Designing a course people actually finish, with the eleven lesson types you have

Course completion rates are dismal almost everywhere, and the standard explanation — people are lazy, people buy aspirationally — is mostly a way of not looking at the design.

The two things that actually predict finishing are pace and evidence of progress. A course that arrives all at once is a 40-hour obligation sitting in a tab. A course that arrives in pieces is a habit. And a member who cannot see how far along they are will assume it is further than it is, then feel worse each time they check.

Here is how to build for both with what Klaas gives you.

Eleven lesson types, and the four people forget

A lesson is not necessarily a video. There are eleven types:

Text · Video (upload) · Video (link) · Audio · PDF · Embed · Assignment · Quiz · Event RSVP · Form · Waiver

Most courses use two of these — video and text — which is why most courses feel like a playlist. The interesting ones are further down the list:

• Assignment turns a lesson into a submission. The member posts their work, and it is visible in the course. This is the single biggest lever on completion, because a course where you have submitted something is a course you are in, not one you are watching.
• Event RSVP puts a live session inside the curriculum. Lesson 6 is not a video about the workshop, it is the workshop, and the member's RSVP happens in place. Cohort courses live or die on this.
• Form collects structured answers. Useful as lesson 1 — "what are you here to do?" — which gives you the single most valuable piece of data you will get about that member, at the moment they are most motivated to give it.
• Waiver is a typed-name consent step. Niche, and if you teach anything physical, financial or medical you know exactly why it is there.

The practical advice: put an assignment or a form in the first 20% of the course. Not at the end. The goal is to convert a viewer into a participant while they are still enthusiastic, and the end of a course is precisely where you no longer have the people you were trying to reach.

Drip: pace it, but pace it for a reason

Each section can be released immediately, on a specific date, or N days or N months after the member enrols.

Those last two are the ones that matter and they are different tools:

• Days after enrolment is a self-paced drip. Everyone gets their own clock starting when they join. Use it for evergreen courses.
• A specific date is a cohort. Everyone moves together, which means the discussion under lesson 3 happens while everybody is on lesson 3, rather than being spread across nine months and dead.

You can attach an email to each release, with merge variables, so the member is told a new section is open rather than being expected to remember to come back. Use it. A drip with no notification is not pacing, it is just content you have hidden from a paying customer.

Two guardrails worth knowing: a drip cannot be activated on a section that is still a draft, and a locked lesson tells the member why it is locked — waiting on a date, waiting on the previous lessons, or locked outright — rather than showing a padlock and leaving them to guess. "When does this open?" should never be a support question.

Where drip goes wrong: dripping a reference course. If somebody bought your thing to solve a problem this afternoon, releasing chapter 4 in three weeks is not pedagogy, it is an obstacle. Drip changes-of-behaviour courses. Do not drip manuals.

Quizzes: two design decisions worth copying

A quiz lesson has a question builder, a passing grade, retake settings, and an attempts history.

Two things about it are deliberate and unusual.

Options are always checkboxes, never radio buttons — even when there is exactly one correct answer. Radio buttons announce "there is exactly one right answer here", and checkboxes with two ticks announce "there are two". Rendering the control based on the answer key leaks the shape of the answer before the member has thought about it. Making everything a checkbox costs a little elegance and removes the tell.

Grading is exact set equality, done on the server. All the right options, none of the wrong ones. Partial credit sounds kinder and makes a passing grade impossible to interpret — 60% on a partial-credit quiz might mean solid understanding or might mean a lucky guess pattern. And doing it server-side means the answer key is not sitting in the browser, which it very often is elsewhere. If you have ever wondered whether your students could read the answers out of a page's network tab: on a lot of platforms, yes.

You choose whether to show correct answers afterwards. Showing them makes the quiz a teaching moment. Hiding them makes it an assessment. Both are valid; know which one you are running.

Completion, and the one thing we deliberately do not offer

Most lessons complete with an explicit Complete and continue button. That is on purpose: auto-completing a video when it reaches the end marks a lesson finished for somebody who left the tab open, and now the progress bar is lying.

Some types do not complete manually — a quiz completes by being passed, a waiver by being signed. And you can set a section to enforce completion, so an assignment or a form has to actually be submitted rather than clicked past.

There is no member-facing "mark as incomplete". Progress moves one way. This is a real constraint and I will not pretend it never itches: a member who ticked something by accident cannot untick it. We accepted that because a reversible progress bar makes every rule built on progress unstable — sequencing, certificates, pathway locks all start flickering — and the failure mode of an accidental tick is much smaller than the failure mode of a certificate that can be revoked by a misclick.

Certificates issue themselves

When a course reaches 100%, the certificate is issued in the same operation that recorded the final lesson. Not by a nightly job, not by you noticing.

The reason to care about that detail: anything issued by a scheduled job can silently stop being issued, and nobody finds out until a member asks where their certificate is — which, for most communities, will be several months and several hundred certificates later.

If you have designed a certificate template, there is one thing to check: a course can be published while the template is inactive, and certificates only issue from an active template. Klaas raises a warning for exactly that combination, because it is the sort of thing that silently produces zero certificates for a month.

When to sequence, and when to leave it alone

You can lock lesson order within a course, and you can lock course order with a learning pathway.

Sequence when later material genuinely assumes earlier material. Do not sequence a library. The most common self-inflicted wound in course design is forcing a linear path through content that people bought for one specific chapter — they hit three padlocks on the way to the thing they wanted and conclude the product is obstructive rather than thorough.

A quick test: if a member emailed asking to skip to lesson 8, would you say yes? If yes, do not lock it.

The short version

1. Put something the member does — assignment or form — in the first fifth.
2. Drip by days-after-enrolment for evergreen, by date for cohorts, and always with the email.
3. Make lesson 1 short. Absurdly short. The first completion is worth more than the first insight.
4. Quiz to teach, not to filter, unless you are certifying.
5. Only sequence what genuinely needs sequencing.

None of this is about the video quality, which is the thing most people spend their money on and the thing least correlated with anyone finishing.