Skip to content
Aplora
EdTech
AI Workflow Automation

Client under NDA

One account manager runs twenty-five cohorts instead of five

An online IT school running intakes and cohorts. Learning operations — schedules, reminders, attendance and deadline tracking, cohort communication — rested on account managers doing it by hand.

Published: 2026-08-24 · updated: 2026-08-24

Hands filling in a paper monthly planner beside a keyboard
Illustrative photo of the working context, not a screenshot of the client's system

  • Cohorts per account manager

    25Baseline: 5

  • Cost of running one course

    −40%

In short

  • Problem: learning operations rested on manual work, and growth in course count hit that ceiling first.
  • Solution: course rules as the source of truth, automatic event-driven reminders and communication, attendance and deadline tracking, and escalation to the account manager instead of a digest.
  • Result: one account manager runs 25 cohorts instead of 5; by the client's own model, the cost of a course is 40% lower.
  • Systems: LMS, messengers, internal cohort records.

Context

  • An online IT school; name and region under NDA.
  • Process volume: several parallel intakes, each with cohorts on their own schedule and deadlines.
  • Team: learning account managers plus a head of operations.
  • Systems: LMS, messengers for cohort communication, internal records; some standards under NDA.
  • The constraint became visible when new programmes launched: sales grew faster than operations could support them.

Baseline

Baseline metrics with their sources. Without them, any later result has nothing to be compared against.
  • Before the work started we fixed: cohorts per account manager, the share of reminders sent by hand, and the time to react to a missed class or a blown deadline.
  • Data source: the client's internal records.
  • The cost of a course was computed from the client's own model; its composition was not disclosed to us, so the number is published as their calculation rather than our measurement.

Diagnosis

Which hypotheses were considered, why this one was chosen, what was assumed, and the condition under which we would have stopped.
  • Three hypotheses were on the table: too few account managers, missing LMS features, or too many operations requiring a manual trigger.
  • The third was chosen: measurement showed most of the time went into actions fully determined by the schedule — things derivable from rules rather than decided afresh each time.
  • The assumption: course rules can be described formally. It was tested on three programmes — and failed on one, whose schedule shifted mid-course; that one was kept out of the first wave.
  • Stop criterion: if describing the rules does not at least halve the share of manual actions, the automation does not pay for its upkeep and the work stops.

What we implemented

  • Data sources: the course schedule and rules, the attendance log, assignment due dates, cohort membership.
  • Business rules: course event → who gets what. All described in one place and editable by operations without a developer.
  • Automation: student reminders, event-driven communication, attendance and deadline tracking, cohort reporting.
  • Integrations: LMS, messengers, internal records.
  • Human checkpoints: the account manager receives an escalation and decides. Withdrawal, deadline extension and conflict handling are never initiated by the system.
  • Monitoring: the share of escalations closed without intervention, and time to react — together they show whether the rules work or have turned into noise.

How the process changed

Before

5 steps
  1. The account manager keeps cohort schedules in their head and a spreadsheet
  2. Reminders are sent by hand down a list
  3. Attendance is checked when there is time
  4. A missed class is noticed at the next one, or later
  5. A cohort report is assembled on request

After

6 steps
  1. The course rules are described once
  2. Reminders go out on schedule
  3. Communication is tied to course events
  4. Attendance and deadlines are tracked automatically
  5. A departure reaches the account manager as an escalation
  6. A cohort report is available at any moment

What was stuck

Account managers kept cohort statuses in their heads. Every new course meant either overload or loss of control: missed classes and blown deadlines surfaced after the fact, when the student had already fallen behind.

  1. 1The course rules are described once
  2. 2Reminders go out on schedule
  3. 3Communication is tied to course events
  4. 4Attendance and deadlines are tracked automatically
  5. 5A departure reaches the account manager as an escalation
  6. 6A cohort report is available at any moment
  7. Measured result

Results

Cohorts per account manager

5 → 255 → 25

Reported by: Client figures: actual workload before and after at unchanged headcount

Cost of running one course

−40%

Reported by: The client's own cost model; the composition of the costs was not disclosed to us

Economic impact

  • The effect is freed operations time: the same people support several times more cohorts.
  • The drop in course cost is the client's calculation from their own model; we did not verify it and do not disclose the cost composition.
  • Cost of ownership: course rules live with the programme. Change the programme and the rules must follow, or the reminders start lying.

Adoption

  • Account managers work to shared rules rather than each in their own way.
  • For students, communication became predictable: the reminder always arrives, rather than when someone remembers the cohort.
  • The solution owner is the head of operations, who also owns the course rules.
«Cohort support used to rest on manual reminders and constant back-and-forth. Now the system runs schedules, deadlines and communication itself — the account managers get more done, without the chaos.»

Client under NDA — Head of operations, EdTech (NDA)

What's next

  • Scaling: programmes whose schedule shifts mid-course — the ones left out of the first wave.
  • Next initiative: spotting students losing pace early, from attendance and submission signals.
  • What we decided against: automatic withdrawals and deadline extensions. A decision about a person is made by a person.

Have a similar workflow? Let's check whether the hypothesis transfers

Another company's result is not a promise. It does show where to look.

All cases