Treat Your Course Like a Product
Hypothesise the audience, interview them, iterate the ICP, and run three weeks — not one.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 6
- Confidence
- 89%
Nika built the most popular AI PM course on Maven by running it through the exact product process she teaches: audience hypothesis, direct customer interviews, iteration when the hypothesis was wrong, a duration tested against outcomes, and a personal relationship with every single buyer. The specific findings — three weeks beats one or two, a Discord community is load-bearing, and every applicant gets a personal message — are transferable to any cohort-based course.
Origin
Marily Nika's account of building her AI Product Management course on Maven, after someone else suggested she should teach.
Core principles
- 01A course is a product; run it through the same hypothesis-and-iteration loop.
- 02Your buyers have full-time jobs and families — the value has to justify the break from all that.
- 03People will pay to learn what you take for granted; it can be game-changing for them.
- 04Teaching is the fastest way to crystallise your own thinking — the uncomfortable student questions become new modules.
How to run it
- 1
Hypothesise who the audience is
Write down who you think this is for and what you think they want out of it. Nika's first hypothesis was software engineers who wanted to become PMs.
- 2
Interview them directly
Reach out and ask two questions: would you want to learn from me, and what specifically do you need answered?
Pro tip The second question is the syllabus. Build the modules from the answers, not from what you find interesting.
- 3
Pivot the ICP when the data contradicts you
Nika discovered the real demand was PMs who wanted to become AI PMs, not engineers who wanted to become PMs, and shifted the whole course accordingly.
Watch out The first ICP guess being wrong is the normal case, not a failure.
- 4
Set the duration by outcome, not convenience
Nika tested and found one week too short, two weeks still rushed, and three weeks correct — because three weeks gives everyone the chance to present their work and to get to know each other in the community.
Pro tip Pair the cohort with an offline Discord community; Nika calls it another important part, not an optional extra.
- 5
Build a personal relationship with every buyer
Message everyone, read every application, and meet some of them to answer questions — because they are trusting and paying a stranger.
- 6
Let student questions rewrite the curriculum
Ship, then absorb the questions you can't answer. On day one Nika was asked how ChatGPT was trained and how to assess trade-offs between two models — so she added a bonus section on how ChatGPT was trained.
In the wild
Nika began aiming the course at software engineers who wanted to become product managers. Interviews revealed the real demand: existing PMs who wanted to become AI PMs. She shifted the positioning, and the course became the most popular AI/PM course on Maven.
→ A course with the right audience, a three-week format, nine workshops with homework, and an end-to-end AI product built by every student — including a pair who raised funding for what they built in it.
Common mistakes
Assuming nobody would want to learn from you
Nika laughed when it was first suggested she teach — 'wait, people would want to learn from me, really?' They did. Underestimating the value of what you take for granted is the reason most experts never build the course.
Compressing the cohort to save effort
One and two weeks both failed because there was no room for students to present their work and build relationships — and the presentation is the most exciting, highest-learning part of the whole course.
Is it for you?
Best for
Operators and experts with a niche skill considering a first cohort-based course.
Not ideal for
Purely evergreen, self-paced info products where cohort community and presentations don't apply.
From the transcript
“I came up with some hypothesis as to who the audience was and as to what they were looking to get out of it and…”
“one week I find it too short two weeks it was still a bit rushed three weeks is excellent because you give the opportunity to…”
“people me want to learn what you take for granted for them it may be gamechanging it can be lifechanging”
From the episode
AI and product management
Marily Nika (Meta, Google)