# Experiment Faster, Own Your Audience, Stay Irreplaceably Human **Speaker:** Steven Bartlett, interviewed by Patrick Bet-David; audience Q&A including Dan Martell **When:** 2026-09-01 · 6:26 PM–8:00 PM ET (3:26 PM Las Vegas) **Theme:** Audience & content **Sources:** et9rqthBK3WF8Sf3ca2Q **Notes page:** https://vault.chels.ai/sessions/d1-steven-bartlett/ > **In a world where the right answer changes every few months, the only durable edge is how fast your team runs small experiments, how obsessively it hunts tiny gains, and whether you own the relationship with your audience rather than renting it from an algorithm.** Steven Bartlett, host of The Diary of a CEO and founder of the creator media company Flight Story, sat down with Patrick Bet-David for roughly an hour, then took audience questions. He ran a large faceless social media company (Social Chain) through his twenties and says he simply ported those audience-growth principles into podcasting. The session was a tour of the operating principles behind the fastest-growing show in podcasting: rate of experimentation, 1% gains, paper walls, listening-led interviewing, packaging tests, audience sovereignty, recruiting, and self-disruption in the age of AI. For the year ahead it is a practical manual for anyone building an audience, a team, or a product while AI collapses production costs. **Best for:** podcasters and YouTube creators at any stage; founders scaling a team and a culture; marketers who depend on social reach; salespeople, recruiters and negotiators; anyone deciding how to use AI without outsourcing their thinking; parents and leaders worried about the next generation and AI ## Claude skills from this session - `audience-sovereignty-audit` — Run Steven Bartlett's sovereignty audit on an audience or customer base. Counts what the user owns (email addresses, postal addresses, paying customers with card details, event attendees) against what they rent (followers and subscribers on each platform), computes the owned ratio, designs an off-ramp from every social channel (newsletter, event, product, physical mail), redraws the funnel with social at the top and owned channels at the bottom, runs the January best-versus-worst variance analysis, and sets a monthly owned-audience target. Use whenever the user talks about followers, subscribers, reach dropping, algorithm changes, "getting zucked", growing a newsletter or email list, launching events or merchandise, direct mail, whether to depend on YouTube, Instagram, TikTok or Spotify, or how to measure their audience — even if they never say "sovereignty" or ask for an audit. https://vault.chels.ai/skills/audience-sovereignty-audit/SKILL.md - `experimentation-cadence` — Install Steven Bartlett's team operating cadence for finding the right answer faster than competitors. Sets up three shared channels (Experiments, 1%, Paper Walls), makes the count of tests run the only incentive, plans each team's first ten tests, runs a 1% audit of one customer touchpoint, hunts paper-wall constraints (with the six-weeks-versus-eight-weeks hiring question), and adds a weekly memo the leader writes in their own words. Use whenever the user wants their team to innovate faster, run more A/B tests, build a testing or experimentation culture, stop playing safe, find marginal gains or small improvements, question "that's how it's always been done" or "it takes eight weeks", screen hires for people who push back on constraints, or asks how to report experiments to the company — even if they never say "experimentation cadence", "1%", or "paper walls". https://vault.chels.ai/skills/experimentation-cadence/SKILL.md - `listening-led-interview` — Prepare and run a conversation the other person leads, using Steven Bartlett's follow-don't-lead interviewing method and the RICE motivation sheet (Reward, Ideology, Coercion, Ego) he learned from former CIA officers. Produces a curiosity brief with zero written questions, one opening question, a listening plan, a filled-in RICE sheet, and an offer or next step tailored to what actually motivates the person. Use whenever the user is preparing a podcast or media interview, a hiring interview, a sales or discovery call, a negotiation, a one-to-one, a difficult conversation with a partner or cofounder, asks "what questions should I ask", wants to understand what motivates someone, or is drafting a job offer or proposal — even if they never say "listening" or "RICE". https://vault.chels.ai/skills/listening-led-interview/SKILL.md - `packaging-test` — Run Steven Bartlett's $100 packaging test before anything is published. Writes ten title or headline variants and three thumbnail or cover variants, runs them as a small fixed-budget paid test to the real target audience, picks the winner by click-through rate rather than taste, then rewrites the first 10 to 60 seconds (or first paragraph) so it delivers the winning promise immediately, and logs the spread between best and worst. Use whenever the user is choosing a title, headline, thumbnail, hook, intro, trailer, book title or cover, product name, packaging copy, landing-page headline or ad creative, asks "which one should I go with?", wants better click-through or average view duration, or is about to launch anything — even if they never mention testing or ads. https://vault.chels.ai/skills/packaging-test/SKILL.md ## Skills to build - **Running a high-rate experimentation cadence** — Treating every team's job as running small scientific tests, reporting the result to the whole company, and judging people on how many clear tests they run rather than whether each one works. _How to practise:_ Each week, every team writes at least one hypothesis, holds all variables but one, names the metric, runs the test, and posts the result in one shared channel whether it failed or succeeded. Track only the count of tests per week. You know it is working when the channel fills up with failures nobody is embarrassed by and the count keeps rising month over month. - **Hunting 1% gains your competitors will ignore** — A daily habit of finding and stacking tiny improvements in areas nobody else bothers with, based on Jeff Olson's book The Slight Edge and Sir Dave Brailsford's marginal-gains work with British Cycling. _How to practise:_ Open a dedicated channel called 1% and ask everyone to post one small improvement a day, however petty (studio scent, water bottle size, the music playing when a guest arrives). Celebrate each post publicly. It is working when the team starts talking about momentum and leaves the office energised rather than depleted. - **Pushing on paper walls** — Asking why whenever you meet a constraint that looks real, because most constraints turn out to be preference and tradition rather than steel. _How to practise:_ Whenever someone says 'it takes eight weeks' or 'that's how it's done', ask why, then ask why again of the person who set the constraint. Log every wall you pushed and what happened in a shared channel. Bartlett's animator went from nine days to two days once someone asked what it would take (a new laptop). It is working when your team brings you supplier pushback stories instead of delay requests. - **Listening-led interviewing (follow, don't lead)** — Preparing by getting deeply curious about a person rather than writing questions, then letting them take the conversation where they subconsciously want to go. _How to practise:_ In your next three interviews, sales calls or one-to-ones, write zero questions in advance. Research the person until you are genuinely curious, ask one open question, then follow what they give you. Time how long you go without interrupting. Bartlett has never written a question down in the history of his show; his most impactful episode came from abandoning his plan within ten seconds. - **Reading motivation with the RICE framework** — A spy-school model (Reward, Ideology, Coercion, Ego) for understanding what actually drives a person before you make them an offer, used by Bartlett in hiring, sales and negotiation. _How to practise:_ In any high-stakes conversation, let the other person offload first. Listen for their ideology (what they care about, the chip on their shoulder, who overlooked them) and their ego before you mention money. Only then tailor your offer. It is working when candidates accept offers that were not the highest bid. - **Testing packaging before you publish** — Running cheap paid-ad tests on titles, thumbnails, intros, book covers and product copy so creative decisions are measured rather than guessed. _How to practise:_ Before any launch, write five to ten title or thumbnail variants and run a small paid test (Bartlett cites a $100 test) to real audience members. Ship the winner and make sure the first 30-60 seconds deliver on it. Bartlett runs hundreds of variants per episode; his early test on one episode showed a 20x gap between the worst and best packaging of the same conversation. - **Building audience sovereignty** — Moving your audience from rented platforms (YouTube, Instagram, TikTok) into channels you own: email, physical addresses, live events and paid transactions. _How to practise:_ Treat social media as the top of your funnel and owned environments as the bottom. Every piece of content should have an off-ramp to a newsletter, an event, a product or a physical mailing. Measure how many email addresses, postal addresses and credit-card relationships you hold, not how many followers. It is working when your revenue no longer moves with algorithm changes. - **Recruiting as your highest-leverage weekly work** — Treating recruitment, not effort or ideas, as the activity most correlated with your outcomes, and moving fast when a strong candidate appears. _How to practise:_ Block real hours for recruiting every week (Bartlett says he spends about 30). Run a written culture test that screens for agency, ambition, hard work, optimism, leaning into change, and willingness to have uncomfortable conversations. Call strong candidates within hours, not weeks. It is working when your best hires say the speed of your response is why they joined. - **Thinking in principles and long time horizons** — Anchoring decisions in first principles that survive change, and setting 10-, 20- and 50-year horizons so you build foundations instead of rushing. _How to practise:_ For each major decision, write the principle behind it in one sentence and ask whether it would still hold if the tools changed completely. Then ask what you would do differently if you had to keep doing this for 50 years. Bartlett only takes guests he personally wants to talk to because he has to love his calendar for decades. - **Writing memos to keep your own thinking sharp** — Deliberately doing the hard cognitive work of writing, rather than deferring thinking to AI, so you keep the deep understanding needed to ask AI the right questions. _How to practise:_ Write at least one memo or essay to your team each week, in your own words, on a problem you are wrestling with. Bartlett says he writes more memos weekly than anyone in the room and that hardship today neurologically sets you up to solve hard problems tomorrow. It is working when your prompts to AI get sharper because you understood the problem first. ## What matters - **The right answer now expires in months, so the skill is finding it faster:** Bartlett cites Ray Kurzweil: the rate of change is accelerating, so any tactic from a book or a talk lasts three to six months. The durable skill is not knowing the answer but running enough experiments to find the next one before competitors do. _(Evidence: 18:34:58-18:36:33 / et9rqthBK3WF8Sf3ca2Q)_ - **Incentivize the input (tests run), never the output (results):** If people are rewarded for experiments that succeed, they stop proposing uncertain ideas because they fear being wrong. Bartlett stays out of the success-or-failure business and only pushes the number of tests his team runs. He has a head of failure and experimentation whose only job is raising that rate. _(Evidence: 18:36:57-18:38:50 / et9rqthBK3WF8Sf3ca2Q)_ - **Asking better questions is a commodity; the system underneath is the moat:** Bartlett says the show's growth (roughly a million subscribers in one recent 30-day period, six to seven times faster than any other show) is not because he asks better questions. It is the rate of trying things across every team, plus tools like a guest-selection database that predicts how a guest will perform. _(Evidence: 18:37:37-18:38:19, 18:46:34-18:47:06 / et9rqthBK3WF8Sf3ca2Q)_ - **Easy to do means easy not to do, and both compound:** From The Slight Edge: brushing your teeth, saving a dollar, improving a thumbnail are all trivial today and invisible for months, then one day you are in the dentist's chair or on an exponential growth curve. Every area of your life and business is already on one of those curves. _(Evidence: 18:39:58-18:42:00 / et9rqthBK3WF8Sf3ca2Q)_ - **Flat for three years, then explosive, is what compounding looks like from the inside:** The Diary of a CEO was completely flat for three years before it exploded. Bartlett warns that you have to keep doing the small things on something like faith, because you will do them tomorrow and nothing visible will happen. _(Evidence: 18:42:00, 19:26:24 / et9rqthBK3WF8Sf3ca2Q)_ - **Tiny wins create the feeling of progress that makes teams stay late willingly:** Sir Dave Brailsford told Bartlett his cyclists went from leaving at 5pm to leaving at 1am once they became obsessed with petty details, because they felt they were going somewhere. Momentum, not pressure, is what makes a team compound. _(Evidence: 18:44:17-18:44:53 / et9rqthBK3WF8Sf3ca2Q)_ - **Be romantic about the impact, unromantic about the method:** Bartlett deleted three of his favourite chapters after eye-tracking tests, replaced human editors when an AI editor produced better watch time, and says if data shows an AI does his job better he will move out of the way. Attachment to the way you deliver value is a tax on your potential. _(Evidence: 18:56:23-18:57:30, 19:20:48-19:21:14 / et9rqthBK3WF8Sf3ca2Q)_ - **Timeframes determine behaviour:** Give someone sixty seconds and unlimited bricks and they stack vertically until the tower falls; give them a year and they build a foundation. A friend who decided to stop being single within a year shifted his goal from the right partner to the best partner available, and will waste time trying to save time. _(Evidence: 18:48:27-18:49:57 / et9rqthBK3WF8Sf3ca2Q)_ - **Distribution is the moat when AI makes making things cheap:** Creators now have the reach of CNN or Fox without the commercial, data or back-office infrastructure. As AI makes software and content trivial to produce, owning distribution and infrastructure is what matters most. That is why Bartlett's company owns a podcast hosting platform used by other major shows. _(Evidence: 18:47:37-18:51:19 / et9rqthBK3WF8Sf3ca2Q)_ - **Follower counts are becoming irrelevant because AI now matchmakes content:** Platforms run every post through an AI to understand its context and route it to whoever wants it. The platform's incentive is retention and ad sales, not honouring who chose to follow you. TikTok already ignores follower counts, and the variance between a show's best and worst episode grows every year. _(Evidence: 18:51:19-18:52:15, 18:58:10-18:58:46 / et9rqthBK3WF8Sf3ca2Q)_ - **Sovereignty is the word of the next decade:** Do you own the relationship with your audience, or does the platform? The biggest podcasts are down 60 to 80 percent and wish they had measured email addresses, physical addresses and card details instead of followers. Social media should be the top of the funnel and owned channels the bottom. _(Evidence: 18:52:15-18:52:49, 18:58:46 / et9rqthBK3WF8Sf3ca2Q)_ - **Infinite AI content plus fixed attention means every view gets cheaper:** Someone anywhere can now generate content while they sleep using agents, but total attention on social media is fixed. That supply shock lowers the view potential of every unit of content, so the only defensible content is what AI cannot make: irreplaceably human material. _(Evidence: 18:52:49-18:53:27 / et9rqthBK3WF8Sf3ca2Q)_ - **Information is being commoditised; lived experience is not:** You can get a fact about the gut microbiome from an AI. What you will still want from a human is relatability and lived experience, the way Formula One would lose its meaning without drivers in the cars. Wrap tactics in the human layer: the struggle, the doubt, the family who thought it was a bad idea. _(Evidence: 18:53:37-18:55:52 / et9rqthBK3WF8Sf3ca2Q)_ - **Human documentaries may outgrow the interview format:** Bartlett has built an in-house team making human documentaries and says early signals suggest they will be bigger than his podcast. MrBeast's first vlog doing 20 million views in a week was, to him, the signal that the whole market is moving this way. _(Evidence: 18:54:32-18:55:11 / et9rqthBK3WF8Sf3ca2Q)_ - **In-person connection gets more valuable as everything else digitises:** People do not come to conferences for information they could find online; they come for connection, which is a basic human need. When a trend rises, its opposite rises too: processed food brought organic, streaming brought back vinyl, screens are bringing back silent retreats and physical mail. _(Evidence: 18:57:30-18:58:10, 19:00:27-19:01:13 / et9rqthBK3WF8Sf3ca2Q)_ - **Physical post is a sovereign channel that is not going anywhere:** Email survived because it is a rail everyone already uses for work. Physical addresses are the same: you will always need somewhere for parcels to arrive. Direct-to-consumer brands in Bartlett's portfolio see some of their best returns from founder postcards with a discount code, and he is testing paid physical packages to his audience. _(Evidence: 18:59:40-19:01:38 / et9rqthBK3WF8Sf3ca2Q)_ - **Money is a lagging indicator; pessimism can be a leading one:** Twice, smart people proved to Bartlett with perfect math that his Facebook pages and then his podcast could never make money. He ignored both because money lags value. If nobody is sceptical about your idea, you may already be late. _(Evidence: 19:02:17-19:03:56 / et9rqthBK3WF8Sf3ca2Q)_ - **Value per minute is the signal to watch when starting anything:** Asked which metrics matter now, Bartlett named value per minute above all else. If your content or product delivers unusually high value for the time it asks, distribution and money follow later. _(Evidence: 19:02:17 / et9rqthBK3WF8Sf3ca2Q)_ - **Build for one honest friend who loves it:** One of Bartlett's best friends told him bluntly when episodes were bad, then asked when the podcast was coming back after a month off because he missed it on his commute. If one honest person genuinely loves what you make, there are many more like them. _(Evidence: 19:04:06-19:04:36 / et9rqthBK3WF8Sf3ca2Q)_ - **Failing experiments can still be right if the underlying technology is improving:** Translating the show into other languages failed for about 18 months, including a partnership with Spotify, and finance said the ad rates would never justify it. Bartlett kept going because the translation tech would eventually get good enough. It is now a seven-figure revenue line and the fastest-growing thing in the show's history. _(Evidence: 19:04:36-19:05:26 / et9rqthBK3WF8Sf3ca2Q)_ - **You are trapped inside constraints you have never tested:** Zara's founder asked why clothes took nine months to make and built factories in Spain; Roger Bannister broke the four-minute mile and others followed within days; suitcases had no wheels until someone asked why. People do not know which of their own constraints are paper. _(Evidence: 19:05:37-19:11:04 / et9rqthBK3WF8Sf3ca2Q)_ - **Only book guests, or take projects, you would still want in year 50:** Bartlett's guest filter has three overlapping circles: will the audience like it, will the algorithm like it, and do I want this conversation. The third is decisive, because sustaining anything for decades requires loving your calendar. He turns down the biggest names when he is not interested. _(Evidence: 19:13:21-19:13:50 / et9rqthBK3WF8Sf3ca2Q)_ - **Ideology motivates people more than money:** Former CIA officers told Bartlett that of Reward, Ideology, Coercion and Ego, ideology is the strongest lever and ego second. A spy listens for months to find what someone cares about, then makes an offer that speaks to it. Bartlett runs two-hour hiring interviews mostly listening for the same reason. _(Evidence: 19:16:11-19:18:53 / et9rqthBK3WF8Sf3ca2Q)_ - **Deliver the packaging promise in the first minute:** Platforms optimise for average view duration. If the title promises how to make tea, the first 10 to 30 seconds must start making tea or people bounce. MrBeast skips the greeting and restates the promise immediately. Bartlett packages first, then cuts the trailer to match. _(Evidence: 19:24:01-19:25:18 / et9rqthBK3WF8Sf3ca2Q)_ - **Not knowing the industry rules is an advantage:** Bartlett says the show succeeded because his team knew nothing about podcasting: they did not know shows did not make trailers or test titles hundreds of times. No preconceptions meant no status quo to defend, and that is where innovation comes from. _(Evidence: 19:47:23-19:48:04 / et9rqthBK3WF8Sf3ca2Q)_ - **Success breeds loss aversion, and your customers will keep you trapped:** Losing $10 hurts so much that you need to find $20 to feel even, so successful people go into defence. The Innovator's Dilemma shows that giving customers exactly what they ask for (faster horses) keeps you out of the next S-curve. Winning again requires self-disruption, which is unnatural. _(Evidence: 19:44:58-19:46:46, 19:48:07-19:50:04 / et9rqthBK3WF8Sf3ca2Q)_ - **Champions rebuild their technique at the top, and so should you:** Shot-putter Ryan Crouser threw out a three-gold-medal technique and rebuilt it from physics; Tiger Woods spent two years losing while rebuilding his swing. Bartlett wrote his team an essay on this as the show approached number one, because reinvention is impossible if you are romantic about what got you here. _(Evidence: 19:43:27-19:44:58 / et9rqthBK3WF8Sf3ca2Q)_ - **AI will shrink research teams but grow sales teams:** After an internal AI-adoption contest, Bartlett found his research team no longer needs to scale because researchers use a flock of agents. Sales, by contrast, remains a relationship business, so he is hiring more salespeople and buying a US sales team. Everyone in the company must now be AI proficient. _(Evidence: 19:28:14-19:31:24 / et9rqthBK3WF8Sf3ca2Q)_ - **Doing hard things you resist physically grows the brain region for hard problems:** Citing neuroscientist Andrew Huberman, Bartlett says the anterior midcingulate cortex grows when you do things you actively do not want to do, and is larger in athletes and older generations than in people raised on swiping. Deferring your thinking to AI shrinks the capacity you need to ask AI the right question. _(Evidence: 19:55:38-19:58:21 / et9rqthBK3WF8Sf3ca2Q)_ ## Frameworks, lists & numbers - **Rate of experimentation (the scientific method at company scale)** — Hold all variables, change one, state a hypothesis, decide what you are measuring, run it, and report the result to the entire company in a central Slack channel whether it fails or succeeds. The input (how many tests the team runs) is controllable; the output is not, so only the input is incentivised. Bartlett's team runs roughly 10 to 15 experiments a day and has a head of failure and experimentation. Inspired by Jeff Bezos's shareholder letters (Amazon will win if it is the best place on earth to fail; one in ten bets pays for the rest) and Booking.com's experimentation platform. Company motto: kill the guesswork. - **Kurzweil's accelerating change** — If you are 10 today, by age 60 you will experience a year's worth of change roughly every 10 days. If you are 40 today, by 60 you will experience a year's change roughly every 90 days. Conclusion: the correct answer to any business question will change faster than books (about a year to publish) or conference talks (useful for three to six months) can deliver it. - **The 50% reach decay** — Running a large social media company through his twenties, Bartlett and every competitor lost roughly 50% of organic reach per year as platforms added ads to the feed. He calls this getting zucked. Investing in an asset losing half its return each year is unstable, which is why he moved to podcasting and now to owned channels. - **The Slight Edge / 1% principle** — Things that are easy to do are also easy not to do, so people skip them. Every area of life follows the same curve: slow and invisible, then fast, in either direction. Don't brush your teeth for a day, a week, a month, and nothing hurts; do it for five years and the dentist is pulling molars. The Diary of a CEO was flat for three years and then exploded. The company runs a Slack channel called 1% for tiny improvements (studio scent, air pressure, guest music) and celebrates each one. - **Marginal gains and the progress principle (Sir Dave Brailsford)** — Brailsford banned the British cycling team from thinking about the podium and obsessed them with 1% gains: how much water fits in the bottles, how soft the hotel pillows are. Some marginal gains turn out not to be marginal, and they are plentiful because competitors overlook them. The team went from leaving at 5pm to leaving at 1am because stacking tiny wins made them feel they were going somewhere. This is the first chapter of James Clear's Atomic Habits. - **Testing volumes at The Diary of a CEO** — About 350 Facebook ad A/B tests per guest interviewed; one guest's name tested 50 different ways to learn how the audience wanted him introduced; 400 tests on the book title and cover before writing began (14.8% clicked the winning title, roughly 'just do it'); 3,000 audience members read randomised chapters with eye tracking and rated each chapter life-changing or cliche, after which three of his favourite chapters were deleted; an early $100 packaging test on one episode showed the worst variant delivering about 1% and the best about 20%, a 2000% difference from the same conversation. - **Growth figures cited** — Two months before the talk the show added about a million subscribers in 30 days; the current month about 600,000 to 700,000, which he says is six to seven times faster than any other show. His growth graph projects around 50 million subscribers within two years. A signed creator's show is growing faster than the flagship did at the same stage. His media company has about 150 staff and 30 to 40 creators signed; the company reportedly raised at around a $450 million valuation with Bartlett holding about 90% (stated by Bet-David, not corrected). - **Three-circle guest filter** — Three overlapping circles: (1) will my audience like it, (2) will the algorithm like it, (3) do I want to have this conversation. The third is the long-termism circle and overrides the others, because doing anything for 50 years requires loving your calendar. - **RICE: Reward, Ideology, Coercion, Ego (CIA motivation framework)** — Taught to Bartlett by former CIA officer Andrew Bustamante. Ranked by power: Ideology first (what a person believes and cares about), Ego second (everyone has one, even Mother Teresa), Reward third (money and the things people want), Coercion fourth (pressure). Everyone has three lives: public, private and secret; a spy can reach any secret within nine months by listening. Application: shut up and let the person hand you their ideology, then tailor the offer. Bartlett uses it in two-hour hiring interviews, sales and negotiation. - **Paper walls** — Ask why whenever a constraint looks real. Zara's founder asked why clothes took nine months, built factories in Spain and got designs on rails within ten days; Roger Bannister broke the four-minute mile and 11 others followed within days; suitcases had no wheels until someone asked why. Interview question: your event is in six weeks, the supplier says eight weeks, what do you do (hire the person who asks the supplier why). Bartlett tells his team that a couple of such breakthroughs over five years mean a team 50% smaller, learning 170% faster and doing things almost 200% faster. His animator went from nine days to two after asking what it would take: a $2,000 laptop. Slack channel: pushing on paper walls. - **Sovereignty scorecard** — Measure your audience by how many email addresses, physical addresses and credit-card details you hold, not followers. Social media is the top of the funnel; owned environments (newsletter, events, transactions, physical post) are the bottom. Some of the biggest podcasts are down 60% to 80% from algorithm changes. Every January, compare your best and worst episode; the widening gap is the AI sorting hat getting more aggressive. - **Average view duration rule** — Platforms rank on AVD (average view duration). The promise made by the thumbnail and title must be delivered within the first 10 to 30 seconds and certainly within 60, then sustained. Package first, then cut the trailer to match the packaging that tested best. MrBeast skips greetings and restates the promise immediately. A 120-minute interview might ship at 120 or around 100 minutes, but Bartlett would not cut it to 60; sections can be moved (a 75th-minute story may open the show) if it fits the story. - **Culture test traits** — Bartlett's team and portfolio founders complete a written culture test (culturetest.com) that screens for agency, ambition, hard work, a capable mentality, a bias towards optimism, leaning in when the world changes, pushing people, and having uncomfortable conversations when it would be socially expedient to be nice. Strong candidates get a call within hours. He spends about 30 hours a week on recruitment. - **What makes a great creator** — Obsession; not doing it for money, because money is a lagging indicator that may take years; humility, because every creator falls at some point and the ones who vanish got romantic about themselves; willingness to reinvent. - **Loss aversion and the Innovator's Dilemma** — Daniel Kahneman: losing $10 is not offset by finding $10; you need to find $20. Successful people therefore go into defence. Clayton Christensen's Innovator's Dilemma: incumbents do exactly what they should (listen to customers) and get trapped sustaining success while the next S-curve passes them. Horse-and-carriage test for a new idea: Is it faster than what we sell? (No, innovation is always worse at first.) Are customers asking for it? (No, they want faster horses.) Is our supply chain set up for it? (No, we have stables.) If you would throw the idea out, that is the danger sign. - **AI at Bartlett's companies** — As reported in a question from Dan Martell: a 60-day AI adoption contest with a $20,000 prize saved about $1.2 million and 60,000 hours. Bartlett works with about 16 agents daily, including one that reads his YouTube back-end data every day and proposes five guests each Monday, and one that triages WhatsApp, iMessage, two inboxes, Slack and a Monday board as a chief of staff. An AI editor beat human editors on average view duration after four months of testing. His roughly $900 million investment fund has had one employee for three years, supported by two agents that source and screen deals. AI proficiency is now almost non-negotiable for applicants. - **Reading list for experimentation cultures** — The Slight Edge (Jeff Olson) on small things compounding; Blue Ocean Strategy on escaping red (crowded) oceans for blue (uncharted) ones as AI collapses moats; Working Backwards (two Amazon insiders) on starting from the customer's pain point, including the empty chair for the customer in every meeting; The Innovator's Dilemma on why successful companies miss the next wave; and his own book. - **Anterior midcingulate cortex** — Per Andrew Huberman, this brain region grows when you do things you actively resist and do not want to do. It is largest in athletes, larger in older generations than in their swiping children, and smaller in people who are sedentary. Hardship today neurologically prepares you for hard problems tomorrow. Bartlett's response is to write memos weekly and to resist deferring thinking to AI, because knowing the right question to ask a model requires deep understanding of the problem. ## Exercises & frameworks to run - **Set up an experiments channel and your first ten tests** _(90 minutes to set up, then ongoing)_ _[speaker (steps written by us)]_ 1. Create one shared channel (Slack, Teams, a group chat) called Experiments that everyone in the company can see. 2. Post a one-line template: Hypothesis / Variable changed / Metric / Result / What we learned. 3. Ask every team to list ten small tests they could run this month. Push for small: a subject line, a call opening, a thumbnail colour, a checkout step. 4. Run them one variable at a time. Post every result, failed or not, within a day of finishing. 5. Count tests per week and make that count visible. Do not track win rate. - **Finish with:** A visible feed of company-wide experiments and a weekly test count you can push upward - **The 1% audit** _(45 minutes)_ _[speaker (steps written by us)]_ 1. Pick one customer or audience touchpoint (a podcast studio, a shop floor, a sales call, an onboarding email). 2. Walk through it as the customer and list twenty details so small you would normally ignore them: smell, sound, temperature, first sentence, load time, the music playing. 3. Mark which ones a competitor would never bother with. Those are your targets. 4. Improve three this week and post each one in a channel called 1%. 5. Repeat weekly with a different touchpoint. - **Finish with:** A running list of tiny improvements and a team habit of noticing them - **Compounding curve check** _(30 minutes)_ _[editorial (speaker idea, steps written by us)]_ 1. List six areas of your life and business: health, money, key relationships, skills, audience, product. 2. For each, write down the small daily behaviour that is either compounding for you or against you right now. 3. Ask: if I keep doing exactly this for five years, where does the curve end up? 4. Circle the one area compounding against you and choose the smallest daily fix. 5. Revisit the list once a quarter. - **Finish with:** A one-page map of what is silently compounding for and against you - **Paper wall hunt** _(1 hour, then ongoing)_ _[speaker (steps written by us)]_ 1. List every constraint your team currently treats as fixed: lead times, prices, formats, 'the way clients want it', how long a task takes. 2. For each, write who set it and when. If nobody knows, it is probably paper. 3. Pick three and ask the source directly: why does it take that long / cost that much? What would it take to halve it? 4. Record what you learn in a channel called Paper Walls, including the small ones. 5. Add the question 'what would it take to do this in a fraction of the time?' to your weekly planning. - **Finish with:** A list of constraints tested, and at least one that turned out to be preference or tradition - **Paper wall hiring question** _(5 minutes per interview)_ _[speaker]_ 1. Add this scenario to every interview: 'You have an event in six weeks. The supplier says it will take eight weeks to deliver. What do you do?' 2. Listen for the three answer types: scale back the event, push the event two weeks, or ask the supplier why it takes eight weeks. 3. Only the third type pushes back on constraints. Weight it heavily in your decision. 4. Follow up with: 'Tell me about a constraint you were given at work that turned out not to be real.' - **Finish with:** A repeatable interview question that screens for people who challenge constraints - **Sovereignty audit** _(1 hour)_ _[speaker (steps written by us)]_ 1. Write down your follower or subscriber count on each platform. Then write down how many email addresses, physical addresses and paying customers you own directly. 2. Calculate the ratio. Most creators are shocked by how small the owned number is. 3. For each platform, name the off-ramp you currently offer: newsletter, event, product, physical mailing. If there is none, design one this week. 4. Redraw your funnel with social at the top and owned channels at the bottom. 5. Set a monthly target for owned-relationship growth and review it alongside views. - **Finish with:** A funnel diagram with owned channels at the bottom and a monthly owned-audience target - **Annual variance analysis** _(1 hour once a year)_ _[speaker (steps written by us)]_ 1. Every January, pull the view counts for every episode or post you published in the past year. 2. Compare your best-performing piece with your worst. Calculate the gap. 3. Compare that gap to the previous year's. A widening gap means the algorithm is sorting more aggressively and your follower base protects you less. 4. Use the result to decide how hard to push sovereignty and packaging tests for the year. - **Finish with:** A one-number read on how algorithm-dependent your reach is, tracked year over year - **The $100 packaging test** _(2 hours plus a 24-48 hour test)_ _[speaker (steps written by us)]_ 1. Before publishing a piece of content, a book, or a product page, write ten title or headline variants and three thumbnail or cover variants. 2. Run them as a small paid ad test to a sample of your target audience with a fixed small budget. 3. Pick the winner by click-through rate, not by which one you like. 4. Rewrite the first 30 to 60 seconds (or first paragraph) so it delivers on the winning promise immediately. 5. Log the spread between best and worst to remind yourself how much packaging matters. - **Finish with:** A tested title and opening, plus a record of the performance spread - **Three-circle guest or project filter** _(20 minutes)_ _[speaker (steps written by us)]_ 1. Draw three overlapping circles: my audience will like it, the algorithm (or market) will like it, I want to do this. 2. Place every pending guest, client, project or partnership on the diagram. 3. Anything outside the third circle gets a no, however impressive, because you cannot sustain it for decades. 4. Prioritise the centre where all three overlap. - **Finish with:** A shortlist you will still be glad you said yes to in years to come - **The no-questions interview** _(One conversation)_ _[speaker (steps written by us)]_ 1. Before your next interview, sales call or one-to-one, research the person until you are genuinely curious, but write no questions down. 2. Open with one broad question and then say nothing until they finish completely. 3. Follow whatever they lean into, even if it abandons your plan. 4. Afterwards, note what they revealed that you would never have asked about. - **Finish with:** A conversation the other person led, and notes on what listening uncovered - **RICE listening sheet for hiring and sales** _(Prep 10 minutes, one meeting)_ _[speaker (steps written by us)]_ 1. Before a key interview or negotiation, draw four boxes: Reward, Ideology, Coercion, Ego. 2. Let the other person talk for most of the meeting. Bartlett spends up to two hours letting a candidate offload. 3. Fill in what you hear: what they want (Reward), what they believe and the chip on their shoulder (Ideology), what pressures they face (Coercion), what they want to be seen as (Ego). 4. Draft your offer to speak first to Ideology, then Ego, before you mention money. 5. Never use Coercion; note it only to understand their situation. - **Finish with:** An offer tailored to what actually motivates the person - **Build your culture test** _(Half a day to build)_ _[speaker (steps written by us)]_ 1. Write ten to fifteen scenario questions that reveal agency, ambition, hard work, a bias toward optimism, leaning in when the world changes, and willingness to have an uncomfortable conversation when it would be easier to be nice. 2. Put it in a simple web form and send it to every applicant before a call. 3. Score answers the same day and call the top scorers within hours. 4. Track whether high scorers become your best performers and adjust the questions each quarter. - **Finish with:** A pre-screen that finds high-agency people and a fast response habit - **Faster horses check (self-disruption review)** _(1 hour)_ _[editorial (speaker idea, steps written by us)]_ 1. List the new ideas your team has raised in the last year that you rejected. 2. For each, ask the three horse-and-carriage questions: Is it worse than what we do today? Are customers asking for it? Are we set up to make it? If the answers are yes, no, no, it may be exactly the innovation that kills you. 3. Pick one such idea and give it a small budget and a separate owner, away from the core business. 4. Repeat every quarter. - **Finish with:** One deliberately protected experiment that threatens your own core product - **Working backwards with the empty chair** _(1 hour)_ _[speaker (steps written by us)]_ 1. At your next product or content planning meeting, put an empty chair at the table to represent the customer. 2. Write a one-paragraph persona of that customer and their three most painful problems. 3. Only then list product or content ideas, and only those that solve a named problem. 4. Reject anything that started from your own bias rather than their pain. - **Finish with:** A product or content plan that traces every item back to a customer problem - **Weekly memo** _(45 minutes a week)_ _[speaker (steps written by us)]_ 1. Choose one problem you are wrestling with this week. 2. Write a one-page memo to your team in your own words, without AI, explaining the problem, what you believe, and what you propose. 3. Only after finishing, use AI to critique it or find gaps. 4. Send it. Keep the memos in one folder and re-read a month later to see how your thinking has changed. - **Finish with:** A weekly written record of your own thinking and sharper questions for AI ## What to do — and how often - [ ] **Create a company-wide experiments channel and post every test result there, failed or successful** — Report back after every experiment; Bartlett's team runs 10 to 15 a day _[daily; speaker-stated]_ - [ ] **Appoint someone whose sole job is to raise the organisation's rate of trying things, not to be right** — Once, this quarter _[once; editorial recommendation]_ - [ ] **Post one small improvement in a 1% channel and applaud it publicly** — Every day, per team member _[daily; editorial recommendation]_ - [ ] **Post a paper wall you pushed on, however small, in a shared Paper Walls channel** — Every day someone posts, in Bartlett's words _[daily; speaker-stated]_ - [ ] **Run the best-versus-worst episode variance analysis on your own channel** — Every single year in January _[annually; speaker-stated]_ - [ ] **Add a newsletter sign-up, an event or a paid product as an off-ramp from every social channel you use** — Once, this month _[once; editorial recommendation]_ - [ ] **Report owned-audience numbers (emails, postal addresses, paying customers) alongside views in your metrics review** — Every month _[monthly; editorial recommendation]_ - [ ] **Test physical mail: send a founder postcard with a discount code to a segment of your customers and measure return** — Once, this quarter _[once; editorial recommendation]_ - [ ] **Run a small paid test on titles and thumbnails before publishing each major piece of content** — Before every major release _[weekly; editorial recommendation]_ - [ ] **Rewrite the first 60 seconds of your next video or episode so it delivers the packaging promise immediately** — Every episode _[weekly; editorial recommendation]_ - [ ] **Block recruiting hours in your calendar and call strong applicants within hours of their application** — Weekly; Bartlett spends about 30 hours a week on recruitment _[weekly; speaker-stated]_ - [ ] **Add the six-weeks-event, eight-weeks-supplier paper wall question to every interview** — Every interview from now on _[weekly; editorial recommendation]_ - [ ] **Make AI proficiency a requirement in every job description and audit which roles no longer need to scale** — Once, this quarter, then review annually _[once; editorial recommendation]_ - [ ] **Write a memo to your team in your own words on one hard problem** — Weekly; Bartlett writes memos on a weekly basis _[weekly; speaker-stated]_ - [ ] **Read The Slight Edge, Blue Ocean Strategy and Working Backwards, one per quarter, and brief your team on each** — One book per quarter _[quarterly; editorial recommendation]_ - [ ] **Review which past ideas you rejected because customers were not asking for them, and fund one as a self-disruption bet** — Every quarter _[quarterly; editorial recommendation]_ ## Questions to ask yourself - If the correct answer in my industry changes every six months, how many experiments did my team actually run last month? - Am I rewarding people for being right, or for trying? - What is the smallest thing in my business that all my competitors ignore, and have I fixed it? - Which area of my life is compounding against me right now in ways I cannot yet feel? - Which constraints in my business are steel, and which are just preference and tradition nobody has pushed on? - If I had to do this for 50 years, would I still say yes to what is on my calendar this week? - If every platform cut my reach by 80% tomorrow, how many people could I still reach directly? - What in my work is irreplaceably human, and what am I doing that an AI will soon do better? - Am I set up to make faster horses, and what idea did I reject because customers were not asking for it? - What is the last thing I wrote or thought through myself, without deferring it to an AI? - In my next important conversation, can I listen long enough to learn the other person's ideology before I make an offer? ## Quotes > “The fishing rod, which will allow you to find the correct answer yourselves, faster than any of your competitors, is quite simply your rate of experimentation and failure.” — Steven Bartlett (18:36:17 · et9rqthBK3WF8Sf3ca2Q) > > Why he refuses to hand out tactics that will expire in months. > “If you incentivize the outcome, you disincentivize the behavior because people will get scared. They'll get scared of being wrong.” — Steven Bartlett (18:38:50 · et9rqthBK3WF8Sf3ca2Q) > > Why he rewards the number of experiments, not their results. > “So I stay out of the business of success and failure. I'm purely focused on increasing the rate of experimentation.” — Steven Bartlett (18:38:50 · et9rqthBK3WF8Sf3ca2Q) > > His one controllable input as a leader. > “In every facet of your life right now, it is compounding for or against you based on how important you think those small things are.” — Steven Bartlett (18:40:11 · et9rqthBK3WF8Sf3ca2Q) > > Summarising Jeff Olson's The Slight Edge. > “We obsess over the smallest possible things that we think all of our competitors will ignore.” — Steven Bartlett (18:42:00 · et9rqthBK3WF8Sf3ca2Q) > > The 1% principle behind the show's growth after three flat years. > “I don't think that I'm a special thing, and I try and build a system that will kill the guesswork.” — Steven Bartlett (18:46:58 · et9rqthBK3WF8Sf3ca2Q) > > On proving his blueprint works for other creators. > “This I think is the word of the next decade for anyone who's got an audience on the internet is, do you own the relationship with your audience?” — Steven Bartlett (18:52:22 · et9rqthBK3WF8Sf3ca2Q) > > Defining sovereignty. > “In the world we're heading towards, it should be seen as the top of the funnel, and your owned environments have to be the bottom of the funnel.” — Steven Bartlett (18:52:37 · et9rqthBK3WF8Sf3ca2Q) > > Flipping the role of social media in the funnel. > “I'm so romantic about the impact I have, and I have zero romance about the way in which I deliver that impact.” — Steven Bartlett (18:56:52 · et9rqthBK3WF8Sf3ca2Q) > > On being willing to step aside if AI does his job better. > “If there's no skepticism and no pessimism surrounding your idea, it's one indicator that you might be a little bit late.” — Steven Bartlett (19:03:17 · et9rqthBK3WF8Sf3ca2Q) > > After ignoring two people who proved mathematically his ideas could not make money. > “Deliver value for people, deliver them good things and the money will take care of itself.” — Steven Bartlett (19:05:16 · et9rqthBK3WF8Sf3ca2Q) > > On persisting with foreign-language translations his finance team said would never pay. > “If I'm going to do this for 50 years, like I told you I was going to, then I have to wake up every day and look at my calendar and love my life.” — Steven Bartlett (19:13:30 · et9rqthBK3WF8Sf3ca2Q) > > Why personal interest is his decisive guest-selection filter. > “People will subconsciously take you to where they want to go if you just listen.” — Steven Bartlett (19:15:37 · et9rqthBK3WF8Sf3ca2Q) > > On never writing a question down and following the guest. > “The most important thing is just to care more than your competitors do about small details and try more things.” — Steven Bartlett (19:25:59 · et9rqthBK3WF8Sf3ca2Q) > > His answer when asked for the single most important factor. > “In a world where the correct answer is changing this quickly, nobody is really that smart.” — Steven Bartlett (19:33:20 · et9rqthBK3WF8Sf3ca2Q) > > Responding to an audience member who challenged his claim that he is not smart. > “I realized that the definition of the word company is group of people.” — Steven Bartlett (19:40:21 · et9rqthBK3WF8Sf3ca2Q) > > Why recruitment became his highest-leverage activity. > “This idea of constant reinvention has to be at the heart of your business and personal philosophy and it's not possible if you're romantic.” — Steven Bartlett (19:44:49 · et9rqthBK3WF8Sf3ca2Q) > > After the Ryan Crouser and Tiger Woods stories. > “The great advantage of the next decade is going to be to resist the temptation to defer your thinking to an AI.” — Steven Bartlett (19:57:00 · et9rqthBK3WF8Sf3ca2Q) > > Answering a parent whose children reject AI. ## Watch-outs - Rewarding successful experiments instead of the number of experiments; people stop proposing anything uncertain. - Copying tactics from books or talks and expecting them to last; they expire in months, only principles hold. - Judging which experiments are 'worthy'; some of the smallest changes had the biggest impact, so there is no experiment too small. - Treating followers as the bottom of the funnel; platforms are now the top and owned relationships are the asset. - Being in the business of just delivering information; AI will do that, so your days are numbered without a human layer. - Getting romantic about the method that made you successful; the biggest creators of a decade ago are gone because they believed the success was about them. - Listening only to current customers; they will keep you trapped sustaining your success while the next S-curve passes. - Compressing timeframes to save time (find a partner within a year, build a tower in sixty seconds); you overlook red flags and waste time trying to save it. - Deferring your thinking to AI; you lose the deep understanding needed to ask it the right question. - Assuming your team knows which of their constraints are paper; people cannot see the walls they are trapped inside, so leaders have to ask why on their behalf. --- Cole's Notes Vault 2026 — https://vault.chels.ai — built by chels.ai (https://chels.ai). Only stage content is published; quotes are verbatim from automatic transcripts with light punctuation. Speaker-stated cadence and editorial recommendations are labelled separately. Times are Eastern; Las Vegas was three hours earlier.