Skip to content
Aplora
EdTech
AI Workflow Automation

Client under NDA

AI across four departments of a school: marketing, sales, content and operations in one loop

An online IT school was growing faster than its processes. Marketing, sales, content production and learning operations each worked their own way, and the customer path broke at the seams between them.

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

Someone putting sticky notes on a glass partition while colleagues work at laptops
Illustrative photo of the working context, not a screenshot of the client's system

  • Revenue

    ×3 in 4 months

  • Operating costs

    −60%

In short

  • Problem: four departments on one funnel each worked their own way, and the losses sat in the seams nobody owned.
  • Solution: one loop along the customer path with shared rules and data; AI components went in wave by wave, not all at once.
  • Result: per the client's figures, revenue tripled and operating costs fell 60% over four months — the automation's own contribution is not isolated from those figures.
  • Systems: CRM, webinar platform, LMS, knowledge base, e-mail and messengers.

Context

  • An online IT school selling programmes B2C; name under NDA.
  • Process volume: the full path from lead to course completion and the next programme.
  • Team: marketing, sales, methodologists and content production, learning operations.
  • Systems: CRM, webinar platform, LMS, knowledge base, e-mail and messengers; some internal processes under NDA.
  • The constraint became visible when every department coped on its own while end-to-end conversion fell: the losses were in the handovers.

Baseline

Baseline metrics with their sources. Without them, any later result has nothing to be compared against.
  • Before the work started we fixed: end-to-end conversion along the customer path, the load on each department, and the share of manual operations at the seams.
  • Data source: the client's CRM, LMS and internal records.
  • The growth figures need saying plainly. Revenue and costs over four months describe the business as a whole, not the measured contribution of the automation. Over the same period the school changed its product line and its team, and the market did not stand still. Separating one from the other would have needed a control group, and there wasn't one. We publish these numbers as the client reported them and do not present them as the effect of the implementation.
  • What is more honestly measurable here are the operational figures per department: those are covered in the separate cases, where the comparison stayed inside a single process.

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: the bottleneck is in sales, the bottleneck is in operations, or the bottlenecks are at the seams between departments.
  • The third was chosen: measurement showed each department handled its own stretch and the losses came at the handovers — a lead without context, a student without support, a material without an update.
  • The assumption: shared rules and shared data remove the seam losses even without changing anything inside the departments. It held only partly: some seams also required changing the processes themselves.
  • Stop criterion: each wave was assessed on its own, and the next began only if the previous one produced a measurable result in its department. A single stop criterion for a project like this cannot exist — otherwise it becomes “do everything and hope”.

What we implemented

  • Marketing: first-pass lead analysis and scoring, personalised nurturing, less noise in the funnel.
  • Sales: lead prioritisation, conversation quality review, automatic next steps and repeat touches.
  • Content production: preparing and updating course materials, drafts of quizzes and supporting materials.
  • Learning operations: student communication, deadline and status tracking, digests without manual assembly.
  • Shared across all four: common working rules and shared customer data — what the seams were missing.
  • Sequence: implementation went in waves, without pausing the business. Launching in four departments at once makes it impossible to tell what worked.
  • Human checkpoints: key decisions in sales and in work with a student stay with a person. Every action is logged.

How the process changed

Before

5 steps
  1. Marketing hands over a lead with no context
  2. Sales start the conversation from scratch
  3. The student reaches operations with no history
  4. Materials are updated separately from feedback
  5. Each department measures its own stretch

After

6 steps
  1. The lead arrives with a score and a touch history
  2. Sales see the priority and the context
  3. The conversation review stays on the shared record
  4. The student reaches operations with their history
  5. Feedback flows back into content production
  6. The end-to-end funnel is measured as a whole

What was stuck

Growth meant one of two things: losing quality, or hiring into every department in proportion to volume. No single department was the bottleneck — the seams were, and nobody owned them.

  1. 1The lead arrives with a score and a touch history
  2. 2Sales see the priority and the context
  3. 3The conversation review stays on the shared record
  4. 4The student reaches operations with their history
  5. 5Feedback flows back into content production
  6. 6The end-to-end funnel is measured as a whole
  7. Measured result

Results

Revenue

×3 in 4 months

Reported by: Client figures: revenue over the four months after launch. The contribution of the automation itself is not isolated from it — the market, the product line and the team all changed over the same period

Operating costs

−60%

Reported by: Client figures: cost trend over the same period; the cost composition was not disclosed to us

Economic impact

  • The effect combines all three types at once: freed time across four departments, additional revenue from higher end-to-end conversion, and losses prevented at the seams.
  • They cannot be merged into one figure — their reliability differs. The first is measurable, the second depends on more than the initiative, the third compares against an event that did not happen.
  • That is exactly why the headline “×3 revenue” and “−60% costs” are labelled here as business trend rather than implementation effect: they contain all three types plus things unrelated to the project.
  • Cost of ownership is higher than for a single implementation: a loop across four stretches needs its rules revised in concert. A change in one department breaks the next.

Adoption

  • The departments started working to shared logic: before that each had its own, and they did not meet at the seam.
  • Manual approvals between departments dropped — some existed only because data did not move automatically.
  • The new processes became the standard rather than an experiment: wave-by-wave rollout gave each department time to settle before the next part arrived.
«We stopped living in manual mode: the funnel became manageable, sales quality became visible, and production and operations scale without adding people.»

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

What's next

  • Scaling: extending the loop to the school's second product line.
  • Next initiative: end-to-end measurement — right now each stretch is measured separately, and the seams' contribution is less visible than it should be.
  • What we decided against: collapsing the effect of four stretches into a single ROI figure. Such a figure looks convincing and cannot be checked, which means it is not an argument.

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