Five mistakes course creators said out loud in 2025 and 2026

The mistakes are listed in the order they show up in a typical launch — from the earliest decision to the latest.

1. Pat Flynn (Lean Learning, 2025) — most creators spend six months building because they are afraid of the validation conversation. The fix is the conversation, not more work.
2. Justin Welsh ($10M retrospective, June 2025) — the single change he wishes he had made earlier was raising prices. Underpricing cost years.
3. Stu McLaren and Arvid Kahl (both 2025) — treating the audience as something you build after the product is the single most common failure mode they see.
4. Tiago Forte (2025 Annual Review) — large cohort programs break at scale. He pulled back from cohorts and moved most of the work to an AI-assisted program.
5. Pat Flynn again (Lean Learning) — the regret of waiting for perfect. First launches are supposed to be ugly. Ship at 70% or you ship six months late.

Flynn has been writing about validation since Will It Fly? in 2016. The retrospective tone in Lean Learning (2025) is sharper. Almost a decade later, he is still seeing the same mistake — and now naming the reason.

Most creators spend six months building because they are afraid of the conversation that would tell them whether to build. — Pat Flynn, Lean Learning (2025)

The argument: the "building" phase feels like progress because it is visible work. You can show somebody what you wrote, recorded, designed. The "validation" phase is uncomfortable because it produces almost no visible work. You sit with a stranger, ask them about their problem, and listen. There is nothing to show afterwards.

Flynn's regret is not that he did this once. It is that he kept seeing other creators do it and assuming the framework was the answer. The framework is not the answer. The conversation is the answer. Without it, the framework is just elaborate procrastination.

Welsh's June 2025 retrospective — written after hitting $10M in cumulative course revenue — is unusually direct about regret. Across 23 steps, the one he names as the change he wishes he had made earlier is pricing.

The instinct to start cheap so people will buy is the wrong instinct. Cheap pricing positions you as cheap. The customers it attracts are the ones who complain the most. Raising prices later is harder than starting higher. — Justin Welsh, June 2025

His point is that pricing is not a knob you turn after launch. It is part of the positioning. A course at $49 attracts a different audience than the same course at $499. Not richer or poorer — different. The $499 audience treats the course as an investment in their work. The $49 audience treats it as entertainment.

Welsh's regret is the years it took him to figure this out. He raised prices repeatedly across the $10M journey. Every time, revenue went up. Every time, customer-support volume went down. The signal was consistent from the first time he tried it. The mistake was waiting so long to try.

Two different vantage points, same conclusion. McLaren has watched hundreds of membership launches from the inside of his program. Kahl writes about audience-building as his entire body of work. Both said in 2025 that the single most common pattern in launches that go nowhere is creators trying to sell to an audience that does not yet exist.

The most successful businesses are built around audiences, not products. Find your people first. Listen to them. Then build something they cannot wait to pay for. — Arvid Kahl, The Embedded Entrepreneur

McLaren's version is more specific. He sees creators with no audience trying to do "big" launches — full sales pages, PLF sequences, ad spend — and getting nothing. His recommendation in 2025: if you have under a few hundred engaged people, a founding-member offer to that small group is the entire launch. There is no point doing the full PLF machinery when there is no one to do it to.

The regret in both their framings is the same: creators treat audience as a separate problem to solve after the product. By the time they realize the audience is the actual problem, they have already burned six months on the course.

Forte's 2025 Annual Review is one of the most unusually direct retrospectives published by a working course creator in the last few years. He spent most of the 2020s scaling Building a Second Brain into a flagship cohort-based course — running it at hundreds of students per cohort. In 2025 he pulled back, hard.

At a certain scale the cohort model stops doing what it was supposed to do. The teaching quality drops. The community thins out. The thing that made the cohort work in the first place — close attention, real connection between students — does not survive the headcount. — Tiago Forte, 2025 Annual Review

His response was to cut the large cohort program and move most of the teaching work into a new structure — Second Brain Enterprise, built with Hayden Miyamoto and centered on AI-assisted learning rather than scheduled cohorts. The first version did 200 participants from 150+ companies and produced $1.5M in revenue.

The regret in Forte's framing is not that he ran cohorts at scale. It is that he kept scaling them for too long after the quality signal had already turned. He admits in the review that he could see the data — engagement metrics dropping, alumni outcomes weakening — for at least a year before he acted on it.

The deeper lesson he names: every model has a scale ceiling. Cohorts work brilliantly between 30 and 200 students. They start to break above 300. The mistake is treating "more students per cohort" as success, when the success metric should be outcomes per student.

This is the final mistake in the launch sequence and, in Flynn's framing, the most expensive one. By the time a creator reaches the point of "almost ready to ship," they have already invested months of work. The regret is the additional months they then spend trying to make it "perfect."

You don't need to know more. You need to ship more. — Pat Flynn, Lean Learning (2025)

Flynn's argument is structural. The first launch is not the final product. It is a feedback mechanism for what to build next. A 70% version that ships now produces real customer feedback. A 100% version that ships six months later produces almost no incremental value over the 70% version — because the 30% of "polish" you spent six months on was guessed at, not informed by buyers.

The Lean Learning frame is built around five steps — Focus, Sprint, Stretch, Skirmish, Lean — designed to break the perfectionism loop. The Sprint phase is deliberately short and time-boxed. The Skirmish phase is the first real exposure to customers, ugly and unpolished. The Lean phase is iteration based on what they said.

The regret Flynn names is the years he watched other creators sit in pre-launch, polishing, until either the market moved on or they ran out of energy. The mistake is rarely the quality of the work. It is the order of when the work meets reality.

None of these creators talks about a mistake in execution. The work was usually fine. Flynn's courses were polished. Welsh's content was sharp. Forte's cohorts were well-designed. Kahl's products work.

What goes wrong is the sequence. Validation belongs before building. Pricing belongs near the start, not as an afterthought. Audience belongs before product. Scale belongs after the model has been tested at a smaller size. Shipping belongs at 70%, not 100%.

The regret in each case is years lost to doing the right work in the wrong order. The lesson is not a longer to-do list. It is a shorter one — done in a different sequence.

Each link below goes to the original primary source — the book, the newsletter post, the podcast episode where the creator named the regret in their own words.

• Pat Flynn — Lean Learning (2025). Mistakes 01 and 05 both trace back to this book.
• Justin Welsh — "My complete $10M journey (all 23 steps)", June 2025. The pricing regret is in steps 8 through 12.
• Stu McLaren — Stu McLaren on The Amy Porterfield Show, May 2025. The "few hundred people" launches.
• Arvid Kahl — The Embedded Entrepreneur (book + blog). Audience-first as a stance, not a tactic.
• Tiago Forte — Tiago Forte's 2025 Annual Review. The cohort-scale retrospective.