window (00:19:28)? - What is the cost of one weekly video in hours, and how did that change once the team grew to five or six (00:13:44)? - Does the weekly cadence hold when a video needs five diagrams and an animation, or does quality drop? - How does Alex separate a video's own effect from the blog, events, and docs in the attribution claims? - What did the Kubernetes operator adoption lift look like in numbers (00:15:09)? - What is the plan for Shorts if the company wants more than long form, given the admitted gap (00:30:15)? ## 🔮 Future research - **Single-operator survivorship.** Alex's channel is a case study of one person who enjoys the topic. A comparison across channels where the presenter has less personal interest would test the "if I wasn't being paid I'd do it anyway" assumption (00:17:20). - **Attribution beyond sales-call search.** Searching call transcripts for a name shows presence, not influence. A matched-pair study of accounts with and without video exposure would test it. - **Length versus retention for technical tutorials.** The 32-versus-14 example is one pair. A larger sample would show where the length curve flattens for tutorial content. - **AI in scripting.** Alex limits AI to extracting points from a rambling transcript. A study of whether that workflow changes distinctiveness or accuracy would help teams set their own policy. - **Refresh cadence.** The talk gives one trigger (a blog page ranking second). A model for when UI drift makes a video wrong enough to replace would help small teams. ## 💡 Implications - **Practice.** Cadence, a fixed recording setup, and a start-line statement are cheap to copy and don't depend on Tailscale's audience. Set them up before the first video. - **Policy.** Teams should state which business metrics a channel owns and who counts them, because Alex says attribution is soft. A written disclosure that each video is a commercial also matches Alex's own framing at 00:08:42. - **Security.** Walkthrough videos show real configuration. Teams should scrub tokens, hostnames, and customer data from screen recordings, a step this talk does not discuss. - **Education.** The start-line habit (what does the viewer already know?) is a transferable instructional design rule, and the Pixar layering gives a way to serve two levels. - **Funding.** The advice that one-off attempts can't be judged supports multi-month budgets for media programs, not per-video grants. - **Career.** A solo DevRel hire given a video remit needs a process to scale into a team of five or six (00:13:44), including a thumbnail and recording system that others can reuse. ## 🧠 Learning styles assessment - **Reading level estimate.** Flesch-Kincaid Grade Level on the transcript: about grade 6 to 7, with Flesch Reading Ease near 75. The register is spoken, conversational English with short sentences. Domain terms (product-led growth, top of funnel, retention-based editing, Kubernetes operator, subnet router) raise effective difficulty for readers new to DevRel or networking. #### 👀 Visual learners - Present: growth chart, thumbnails (some pulled with ChatGPT for competition), a traffic chart, and two side-by-side videos for the length comparison. - Improve: a single diagram of the discover, understand, succeed, grow cycle (00:16:35) would help; no image of the attic setup beyond a mention. Describe chart axes aloud, because the numbers are on screen only. #### 🔊 Aural learners - Present: full spoken delivery with story anecdotes (2008 iPhone jailbreak, voice memos on a dog walk), plus audience laughter and Q&A. - Improve: the transcript shows overlapping speech in Q&A; repeat each audience question before answering so audio-only listeners can follow. #### 🗣️ Verbal learners - Present: analogies (Pixar layering, air fryer versus microwave, nachos) and plain explanations of product-led growth. - Improve: define top of funnel and retention-based editing in one clause each. #### 🧩 Kinesthetic learners - Present: concrete setup steps (fixed desk, window presets, 15 shirts for 50 face variants). - Improve: a downloadable checklist or template (brief, start-line statement, thumbnail bank) would make the process doable the same day. #### 🧠 Logical learners - Present: a clear cause chain (cadence, algorithm, views) and four-purpose taxonomy; before and after example with a measurable cut. - Improve: give the numbers behind the attribution claims, and show the data for the cadence change (views per video before and after March). #### 🧑‍🤝‍🧑 Social learners - Present: live Q&A with four questioners, audience polling by show of hands, team references (a colleague named Kevin, the social media manager). - Improve: no community channel or discussion prompt is given for continuing the conversation. #### 🧘‍♀️ Solitary learners - Present: self-contained talk with a stated checklist of four principles at 00:24:30. - Improve: add a reflection prompt such as "write the start line for your next video in one sentence." #### 🌿 Naturalistic learners - Present: the walk with the dog as a thinking process and a time-lapse; photography is mentioned in the intro. - Improve: not addressed beyond incidental references; a sustainability angle (a long-haul cadence that doesn't burn out one person) is named at 00:24:30 but not developed. ## 📊 Accessibility assessment #### Screen reader semantics - Present: YouTube supplies the page landmarks, a transcript panel, and a description with the speaker name and conference link. - Improve: the description has no chapters, and the transcript has no speaker labels, so a screen reader user can't tell Alex from audience questioners in the Q&A. WCAG 2.2 AA 2.4.6 (Headings and Labels) and 1.3.1 (Info and Relationships). #### Keyboard navigation - Present: the YouTube player is keyboard operable with visible focus. - Improve: the video adds no barriers, but with no chapters, keyboard users must scrub to skip to the Q&A at 00:25:12. WCAG 2.4.1 (Bypass Blocks) is met by the platform only partly. #### Contrast and readability - Present: unverifiable from a transcript; slides and a chart are shown on screen. - Improve: slide text and chart labels should be checked at 4.5:1 and at small-screen size, per 1.4.3 (Contrast Minimum). #### Text resizing and reflow - Present: the player and captions scale on their own. - Improve: burned-in slide text doesn't reflow (1.4.4 Resize Text, 1.4.10 Reflow). A downloadable slide deck would help. #### Media and captions - Present: auto-captions exist, which is how this analysis was produced. - Improve: they mangle names and terms: "KTZZ Systems" for a channel name, "image" for what is probably Immich, "Bento Box" for a window-layout app, and "Ben Hur link epic" for what reads as "Ben Hur length." A live-event recording also has no audio description for slides and charts. WCAG 1.2.2 (Captions, Prerecorded) in substance, 1.2.5 (Audio Description, Prerecorded) for on-screen charts, and 1.2.4 does not apply because the video is not live. #### Forms and interactions - Present: no forms. Audience Q&A is verbal. - Improve: the speaker asks viewers to like, comment, and subscribe at 00:25:12 but says that's rare; a text call to action in the description would be selectable. 3.3.2 (Labels or Instructions) applies to any landing page that follows. #### Standards reference - Primary gaps against WCAG 2.2 AA: **1.2.2** (inaccurate captions on proper nouns), **1.2.5** (no description of the charts and thumbnails), **1.3.1** (no speaker labels or chapters in the transcript), **1.4.4 / 1.4.10** (burned-in slide text). The caption fix is the cheapest and highest impact. ## 🌍 DEIB assessment #### Cultural breadth - Present: the speaker is an English-born presenter living in the US, the examples span US and UK life (10 pounds a month versus 9 bucks), and several products serve a global self-hosting audience. - Improve: examples are English-language and home-lab-centered; add examples from non-English YouTube markets, and discuss subtitles and translation as a growth lever. #### Inclusive language - Present: jargon such as product-led growth, top of funnel, and retention-based editing is explained or contextualized by example. Alex repeatedly names newcomers as the audience. - Improve: idioms (cog in the machine, vibe merchants, heavy lifting, reheating nachos) and Ben Hur-length references are US and UK slang that second-language readers may skip. "Insane" and "crazy" appear as intensifiers (00:05:09, 00:31:41), and plainer words would work. #### Neurodiversity support - Present: talk structure is clear (growth, strategy, four purposes, process, Q&A). The four principles at 00:24:30 give a memorable summary. - Improve: the talk has asides and tangents (Huntington birthplace, dog names) without chapter markers, which makes scanning hard for viewers who need predictable structure. Retention-style editing is called "exhausting," and Alex's own long-form pace is a plus. #### Representation - Present: one named speaker, Alex, plus several unnamed questioners; a colleague named Kevin and Justin Garrison are mentioned by name. - Improve: the audience questioners are not named or described, and the talk doesn't discuss how company channels represent diverse presenters or audiences. #### Psychological safety - Present: Alex admits early videos "are not very good" (00:20:53), shares that they used to record unscripted, and invites questions with humor. The audience response is warm (laughter, a questioner calling it cathartic). - Improve: the "I don't mean to single him out" aside at 00:28:48 about another company's product could land as a jab; naming no one would avoid that. ## ✅ Logical validity #### Strong reasoning - **Causal, 00:02:17.** Weekly cadence is paired with a visible curve change, and the claim that regular releases train audience and algorithm is stated along with the date of the switch. Strengthen by plotting views per video before and after, not total views. - **Inductive, 00:19:28.** The 32-minute versus 14-minute comparison is a before-and-after on one topic, with a stated cause (comments and product feedback) and a result (more views, embedded in docs). Strengthen with view counts and the same time window. - **Counter-case handling, 00:19:28 to 00:20:11.** Alex keeps the long video for viewers wanting kernel-level depth. Serving both groups avoids a false either-or and is the talk's best-argued point. - **Analogical, 00:12:17.** The Pixar layering (something for kids, something for grownups) maps cleanly onto the beginner and expert split described earlier at 00:11:34. - **Self-limiting disclosure, 00:18:02 and 00:30:57.** Alex says DevRel can't always attribute to metrics ("vibe merchants") and says Shorts planning is a company gap. Conceding these raises credibility of the rest. #### Fallacies - **Survivorship and single-case generalization, 00:24:30.** "All of my traffic, what's that, like 60, 80% or so, comes from YouTube." and the advice to post weekly for 6 months to a year rest on one channel with a product the presenter loves. Why weak: no comparison with channels that stopped, no sample. Fix: cite two or three counter-cases or say "this worked here." - **Unsupported quantification, 00:03:42.** "over 6 million annual views in 2025" conflicts with the description (3 million views in 2025). Why weak: two numbers for one fact. Fix: state the source (YouTube Studio, date range) and align the description. - **Post hoc, 00:15:09.** "we know that our Kubernetes operator, as a result of this video, saw more adoption" attributes an outcome to one video. Why weak: other launches and KubeCon follow-up happened at the same time. Fix: show the adoption chart with the publish date and controls. - **Appeal to a hearsay authority, 00:02:17 and 00:18:02.** A Mr. Beast video and an unnamed conference speaker are cited by memory ("I forget who said it"). Why weak: neither is checkable. Fix: link or omit. - **Possible conflict of interest, throughout.** The speaker is paid to grow the channel and presents the program's success as evidence the method works. Why weak: self-reported results. Fix: share the raw analytics export or an independent figure. - **Equivocation on "useful," 00:10:10.** Usefulness is used to mean educational, entertaining, and brand-building across the talk. Why weak: the claim "be useful" can't be tested without a metric. Fix: define one measure of usefulness, such as retention at the median point or return viewers. #### Tone of voice - Warm, self-deprecating, and conversational, with jokes (the dog named Archie, "the pet tax," "reheating our own nachos"), as at 00:00:52. The register firms up around 00:20:53 to 00:24:30, where Alex gives direct advice ("remove friction at every point") after earlier storytelling. Claims are personal and hedged ("for me," "I think"), and the talk closes on a modest ask at 00:25:12 about liking and subscribing, which Alex says is rarely made in a video. ## 🛡️ DARVO analysis Not applicable to this source. DARVO describes a defensive response to accusation or criticism, and this is a conference talk with friendly Q&A. For completeness across the three prongs: **Deny**, nothing is denied; Alex concedes the unplanned Shorts gap (00:30:57) and soft attribution (00:18:02). **Attack**, no critic is attacked; the nearest edge is the aside about making a MongoDB video every week (00:28:48), which Alex flags as unintended to single anyone out. **Reverse victim/offender**, no such move appears. Emotional cues are enthusiasm, gratitude (the "this was incredibly cathartic" comment from an audience member at 00:25:12), and humor. ## 🎯 Conclusion The transferable idea is that a company channel works when each video is built for a named viewer at a named start line, shipped on a fixed cadence, and cut to what that viewer needs. Alex's evidence is a single channel's growth from 1,600 subscribers to over 75,000, with the weekly switch and the 32-to-14 minute rework as the two clearest mechanisms. The weakest points are the figures themselves (description and speech disagree) and attribution that Alex calls vibes. **Next steps for readers:** write the start-line sentence for your next video, set up a recording space that needs no setup, and shoot a bank of thumbnail faces in one session. Before citing any growth figure, check subscriber and view counts against the channel's own analytics. Then pick a cadence you can keep for six months, since Alex says anything shorter can't be judged. ## 🔄 Refactor Appendix — Finding success with a company YouTube channel — Alex Kretzschmar > Density: 2–5 Literature atoms (H3 pad), 0–3 Permanent candidates (H4 pad). > Emitted: 5 Literature, 2 Permanent. Resonance filter applied: the Mr. Beast > and thumbnail-shirt anecdotes were dropped as unevidenced or incidental. > > Konik claim-header rules: present tense, declarative, ≤10 words, searchable, unique. > > Atomicity: one declarative sentence per atom. "and"/"also" → split. > > Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms ### 📚 Literature Split Pad #### [H1] Regular uploads train both audience and algorithm > [!quote] [00:02:17] > the regular videos, being regular, feeding the algorithm regularly with what you're doing, it trains the audience to expect that. **Context & use**: - 00:02:17, after the chart showing the curve change when Alex moved from biweekly to weekly. - The claim is causal but uncontrolled: one channel, no comparison group. Keep the cadence rule, flag the evidence as a single case. **Tags**: #xref #share <!-- atomicity: 1 idea ✓ --> #### [H2] Videos should exist to be seen > [!quote] [00:10:10] > I I strive to make videos which exist to be seen, not videos that are merely seen to exist. **Context & use**: - 00:10:10, right after the point that a company video competes with every other use of a viewer's time. - Usable as a one-line test for any content brief: who sees this, and what do they do next? **Tags**: #xref #aphorisms <!-- atomicity: 1 idea ✓ --> #### [H3] Expert blindness hides the viewer's real start line > [!quote] [00:11:34] > half your audience are probably you 15 years ago thinking, "What's he on about?" **Context & use**: - 00:11:34, following the 2008 iPhone jailbreak guide that said to "just" SSH in. - Applies directly to docs and tutorials: the hardest question is what the reader already knows. **Tags**: #xref #share #articleseed <!-- atomicity: 1 idea ✓ --> #### [H4] One useful video can land a whole company > [!quote] [00:15:54] > if you're useful to one CISO, and that entire company decides, "I'm going to adopt Tailscale from the top down," **Context & use**: - 00:15:54, in the argument against a zero-sum views and subscribers mindset. - Evidence is anecdotal. The 00:30:57 sales-call search is the supporting data point, and it measures presence, not influence. **Tags**: #xref <!-- atomicity: 1 idea ✓ --> #### [H5] Removing friction makes regular publishing sustainable > [!quote] [00:22:20] > Remove friction at every point in the process. **Context & use**: - 00:22:20, introducing the fixed attic desk, dedicated laptop, window presets and headshot preset that follow through 00:23:46. - The sibling claim at 00:24:30 ("DevRel is a long-term investment") is why friction matters: the cadence has to last 6 months or more. **Tags**: #xref #aphorisms <!-- atomicity: 1 idea ✓ --> ### 🧠 Atomic Split Pad #### [P1] Aim every video at a stated start line **Claim** (one sentence, Courtney's voice): A tutorial works for its reader only when I know what that reader already knows and start exactly there. **Evidence pointer**: [[#H3]] · [[#H2]] **Scope & conditions**: Holds when the audience level is knowable, as with a subnet-router video versus a what-is-the-product video. It breaks when one piece must serve two levels, which Alex handles by layering (00:12:17) or splitting (the 14-minute and 32-minute versions). Evidence is one presenter's experience, not a test. **Links**: - [[Docs audience mapping]] — because extends - [[Diátaxis tutorial versus how-to]] — because contrasts on how much background each form assumes - [[Onboarding documentation]] — because applies-in first-run readers <!-- atomicity: 1 idea ✓ --> #### [P2] A sustainable cadence needs friction removed first **Claim** (one sentence, Courtney's voice): Publishing on a fixed schedule only lasts if setup, recording, and packaging steps are scripted so starting takes no decisions. **Evidence pointer**: [[#H1]] · [[#H5]] **Scope & conditions**: Applies to solo or very small teams producing recurring media. Alex gives no time-saved numbers, so treat the effect as unmeasured. It does not cover content that needs original research for each piece. **Links**: - [[Content calendar discipline]] — because extends - [[Reducing friction in writing workflows]] — because applies-in documentation drafting - [[Quality versus cadence]] — because contrasts on whether speed trades off against polish <!-- atomicity: 1 idea ✓ --> ### ✏️ Fleeting Queue - [F1] Check the real subscriber and 2025 view counts: the speech says 75,000 and 6 million, the description says 66k and 3 million. https://www.youtube.com/watch?v=tiEefGcH-OY #fleeting #inbox - [F2] Confirm the "image" app at 00:13:01 is Immich and the "Bento Box" app at 00:23:46 is the window-layout tool Alex means, from the on-screen logos #fleeting #inbox - [F3] Open question: what is the per-video hour cost of weekly publishing, and how did it change when the team grew to five or six (00:13:44)? #fleeting #inbox - [F4] Article seed: "Start-line statement" as a one-line field for every doc and video brief, based on Alex's 2008 iPhone SSH story at 00:10:52 #fleeting #inbox #articleseed - [F5] Idea: keep a running "content brain" note of weekend-project ideas, as Alex does in Obsidian with hundreds of entries (00:28:06) #fleeting #inbox - [F6] Track whether short, sub-90-second UI demos with the face at the bottom and large type (00:30:15) work for docs-style how-tos #fleeting #inbox ### → Things 3 (max 2 actions) - [ ] Write a start-line sentence at the top of the next tutorial or video brief. docs and DevRel workflow, 15 min #things - [ ] Reconcile the subscriber and view figures with Tailscale channel analytics before citing the talk anywhere. research, 15 min #things <!-- deep-read:sections end --> 📄 **Full text:** [[Full Document Contents/Articles/Finding Success With a Company YouTube Channel — Alex Kretzschmar.md|Finding Success With a Company YouTube Channel — Alex Kretzschmar by Alex Kretzschmar]] ## Highlights - [00:01:33] the last 3 years. Um I picked up the channel on August 1st, 2023 and it at that point had 1,600 subscribers and it had 23,000 annual views. Basically no presence of any kind or program. There was there was a few efforts with like a tailscale.dev website and there was a few people doing a few things uh with NixOS. It was it was all very kind of just sort of back of an envelope best effort type stuff. Um and I was hired in as a developer advocate with the remit of make videos and stuff. That was pretty much it. So I I had to figure it out. Now this is the this is a chart I I just keep a spreadsheet of all the I have like I've written a little bot now that goes and just queries the Google API and pulls [00:02:17] all of the view stats and things like that and you can you can see the the curve that we've had here. Um and in in uh probably 18 months ago I switched to going from bi-weekly to doing weekly videos. So where would that put us? Roughly March, I guess, last year. And look how the curve's been since March. So the first takeaway is the regular videos, being regular, feeding the algorithm regularly with what you're doing, it trains the audience to expect that. It trains the algorithm and that those two words kind of are interchangeable. There's um there's a famous Mr. Beast video. Didn't expect to be quoting Mr. Beast here, but um there's a famous video of his where he says if you just replace ([View Highlight](https://read.readwise.io/read/01m50s7cdf4d0yf0dxwsmyjrg1)) - Note: [H1] Regular uploads train both audience and algorithm - Tags: atom/literature/h1 - [00:10:10] them. And so I I strive to make videos which exist to be seen, not videos that are merely seen to exist. And the way I do that is I I ask a few questions along the way. Like, someone comes to me with that request of, "Can you just make me a video?" Like, do you know your target customer, your target market? What What's your product fit? For us at Tailscale, it is self-hosted, it's people wanting to do something around digital sovereignty, it's people wanting to move off OpenVPN, it's people wanting to connect different things together or share a photo with a friend, those kinds of things. And [00:10:52] I think what we lose sight of a lot in the IT industry, particularly, you know, I've been in the industry for 15 years at this point. I can't remember what it was like to be new. I remember it sucked, but I can't remember like one example I will share with you. Uh the very first iPhone, so what was that? 2008 or so? There was a jailbreak came out for that and there was like a Cydia App Store. Anybody remember that? Yeah? Yeah? Okay, good, a few people. And one of the instructions in there was, "Just and the word just was doing some heavy lifting that day, just SSH into your iPhone." And I'm like, "What's SSH?" I mean, now I use it every day and it's like second nature to me. ([View Highlight](https://read.readwise.io/read/01m50s7f53qctv7jwbgfcq6et3)) - Note: [H2] Videos should exist to be seen - Tags: atom/literature/h2 - [00:11:34] What you got to remember is that half your audience are probably you 15 years ago thinking, "What's he on about?" And the hardest part of making a video is figuring out what your audience already knows, where the where the start line is. So, in a video about subnet routers for Tailscale, which is a very deep deeply technical topic that you probably understand subnets, IP addresses, firewall rules, the basics of networking already, I can skip all that and I can start with the actual subnet routing piece. But, in a video about what is Tailscale, well, I have to start at a much earlier primitive. And so, by doing that, I'm making the video useful to that specific audience segment that I'm trying to reach. [00:12:17] I will then try and do the Pixar thing. That's meaning, I'll try and bifurcate my content. I'll try and have something for the kids and something for the grownups. Whether that's just me talking about Vim key bindings alongside where I'm talking about basics of SSH, whatever. Like, it's something for everybody. I'm trying to keep it interesting at any any given moment. And it it can be really difficult to do um and you don't you won't always get it right. But, when you think about the purpose of a video, it can have a few a few uh prongs to it. And so, one of the things you can do a video about is something brand awareness. And you do this by putting your company logo next to another popular product or service or project that your audience is already ([View Highlight](https://read.readwise.io/read/01m50s7hhnr2x5w0a9b2neybkp)) - Note: [H3] Expert blindness hides the viewer's real start line - Tags: atom/literature/h3 - [00:15:54] actually don't perform very well. And that speaks to a wider thought process you might have of like, "Well, surely more views is better. More subscribers is better. More, more, more." And this growth mindset of content, like it it's just it's a zero-sum game. Like, you can never have enough views. You could always get one more view. You could always get one more subscriber. But, if you're useful to one CISO, and that entire company decides, "I'm going to adopt Tailscale from the top down," as opposed to what we typically do in DevRel of bottom-up, you've been useful not only to that company and that CISO, you've also been useful to your sales [00:16:35] team and the pipelines and all those other marketing uh machinery that operates around DevRel cuz we're just a cog in a wider machine, aren't we? And so, all of those things, all of these four things, they're all about discovery, understanding, allowing users to succeed, and ultimately grow in their understanding and use of your product. And this is the cycle of YouTube. Users discover you, they understand you, and they succeed with your product where before, perhaps, there was only frustration of like, "What is SSH?" Well, now they know, and they know that Tailscale does it really well. That's the idea, and the cycle continues. I've talked a lot about being useful. And the best way to do this is just make ([View Highlight](https://read.readwise.io/read/01m50s7kwwxdpq4fnp2qyyh4cq)) - Note: [H4] One useful video can land a whole company - Tags: atom/literature/h3 - [00:21:36] And then I'll come back, and I'll throw that transcript into chat, and I'll be like, "What did I just say?" >> [laughter] >> And it comes up with a a list of points, and then I'll sort of get I'll repeat the process maybe sometimes five or six times. It takes a long time sometimes for the the more boring ideas, the more businessy ideas. The self-hosting stuff, like I talk about it all day, but the the product updates, they can be really tough to tease out anything that's vaguely entertaining. And so, going Didn't mean that to be a joke. Um The point is is that essentially, once you've got to that point, you've got to not read what's ChatGPT has just vomited out to you. Because otherwise, you are just presenting the statistical mathematical [00:22:20] mean of the English language, and you become the same as everybody else. What people are tuning into is me. Rightly or for better for worse, right? You know, um and they want to hear me in all of my imperfect human glory. They want to hear my opinions, the jokes I didn't mean to be jokes, that kind of thing, right? You are your secret weapon. Now, this is a key thing. You've got to make starting easy. Remove friction at every point in the process. Do not, if you can avoid it at least, think that you're going to set your lights up every single time. You're going to set your camera up every single time. You're going to set a dedicated recording monitor. You're going to log into the specific accounts, and close Slack, and make sure your do not disturb is on, and ([View Highlight](https://read.readwise.io/read/01m50s7p8kx49k2xm8hvzevh64)) - Note: [H5] Removing friction makes regular publishing sustainable - Tags: atom/literature/h3