Karina, who owns a UX design school in Almaty, wants to put "75% of students mastered UX design" on the landing page. A hundred people paid, forty sent in the final assignment and thirty met the criteria. That 75% is the share among those who submitted. About the other sixty, the school knows nothing.
Take an eight-week programme at 150,000 ₸; the numbers are illustrative. The marketer needs two tables about the same cohort: what the students learned, and how much money the school kept after teaching, support, refunds and acquisition. Video views will not build the second, and payments will not tell you whether the skill was learned.
Turn the promise into a task
Instead of "become an in-demand UX designer", the school describes what a graduate will be able to do: run a user test, find where users get stuck and justify changes to the mockup. The course page shows the entry requirements, the workload, a sample assignment and how feedback works. That way a person knows before paying whether the course suits them and whether they can find the time.
The Eberly Center at Carnegie Mellon recommends aligning objectives, activities and assessment. If the school promises UX design, the final assignment has to require design work. A terminology quiz checks terminology; a user test and mockup changes call for a different kind of work.
The curriculum lead sets the criteria in advance: the share of participants who completed the scenario is calculated, the causes of errors are explained, the mockup changes are checked with a repeat test. In class, students practise on a different task and get feedback. Otherwise the final assessment asks for something the course never prepared them for.
Example of assessing a learning outcome
Course screen
100%
- Separate skill assessment
- Solve a new task
Final-task criteria
- Calculated
Share of participants who completed the scenario
- Explained
Causes of errors
- Checked
Mockup changes with a repeat test
Set criteria in advance; during the course, practise on another task and receive feedback.
A promise of a job or an income needs other grounds: being able to do the assignment is only part of what the profession asks. A graduate's story can be told as their own experience, with its conditions and path, but their result must not be passed off as the whole cohort's.
Three denominators instead of one percentage
For each of the hundred who paid, 90 days have passed since the course started. Those who were later refunded stay in the group. If you remove the inconvenient rows, the cohort will look better than it really was.
| What happened by day 90 | Students | Share of the original cohort |
|---|---|---|
| Started the course | 90 | 90% |
| Watched all required videos | 70 | 70% |
| Submitted the final assignment | 40 | 40% |
| Met the final assignment criteria | 30 | 30% |
The thirty who succeeded are among the forty who submitted. Watching every video, though, is not required for the assignment in this example: someone may already know part of the material. So the rows do not form a funnel where each step follows from the one before.
That gives three metrics. Final assessment coverage is 40 / 100 = 40%. Success among those who submitted is 30 / 40 = 75%. A demonstrated result exists for 30 / 100 = 30% of the original cohort. All three go in the report, each labelled with the group it is counted from.
For the sixty who did not submit, the result is unknown. Some did not learn the material, some had learned it before, some solved the task and never sent it in. Calling them all failures is as wrong as calling them all successes.
A learned skill and its growth are different things
The final assignment shows what a person can do at the time of assessment. To judge what changed, you need a comparable entry assignment: some students may have had the skill before the course.
The entry and final assignments must test the same objective with comparable criteria. If a student repeats familiar answers, the score rises while they may still fail a new task. The curriculum lead owns what the assessment tests, and the analyst keeps each student's results linked.
Even the difference between entry and exit does not show what one course contributed: the person may have been studying elsewhere and practising at work. For the landing page it is enough to say how many students completed the final assignment. That claim does not need to credit the course with everything that happened to someone over those months.
Help depends on where students stop
An email saying "We haven't seen you in a while" says almost nothing about where a student got stuck. First look at their path: got access, found the first step, started the work, read the feedback.
| Observation | What the team looks at |
|---|---|
| First lesson not opened | Access, schedule and how clear the first step is |
| One assignment rejected several times | The error, how the criterion is explained and the chance to practise |
| A group stalled on one module | Instructions, materials and whether the task is manageable |
| Course completed | Applying the skill and the next need |
The cause comes from a conversation or a support request; a single analytics event will not explain it. An unopened lesson does not tell you that a person is lazy or that the lead was bad.
The support you promise has to fit the team's schedule. If the school runs group consultations, the landing page cannot promise everyone a personal mentor. The budget includes marking, resubmissions and answers in the group's Telegram chat, including whatever was promised after the last class.
The course is over and the graduate stops logging in, which is normal. A cohort helps find where students stall during the course, and buying the next course is a new need. There is no reason to push graduates back onto the platform for an activity metric.
This cohort's money
A hundred students paid 150,000 ₸ each, 15,000,000 ₸ in total. Within 90 days five people got a full refund, which is 750,000 ₸. The school spent 5,000,000 ₸ on teaching and supporting the cohort and, by the agreed budget, another 3,000,000 ₸ on acquisition.
| Item for the period | Amount |
|---|---|
| Payments received | 15,000,000 ₸ |
| Refunds | 750,000 ₸ |
| Left after refunds | 14,250,000 ₸ |
| Teaching and support | 5,000,000 ₸ |
| Acquisition | 3,000,000 ₸ |
| Remainder on the items counted | 6,250,000 ₸ |
From 15,000,000 ₸ we subtract 750,000 ₸ of refunds and 8,000,000 ₸ of costs, leaving 6,250,000 ₸. The calculation is simplified: fixed costs, taxes and obligations beyond the period are not deducted. If graduate support continues, money has to be set aside for it.
Acquisition cost by the budget is 3,000,000 / 100 = 30,000 ₸ per person who paid. Counting only the 95 without a refund gives about 31,579 ₸. When comparing cohorts, stick to one definition or show both: quietly change the denominator and CAC starts to mean something else.
Full CAC includes every acquisition cost. If the three million is only part of it, say so in the label. Teaching and support sit next to acquisition, not inside it, and each item is counted once in the overall economics.
Instalments add a schedule of receipts. Future payments under a contract are not yet in the school's account. If a financial partner pays, the calculation uses its receipts and fees on its terms. Match the ad budget and teacher pay to when the money will actually arrive.
Better-suited students, fewer sales
Suppose the school added a sample assignment and the real workload to the course page. Payments fell, while the share submitting the assignment rose. One percentage is not enough to conclude anything: look at the number of demonstrated results, refunds, support workload and the cohort's money.
Compare at the same time from the start and with the same criteria. Note what else changed: price, teacher, programme, audience. If any of these changed at the same time as the landing page, you cannot put the whole difference down to the page.
For the next test, the team changes one thing and agrees in advance which result will settle the decision. Views, attendance and sales work as intermediate metrics, but the skill promised to students is checked with its own assessment.
Exercise: copy for the page and a table for the owner
Take a finished cohort from your project, or draft a report form for a future one. The reader of the report should understand who is in the calculation and what the school actually measured.
0 of 6 done
Karina brings the next cohort: a hundred paid, by day 90 sixty submitted and forty-five met the criteria. "75% of students mastered UX design" again says more than the school knows. 45 / 60 = 75% is the share among those who submitted; coverage is 60%, and a demonstrated result exists for 45% of the original cohort.
For the page, "45 of 100 participants met the final assignment criteria by day 90" works, with the assignment described next to it. Claiming the skill came from the course needs other grounds, such as a comparable entry assignment. And Karina, as the owner, also needs the economics of the same cohort next to these figures.