School trip administration workflow

The problem

Organising a school trip involves a lot of administration: sending consent forms, collecting responses, updating spreadsheets and chasing missing replies. Office staff can end up entering the same information several times and checking different sources before the trip leader has what they need.

I wanted to reduce the staff time spent on repetitive tasks, so it could go on parents’ questions, checking important information and decisions that need human judgement.

Redesigning the process first

I mapped the existing process with an As-Is flow diagram to find where information was duplicated and where staff re-entered data. Then I considered which stages could be eliminated, simplified or automated. The principle: improve the process before automating it.

I proposed a single consent form, a central SharePoint list of pupil records and an automated workflow to update it. The workflow would also identify information needing particular attention. The aim was not to remove people from the process, but to cut routine administration while keeping human oversight where it matters.

Choosing the technology

I considered Make and n8n. Power Automate suited this scenario because the school was assumed to use Microsoft 365 already, and it connects directly to Forms, SharePoint, Outlook and Teams. Keeping data in Microsoft 365 doesn’t make a system compliant with data protection law, though; permissions, sharing, retention and security would still need review before deployment.

The workflow

An automated cloud flow starts when a parent or guardian submits a Microsoft Form (consent, contact email, medical or dietary information, questions). The demonstration used entirely fictional data and pupil reference numbers rather than names.

1. Forms trigger. A new response starts the flow.
2. Get response details. Retrieves the submitted answers.
3. Get items (SharePoint). Finds the pupil record using the reference in the form.
4. Apply to each. Get items returns an array, so results are processed in a loop. Normally one record matches, so it runs once.
5. Update item (SharePoint). Updates the existing record, with no duplicate. It maps consent, parent email, medical or dietary notes and trip questions, replacing manual transfer.
6. Send an email (Outlook). Sends a confirmation after the update, with no individual staff email.
7. Condition. If the medical or dietary field is completed, the flow posts a Teams notification (via the Flow bot) to the administrator in the demonstration. If empty, nothing more happens. The information is already saved; the notification exists to draw attention to something that may need human review.

Routine responses are processed automatically, and potentially significant information is flagged for staff.

Data protection and safeguarding

Even with fictional data, a real school system would process personal data about children and parents, and medical information may be special category health data under UK data protection law. I treated this as part of the design, not an afterthought.

• Pupil identification: reference numbers reduce the exposure of names, but aren’t anonymous if the school can link them to pupils. A live system would need a reliable way to confirm the respondent was authorised to give consent.
• Access: SharePoint permissions must restrict records, especially medical information, to authorised staff. I’d also review who can see run history, form responses and Teams notifications. The prototype’s Teams notification includes the medical or dietary text; in a live system it should point authorised staff to the restricted record instead of repeating sensitive details in a chat.
• Safeguarding: the Department for Education’s Keeping Children Safe in Education 2026 sets out schools’ safeguarding responsibilities and links the safe use of AI to them. Automation must not delay or replace the school’s procedures; concerns go promptly to the Designated Safeguarding Lead.
• Assessment: before a live deployment I’d involve the school’s Data Protection Officer and review lawful basis, health data, access, retention and whether a DPIA is needed.

Testing

I used the prototype’s testing tools and run history to demonstrate information moving from the form into the SharePoint list and the connected actions. More systematic testing would be needed:

• consent mapped to the correct pupil record;
• missing, incorrect or duplicate pupil references;
• confirmation emails sent only after a successful update;
• responses with and without medical or dietary information;
• a Microsoft 365 service being temporarily unavailable;
• sensitive information accessible only to authorised staff.

This would show whether it’s reliable enough for a real working environment, rather than only that it runs.

Where AI could add value (proposed, not implemented)

The prototype is rules-based, which suits structured consent data. Parents also send free-text questions about equipment, timings or arrangements. I propose an extension where AI helps classify routine enquiries and prepare draft replies from approved school information, and a member of staff checks, edits and approves every reply before it’s sent. AI could help prepare a response, but a person stays responsible for its accuracy.

Applying the DfE’s generative AI guidance:

• Rules for forms, AI for free text. Structured consent stays rules-based.
• Protect personal information. Avoid sending identifiable pupil or medical details to a generative AI service wherever possible; the school would approve the tool, review its data processing and consider a DPIA.
• Retain human judgement. Staff review every draft.
• Safeguarding and medical matters. These follow established routes to authorised staff. AI would not assess a safeguarding concern, decide about a pupil’s needs or reply to sensitive enquiries, and the process would need safeguards against sensitive cases being missed or misclassified.

Automation handles predictable tasks, AI may assist with less structured information, and people remain responsible for decisions.

Future development

Reliable consent and identity verification; populating the pupil list from an approved MIS export, or a secure API integration; automatic reminders for outstanding consent (part of the redesign, not demonstrated); error handling for missing or duplicate references, failed updates and undelivered email; an information governance review; AI evaluation on fictional examples, including ambiguous and safeguarding-related cases, before any AI is introduced.

What I learned

Understand the process before choosing the tool; the useful starting point was where staff repeated tasks, not Power Automate. I gained practical experience with triggers, actions, dynamic content, SharePoint updates, Apply to each loops and conditions. Decisions about automation and AI can’t be separated from the environment a system operates in.

Good automation isn’t about automating everything. It’s about removing unnecessary work while keeping people involved where their judgement is needed.

References

• DfE (2026), Keeping Children Safe in Education 2026.
• DfE, Generative Artificial Intelligence (AI) in Education.
• DfE (updated October 2026), Generative AI and Data Protection in Schools.
• DfE (2026), Generative AI: Product Safety Standards.

← Back to Work