SEAtS Engage

The Engage score is a weighted composite of data from multiple sources. Each source you connect adds depth — but each requires clean, consistent data. A low-quality source reduces score reliability; a high-quality source significantly improves the model's predictive power.

• Available sources: SEAtS attendance, VLE logins (Moodle, Canvas, Blackboard, Teams), library entry swipes, digital resource access, submission records, campus Wi-Fi, card door access
• Enable a source in Engage → Config → Data Sources — it begins contributing from the next weekly run
• Each source shows a data quality score (0–100%) — sources below 80% are contributing noise; investigate before enabling
• Disabling a source doesn't change historical scores — it only affects future calculations
• Attendance is typically the highest-weight source as it is the most reliable and consistent

Engage scores can trigger automated workflows independently of attendance alerts. A student whose score drops below a threshold can trigger a tutor notification, open a welfare case, or send a check-in message — all without manual intervention, transforming Engage from a reporting tool into an active early-warning system.

• Configure Engage-triggered workflows in Engage → Config → Alert Rules → New Rule
• Trigger options: score below threshold for the first time, below threshold for N consecutive weeks, or drops more than X points in a single week
• Actions: email the student, notify their tutor, create a CRM case, or trigger a full workflow sequence
• Combine Engage triggers with attendance for compound rules: e.g. score below 300 AND attendance below 75% triggers a priority welfare referral
• Preview impact before saving — the system shows how many students would currently be in scope

Engage data becomes most powerful when directly connected to CRM casework. A student flagged by an Engage alert who already has an open welfare case needs a different response than one appearing for the first time. SEAtS integrates both views in the student profile — the Engage score trend appears alongside case history.

• The student profile shows the Engage score trend graph alongside the CRM case list — no switching between screens
• When an Engage alert fires on a student with an open case, it's added to the existing case timeline rather than creating a duplicate
• Case workers can reference the Engage score in case notes — providing objective evidence for referral paperwork
• The "Engage + No Case" CRM view surfaces students with significant score drops who have no open support case
• After an intervention, monitor whether the score recovers over 4–6 weeks as a measurable outcome indicator

A single weekly score is less useful than a trend. Three consecutive weeks of decline is a meaningful signal; a single week of low score followed by recovery is often noise. Distinguishing genuine disengagement from one-off anomalies (exam week, student holiday, system data gap) is a key practitioner skill.

• The score trend graph in the student profile shows weekly scores for the academic year — hover over each point for the exact value and date
• A consistent downward trend over four or more weeks is a strong signal; a single anomalous week is not
• If both engagement score and attendance decline together, the signal is stronger than if only one source drops
• A score drop in an entire cohort simultaneously suggests a shared cause — an exam period, a cancelled event, or a data gap
• Peer group comparison shows how a student compares to a filtered subset of peers — useful for contextualising performance within a specific programme

The Engage scoring model is configurable at institutional level. Administrators can adjust source weights, define custom peer groups, and choose how the model handles missing data. These decisions have significant impact on score quality and should be made deliberately.

• Source weights are in Engage → Config → Model Settings; default: attendance 40%, VLE 30%, library/door access 20%, other 10%
• Adjust weights only after reviewing data quality scores — a low-quality source should carry lower weight
• Custom peer groups can include cohort, entry qualification band, or study mode in addition to programme and year
• Missing data handling: "exclude from calculation" reduces that week's score weight; "impute from prior weeks" carries forward the previous value
• Document every model configuration change — configuration history is not stored in the audit log by default

Engage AI extends the weekly score model with a predictive layer identifying students statistically likely to disengage in coming weeks — before their score has fallen to a standard threshold. The model is trained on your institution's historical data and improves in accuracy over each academic year.

• Engage AI generates a "risk trajectory" for each student: Rising, Stable, Declining, or At Risk
• At Risk fires based on current score, rate of decline, historical patterns, and data source signals combined
• AI predictions appear as a trajectory banner in the student profile and as a filterable column in the dashboard
• Model accuracy metrics are in Engage → AI → Model Performance — review quarterly
• New students in weeks 1–8 show "Insufficient data" — the model requires 8 weeks of score history minimum

Engage aggregates data from multiple external systems. Each integration has its own technical characteristics, delivery schedule, and failure mode. Managing the pipeline — monitoring for failures and co-ordinating with source system owners — is an ongoing operational responsibility.

• VLE integrations use API polling (Moodle, Canvas, Teams) or nightly SFTP (Blackboard, legacy systems)
• Library and door access data typically arrives via flat-file SFTP from your access control or library system overnight
• Each source has a data freshness indicator in Engage → Config → Data Sources — "Stale (>48hrs)" needs immediate investigation
• Missing data for one week is handled gracefully; missing for three or more consecutive weeks degrades score quality significantly
• Maintain a contact list of the technical owner for each source system — when a feed fails overnight, you need to know who to call

Engage data raises important questions about student privacy, consent, and the ethical use of behavioural data. Institutions deploying Engage have a responsibility to be transparent with students about what is collected, how it is used, and what rights they have.

• Inform students about Engage data collection via your institution's Privacy Notice — confirm with your DPO that the notice covers all active sources
• Students have the right to access their own engagement data under UK GDPR — the platform supports this via the student portal and staff-assisted SAR exports
• Configure data retention periods in Admin → Data → Retention Policy in line with your institutional student data retention schedule
• Engagement data should never be used in academic decision-making without explicit institutional policy approval and student disclosure
• Governance reports showing aggregate trends and intervention outcomes are generated from Engage → Reports → Governance Summary