Cohort or self-paced?

You can run both, and plenty of course businesses do: a self-paced library and a live cohort, sold side by side. The real work is knowing which buyer each one serves. Some people want the material and their own calendar: they'll buy the recording and get to it when they get to it. Others want an outcome, with you in the room, and will pay for the difference.

Self-paced is McDonald's. Open all hours, same product every time, one kitchen that serves thousands. Nothing wrong with that: it's convenient, it's predictable, and it scales without you. A cohort is the nightclub. Capacity is strictly limited, the room itself is the product, and the price of admission reflects who's in it. The nightclub doesn't try to serve a million customers a day, and it doesn't apologize for that. Volume was never the goal. Same audience, different aspirations, two doors.

The room is what you're selling

What a cohort sells, underneath the curriculum, is the allure and promise of instructor involvement. You (or someone you pay) are actually engaging with your members: answering this group's questions, reviewing this group's work as it comes in. That's the the biggest thing a recording can't deliver, and it's why cohort courses command live prices. Recording scales your content. It doesn't scale you.

The seats are limited because you and your expertise are limited, and that limit is a feature. It's what makes the cohort a high-ticket offer instead of another library competing on price.

Two ways to work the room harder:

Unbundle the involvement. "How much of me" is a product line, not a binary. A kickoff call, weekly office hours, personal work review, a graduation session: each is a tier. Offer the same content at three levels of access and buyers sort themselves by budget without you negotiating anything.

Treat sell-outs as pricing information. The nightclub raises the door price; it doesn't add tables. A cohort that fills in 48 hours is telling you the next one should cost more. Self-paced pricing moves with production cost; cohort pricing moves with demand, and demand is right there in your enrollment numbers.

The library is the long tail

Self-paced is the volume business. Record the course once, put it up, and sell it while you sleep. A student can buy at 2 a.m. and start lesson one at 2:01. For reference material, tool tutorials, and anything people look up rather than complete, it's the honest format; nobody needs a cohort to learn CSS flexbox.

Two honest weaknesses. With no deadlines and no group, many buyers drift, and people who don't finish your first course rarely buy your second. And a library of videos competes with every other library of videos, including free ones, which caps the price.

But notice who the library is for: the same audience, split by commitment rather than interest. Plenty of people want your material and can't clear a Tuesday-evening obligation for eight weeks. The library serves them without cheapening the room. And the two products feed each other. Library buyers are your warmest leads for the next cohort: they know the material, and some of them will pay to finally do it with you in the room.

If drift worries you, steal lightly from the cohort world: a suggested weekly pace by email, a progress email at week three, even a "here's who else bought this month" list. The drift problem is partly a design problem, and deadlines can be suggested without being enforced.

The students teach each other

The underrated product of a cohort is the other students. A group that's moving through the same material at the same time asks questions you'd never think to answer, swaps fixes and shortcuts, and keeps each other moving. Self-paced buyers work alone whether they like it or not; cohort members have a room.

That only pays off if you design for it. Left alone, a group chat goes quiet in two weeks. What works:

Pair people up. Peer review partners or study pods of three or four. A person who'd drop out alone sticks around because someone would notice.
Answer second. When a question lands in the community, wait a beat before answering. Students are often faster and better at explaining to each other than you'd expect, and every peer answer is teaching you didn't have to deliver.
Assign work that needs the room. Group deliverables, crit sessions, pair exercises. Content that makes sense alone gets consumed alone; make collaboration the cheapest path to the result.
Bring back alumni. Last run's graduates make this run's teaching assistants. It solves your scaling ceiling, it costs you a discount instead of a salary, and being asked is an honor that keeps alumni attached.

The room also leaves artifacts: shared notes, student showcases, before-and-after examples. That residue is next cohort's marketing, and it's better than anything you'd write yourself.

For private corporate cohorts, the effect is the point. An employer who puts a whole team through your course isn't buying content, they're buying shared vocabulary and a team that learned it together. Who's in the room is the deliverable.

The calendar is an asset

Self-paced has no clock, which sounds like freedom until you notice it also has no urgency. A cohort's calendar works for you:

"Enrollment closed" is real scarcity. The line outside the nightclub is the waitlist, and it's honest: no fake countdown timers, just a door that actually closed. People who miss this run sign up for the next one while their motivation is hot.
The year gets a heartbeat. Launch seasons, an off-season, alumni early-bird windows. Pulses of activity are easier to market and easier to plan than a flat, endless "available now."
Founding-member pricing fills the first room. The first run is the cheapest the course will ever be, and you say so: it's a discount in exchange for messiness. That's an honest trade that fills seats and buys you testimonials.

And there's a middle path most people miss: the rolling cohort. Instead of waiting to fill one big fall class, start a new small cohort every month. Each group gets its own start date, its own drip schedule, its own instructor. Enrollment is always open, but every student still gets the room. You trade the big launch for a steady drumbeat, and in Torii it's not even extra machinery: tag-driven enrollment plus each cohort's own start date does it natively.

So which do you build first?

What are you actually selling? If the answer is "the material," record the library. If it's "an outcome, with me involved," run the cohort and price the involvement.
Can you commit to the calendar? A cohort with a distracted instructor is worse than a good self-paced course. If your next quarter is unpredictable, sell the library until it isn't.
Does the work happen between lessons? Coaching programs, certifications, implement-a-thing courses live on practice and feedback. That's cohort territory: the group keeps people doing the work.
Run the refund test. A library refund costs you nothing; a cohort refund costs a real seat that could have gone to someone on the waitlist. If your audience is refund-prone, that asymmetry should shape both your price and your policy.
Run the lifestyle test. A library travels. A cohort chains you to the calendar. Two launches a year is a different life than twelve, and only one of those is right for some seasons of your business.

Whatever you choose, the mechanics are the same in Torii: memberships gate the content, tags drive enrollment, and cohorts simply attach a shared schedule and shared variables to a group of members. If you're running the cohort model, continue to Running a Cohort.