A training center goes digital, part 4: admin learns to follow up with data, and students get a first portal to log into

This series follows one training center through several phases of going digital. The journey is built from real implementation experience in this industry — the stages and challenges should feel familiar if you run this kind of school. Each part covers what was built, what was deliberately not built, and what evidence justified moving to the next phase.

This phase actually has two leads

After part three, the center had clean enrollment leads, and admin was used to following up forms. But two separate problems had to be solved together:

• For admin: follow-up cannot run on memory and luck. Who enrolled, which step are they at, should we contact them again — that needs a simple way to be recorded, or leads will leak out during follow-up.
• For students: after enrolling, students and parents keep asking “where’s the schedule, what about homework, what about grades?” If they always have to ask a teacher on LINE, the portal will never have a reason to be used.

So this phase built two things at once: a follow-up record for admin, and a login entry point for students.

Admin: from “reading forms” to “following up with data”

No full CRM here — just one small but crucial thing: every enrollment lead carries a follow-up record.

• who owns this lead;
• when they last contacted it;
• what the outcome was (enrolled, still considering, paused, given up);
• when to contact it again.

The method is plain: maintain this follow-up list in a place the school controls, and admin opens it, updates it and closes items every day. The point is the habit, not the tool. Only when admin genuinely keeps following up through this list for weeks — instead of going back to handling each chat — can the school talk about “running on data.”

A few months later, the school saw some real patterns for the first time:

• one course attracted noticeably more enrollment leads, but did not convert well after follow-up;
• another channel brought fewer enquiries, but a higher enrollment ratio;
• most leads were decided within a week of enrolling; those that went past two weeks rarely came back.

These patterns were not earth-shattering, but they were real data, not feelings. They decided what to build next.

Students: the first portal to log into

The learner portal phase one deliberately did only four things:

1. Login — log in with the contact details used at enrollment. No complicated new account flow to learn.
2. See the schedule — when my class runs, in which room, and what time the online class starts.
3. See homework — teacher-posted homework lives here, no more hunting through chat history.
4. See grades — test and homework results are here.

Why so little? Because this phase’s goal was not “feature-complete portal” but getting students to actually log in. If phase one were stuffed with forums, livestreams and an AI assistant, students would only get more confused. Four simple, genuinely useful entry points build the login habit better than a feature-rich empty house.

Teachers also had to keep it light: posting homework and recording grades must be fast enough not to feel like extra work. If teachers think it is a system that adds to their workload, it will not survive.

What this phase deliberately was not

• Not a full student management system. No payment management, attendance statistics, automatic term-start notices or refund flow.
• Not an online testing system. Students do not see test papers yet, and teachers cannot write tests online yet.
• Not AI. No test generation, no online Q&A. The portal is still forming a login habit; AI comes later.
• Not a mobile app. Just a web login entry point, usable from a phone browser.

The evidence that would justify phase five

Before the student portal went live, the center agreed on a rule: no AI talk yet — let the portal survive one or two full terms. It would watch:

• whether the student login rate stays stable, rather than spiking with novelty and fading;
• whether teachers genuinely post homework and grades in the portal instead of continuing on chat;
• whether students move from “asking teachers for the schedule” to “checking the portal themselves,” and whether those repeat questions actually drop.

Only when the portal is used steadily, the data is reliable, and teachers are on board does the school earn the right to talk about the thing it has been waiting for — plugging in AI over an API, so AI helps write tests and answers students’ questions online.

Part 5 looks at how AI arrived late: only after the entry point, the data, the portal and the habit all held.

---

The boundary matters: admin’s follow-up records and the student login portal are the school’s own independent system projects. Online testing, CRM, online payment, and AI test generation and online Q&A plugged in over an API are separate system projects with their own scope — they are not part of a standard website package, and the AI features in particular are a later phase that comes after the entry point, the data and the portal all exist.