How to Improve Course Completion Before Buying More Ads

Before buying more ads, answer a delivery question: where does the learner journey first stop producing meaningful progress? Start with a defined group of learners, inspect the earliest unexplained drop-off, and test one change. The aim is not to make everyone finish every lesson. It is to understand whether your course helps learners reach the outcome they came for—and whether you can support additional learners sustainably.

Khan Academy’s first-party article describes building motivation into practice, progress, mistakes, and learning. That offers a useful starting point for examining course delivery, not evidence that the same practices improve completion in independent paid courses. The workflow below is an editorial diagnostic framework, not a validated retention formula.

1. Define meaningful completion before counting it

Write down the course outcome, the evidence required, and the observation window. For example: “Within six weeks of enrollment, learners submit a usable spreadsheet budget and explain their formulas against the published rubric.” Six weeks is an example, not a universal benchmark; choose a window that fits your course promise and expected workload.

Keep three measures separate:

• Course completion: the learner satisfies your stated participation or submission requirements.
• Demonstrated learning: the learner’s work meets the outcome criteria.
• Reported goal achievement: the learner says they obtained what they needed, possibly before finishing the full course.

A completion label should not substitute for reviewing learner work. NWEA distinguishes checks that guide instruction during learning from assessments that evaluate learning at the end. Use that distinction to decide which evidence belongs at each checkpoint.

Define inactivity separately—for example, no recorded task submission or reply during an agreed interval. Record it as an observation, not a diagnosis. If someone reports achieving their goal early, retain their original completion status and record goal achievement alongside it; do not quietly remove them from the denominator.

2. Map the journey and locate the first meaningful drop-off

Choose one enrollment group whose members have had the full observation window. Do not compare last week’s new learners with people who enrolled months ago. Record unique learner counts at these checkpoints:

1. Enrollment and successful access.
2. First meaningful task attempted.
3. First task submitted.
4. First feedback delivered and, where observable, acknowledged.
5. Next task or revision attempted.
6. Final requirements completed.
7. Final work meeting the learning criteria.

Define each event before counting. Opening a lesson is not the same event as attempting its task. Feedback sent is not feedback read. If your records cannot establish an event, mark it unknown rather than treating missing data as learner failure.

Inspect the earliest substantial unexplained loss between adjacent checkpoints. If learners gain access but never attempt the first task, investigate that transition before redesigning the final module. Show counts with denominators—such as “12 of 20 enrolled learners submitted task one”—and retain the enrollment-based view alongside checkpoint-to-checkpoint comparisons.

3. Separate possible explanations from evidence

For the checkpoint you select, list competing explanations. These are hypotheses to investigate, not labels to assign automatically:

• Onboarding friction: access instructions, navigation, or submission steps may be unclear.
• Missing prerequisites: the task may require knowledge learners have not yet demonstrated.
• Workload mismatch: the expected effort may not fit learners’ available time.
• Unclear value: learners may not see how the task connects to their goal.
• Insufficient support: a question, error, or revision may remain unresolved.
• Goal already achieved: a learner may have obtained the specific skill they wanted.

Match each explanation to something you can inspect: the actual access instructions, an unfinished response, a prerequisite check, a support message, or an optional learner reply. Login records alone do not establish which explanation applies.

Ask one short, optional question

Try: “What best describes your pause: trouble accessing the course, an unclear next step, difficult material, limited time, already got what I needed, or something else?” Allow learners to skip the question and avoid requesting sensitive personal explanations.

Compare replies with available work and support records. Record how many learners you contacted and how many answered. Do not assume respondents represent everyone who paused, or interpret silence as low motivation.

An exit ticket is one option for checking understanding among learners who reach a lesson checkpoint. NWEA describes exit tickets as formative checks used to inform the next instructional move, rather than graded tests. For implementation details, use SubSchool’s existing guide to ending each course week with a learner decision or action. An exit ticket cannot explain the experience of learners who never reached it.

4. Copy this completion-friction audit worksheet

Use one worksheet per enrollment group. Replace every bracketed field with your own definition or evidence. Keep identifying learner details outside the shared audit.

• Group and observation window: [Enrollment dates, eligibility rules, review interval].
• Selected checkpoint: [Earliest meaningful unexplained drop-off].
• One proposed intervention: [Change and evidence supporting it].
• Teacher time required: [Setup minutes and ongoing minutes per learner].
• Success and guardrail measures: [Checkpoint count, learning evidence, support workload].
• Owner and review date: [Responsible person and calendar date].

Hypothetical worked example

All numbers in this example are invented for illustration. A six-week spreadsheet course has 20 enrolled learners. Eighteen gain access, 10 attempt the first task, and eight submit it. All eight receive feedback; six attempt the next task. Five complete the course requirements, and four of those five meet the final learning criteria.

The creator investigates access-to-first-attempt: eight learners accessed the course without a recorded attempt. Three of those eight answer the optional question. Two report confusing submission instructions; one reports limited time. The other five remain unexplained.

The creator proposes one intervention: replace the first task’s submission instructions with a short numbered sequence and an example of the required file. The task and rubric remain unchanged. The worksheet allocates 45 minutes for setup, logs ongoing support separately, and sets a review date six weeks after the next group starts. The replies justify testing the change—not claiming it will improve completion.

5. Run a small pilot without promising uplift

Select one change that fits the evidence: simplify the first task, clarify the next step, add prerequisite practice, or offer targeted teacher support. Avoid changing all four together if your purpose is to learn which adjustment is useful.

Before launch, record the definitions, observation window, comparison group, teacher-time budget, and review date. At review, compare checkpoint counts, learner work, optional feedback, and actual support minutes. Keep the learning criteria consistent so that easier participation requirements do not masquerade as stronger learning.

A small before-and-after comparison is directional evidence, not proof of causation. Note differences in audience, enrollment timing, course version, or tutor availability. If more learners start but final learning evidence does not improve, investigate the next checkpoint rather than declaring the delivery problem solved.

6. Decide whether to improve delivery again or increase acquisition

Use the audit to make a bounded decision:

• Improve delivery again when the selected obstacle remains unresolved or learning evidence reveals another teaching need.
• Gather more evidence when the observation window is incomplete, records are missing, or groups are not comparable.
• Consider a limited advertising increase when the pilot supports a workable learner journey and you have capacity to deliver the resulting support.

Estimate the additional teaching workload before spending: expected new learners multiplied by observed support minutes per learner, plus fixed preparation time. Review acquisition spending alongside receipts, refunds, payment and platform costs, and the value of your teaching time. Set your own spending limit and stop conditions; no completion percentage automatically makes advertising worthwhile.

SubSchool offers editable course and practice drafts with teacher-reviewed assessment support. It can support repetitive preparation and review work while teachers retain authorship and the final educational decision. It is not evidence that an intervention works. Once your audit identifies the workflow you need, review SubSchool’s pricing against your delivery model and support capacity before committing.