Solutions · By need · Student success

Retention is decided in week three, not at the exam board.

Every institution knows who withdrew. Almost none can say who was about to. The signals arrive on time and separately: attendance in a register, marks in a gradebook, engagement in the learning environment, an unpaid instalment in finance. Creatrix reads them from one student record, explains the risk, and puts an alert in front of a named person while there is still a term left to change.

Request a demo See the difference

Why it goes wrong

The problem was never prediction. It was the handoff.

Most early alert programmes are built on a reasonable premise and fail on the mundane part. A list is produced, it lands in an inbox, and nobody owns the next step. By the time a concern reaches the person who could act, the student has already stopped attending, and the conversation has become an exit interview.

Three things go wrong every term. Risk is discovered late, because the signals sit in different systems. Concern is raised without an owner, so a shared inbox absorbs it. And nothing records what was tried, so the institution repeats the same interventions without ever learning which of them worked.

The problem: a list nobody is accountable for.
Six reasons the alert arrives too late None of them are about the accuracy of the model.
  • Signals live in separate systems Register, gradebook, LMS, finance
    Nobody sees the pattern until it is a withdrawal
  • The data is a day or a week old An overnight extract into a dashboard
    You act on last week's version of the student
  • The score cannot explain itself A number, with no reasons attached
    Advisors do not trust it, so they ignore it
  • The alert has no owner Emailed to a list, or a shared inbox
    Everybody assumes somebody else called
  • The intervention is not recorded A conversation, remembered by one tutor
    The next person starts from nothing
  • Nobody can say what worked Retention reported as a rate, once a year
    The same effort is spent again, blindly

Six gaps, one consequence: the institution learns about risk after it has become an outcome.

The difference

A report, a dashboard, and a loop are three different things.

All three describe risk. Only one of them changes what happens to a student, because only one of them ends with a person and a due date.

Option one

The retention report

Produced annually, read by a committee.

Signals fromExtracts pulled after the fact.
What you getA rate, and a comparison with last year.
Who actsNobody in particular.
ExplainableAt cohort level only.
Proves impactNo. It describes the outcome.
Option two

Analytics bolted on the side

A dashboard beside the systems of record.

Signals fromAn overnight extract, one day behind.
What you getA ranked list of names.
Who actsWhoever opens the email.
ExplainableA score, rarely its reasons.
Proves impactNo. It cannot see what was done.
Option three

An intervention loop on one record

Where the signals and the action are the same system.

Signals fromThe register marked this morning.
What you getAn alert with an owner, a date and a state.
Who actsThe named advisor or tutor responsible.
ExplainableThe reasons travel with the score.
Proves impactYes. Action and outcome sit on one record.
The test of a success programme is not whether it can rank students. It is whether anything happened afterwards.

How it is done

Six steps, and the loop closes on the sixth.

The first three are about seeing risk in time. The last three are the ones most programmes never build, and they are what make the next term different.

Nothing visible Too late to change Seen and owned Being worked
  1. 01

    Read the signals live

    Attendance marked this morning, the mark entered yesterday, engagement in the learning environment and a hold placed by finance, read from the record rather than from an extract.

    Attendance · Assessment · Finance
  2. 02

    Score risk, and show the reasons

    Ranking without reasons gets ignored. The factors that produced a score travel with it, so an advisor walks into the meeting already knowing what to ask about.

    Explainable · Ranked
  3. 03

    Give the alert an owner

    When a threshold is crossed, the alert is raised against a named student, routed to the advisor or tutor responsible, and given a due date and a state. A tutor can raise one manually too.

    Owner · Due date
  4. 04

    Act, and write it down

    The meeting happens, the case note is recorded, and a referral goes to support services where it is needed, all against the same student record rather than in somebody’s notebook.

    Advising · Referral
  5. 05

    Follow up until it closes

    An alert stays open until it is resolved, which is the difference between a system and a good intention. A concern cannot quietly disappear between a meeting and the next term.

    Open · Resolved
  6. 06

    Measure what actually worked

    Because the alert, the action and the student’s subsequent attendance, assessment and progression sit on one record, you can compare students who received an intervention with those who did not.

    Outcome · Evidence

Who acts

One record, six people, six different views of it.

A success programme only works if each person opens something useful to them. Same student, same signals, a different question in each pair of hands.

Academic advisor
SeesA caseload ranked by risk, and why each student is on it before the meeting starts.
Tutor or lecturer
SeesThe class list, with the ability to raise a concern in a category with a note.
Head of department
SeesWhich cohorts are drifting, and whether alerts in their department are actually being closed.
Student services
SeesReferrals arriving with context, so the first question is not what happened.
Registrar
SeesProgression consequences of the term as it is running, not as it is reported afterwards.
Leadership
SeesWhether interventions changed outcomes, at cohort and programme level, with the evidence attached.

Go deeper

Success is not a module. It is the record, read attentively.

The signals come from registration, attendance, assessment and finance. These are the pages that explain where they originate and what runs the loop.

The product Student Success & Retention The application itself: the risk board, a student’s risk detail, the alert queue, advising and the analytics. See the product → Lifecycle · L05 Student Success & Records Registration, progression, records and graduation: the record the signals are read from. See the lifecycle → Lifecycle · L03 Academic Operations Where attendance is actually marked, which is the earliest reliable signal you have. See the lifecycle → Lifecycle · L04 Assessment & Outcomes Missed and failed assessment, and the attainment picture behind a struggling student. See the lifecycle → Lifecycle · L08 Student Financials Outstanding fees and holds, one of the strongest and most ignored risk signals. See the lifecycle → In practice Institutions already doing this Universities and colleges across Europe, the Middle East, Southeast Asia and Oceania. Meet our customers →

What it runs on

The products that make the signals trustworthy.

Early alert is only as good as the data underneath it. These are where the signals come from, and they can be adopted one at a time.

The loop
Student Success & Retention →

Explainable risk, early alerts with owners, advising caseloads, referrals and the analytics that show what worked.

RiskAlertsAdvising
 The record
Student Information System →

Registration, progression and records: the student the signals are attached to.

RecordsProgressionRegistration
 Attendance
Class Scheduling →

Sessions and registers, which is where the earliest and most reliable signal is generated.

SessionsRegistersContact hours
 Assessment
Examination Management →

Missed assessments, marks and results, read as they are entered rather than at the board.

MarksEligibilityResults
 Finance
Student Finance & Billing →

Outstanding instalments and holds, tied to academic events rather than tracked in a separate silo.

FeesHoldsAid
 Attainment
Outcomes & Attainment →

Where a student is struggling against the outcomes, not just against a grade average.

OutcomesAttainmentCohorts
Start with the loop. Every lifecycle you connect makes the signal sharper.

Institutions running the student record on Creatrix see the term as it happens.

Creatrix Campus has helped us centralise all student information in one reliable and intuitive platform, improving efficiency and reducing errors. We use it to manage student data, schedule sessions, and automatically generate transcripts, which has simplified many of our academic processes.
Omar Mansour · Head of Registry Forward College
The system ensures proper data management and better information flow across departments. The automation of processes allows us to focus on more important tasks.
Nurashikin · IT Manager i-CATS University College

Frequently asked

Plain answers about student success.

Why do early alert programmes usually fail?

Three reasons, and none of them are the model.

Risk is discovered late, because the signals arrive in separate systems. A concern is raised without an owner, so nobody is accountable for the next step. And nothing records what was tried, so the institution cannot tell which interventions worked and repeats all of them.

Which signals actually predict risk?

Attendance decline, missed or failed assessments, engagement with the learning environment, outstanding fees and financial holds, and prior-term performance.

Institutions can weight any programme-specific factor of their own on top. The point is not a universal model, it is that the factors are visible and yours to adjust.

Does AI decide who is at risk?

No. Risk is ranked and explained, and a next-best action can be suggested, but a person decides what happens.

Nothing is placed on a student’s record and no student is contacted without somebody choosing to do it. That boundary matters, because an advisor who does not trust the list will not work it.

How is an alert different from a report?

An alert has an owner, a due date and a state. It is raised against a named student when a threshold is crossed, routed to the person responsible, and it stays open until it is resolved.

A report has none of those properties, which is the whole reason reports do not change outcomes.

Can tutors raise a concern themselves?

Yes, and they should. A tutor or lecturer can raise a flag with a category and a note.

It enters the same queue as a system-generated alert, with the same routing, ownership and follow-up, because human judgement is a signal too and often the earliest one available.

Can we prove an intervention worked?

Yes, and this is the part most programmes cannot do.

Because the alert, the action taken and the student’s subsequent attendance, assessment and progression sit on one record, outcomes for students who received an intervention can be compared with those who did not, which turns a retention strategy into something you can direct.

SEE IT ON YOUR OWN COHORT

Bring one cohort and last year’s withdrawals. We will show you when they were visible.

Using your attendance, assessment and finance signals and your own advising structure, not a canned demo. Most institutions are surprised how early the pattern appears.

Request a demo Talk to us