# Study-room cancellation: a decision brief

Original fictional practice case by InternshipDeadlines. No real research, product release or outcome is claimed.

## Decision
Investigate whether a clearer cancellation path can release unused bookings before building a waitlist. Run a bounded test only after confirming that difficult cancellation is a material cause of unused room time.

## User and problem
A student who no longer needs a room wants to release it quickly. Another student wants to find an actually available room. Staff report empty rooms with active reservations. These reports are starting hypotheses, not measured findings.

## Evidence needed before committing
- Check a defined sample of bookings against observed use, using an authorized method. Record the period and missing observations.
- Review the existing cancellation path and support requests.
- Distinguish forgotten cancellations from inaccurate booking records, unauthorized occupancy and last-minute schedule changes.
- Ask whether released slots become available promptly in the present system.

## Alternatives
| Option | Mechanism | Main uncertainty | Cost or drawback |
|---|---|---|---|
| Reminder | Helps people remember to release a room | Is forgetting a major cause? | Extra messages; may be ignored |
| Clearer cancellation | Reduces friction in releasing a reservation | Are people willing to cancel when it is easy? | Requires correct status handling |
| Waitlist | Reallocates released or unused capacity | Does capacity become available reliably? | Additional allocation and notification rules |
| No change yet | Avoids implementing the wrong explanation | Can we obtain better evidence soon? | Current frustration continues |

## First scope
Improve discovery and confirmation of cancellation for a user's own reservation. Preserve existing building-access and booking rules. No automatic penalty, new priority system or cross-campus redesign in this version.

## Acceptance criteria
1. An authenticated authorized user can cancel their own eligible reservation.
2. Cancellation releases the slot once. Repeated requests leave a consistent cancelled state.
3. The interface confirms success only after the service confirms the change.
4. Failed requests show a recoverable error and do not falsely say the room was released.
5. A user cannot cancel another person's booking through this path.
6. Support can investigate the change through an approved history record, without unnecessary personal-data collection.

## Measurement
Primary outcome to investigate: unused booked room time in a defined eligible cohort and period.
Diagnostic task measure: successful cancellation completion among eligible attempts.
Checks: accidental cancellations, support requests, delayed availability updates and repeat-event logging.

A synthetic task test has 40/200 completions in A and 50/200 in B. The observed rates are 20% and 25%: a five-percentage-point difference and a 25% relative increase against A. This does not establish a population effect, causality or improved room use. Check assignment, repeated users and logging; estimate uncertainty with an appropriate method before making a deployment claim.

## Review and rollout decision
Have engineering review the state transitions and failure handling, and have the service owner confirm the booking rules. Start with an authorized small test and a rollback path. Expand only when the outcome and failure measures support doing so. If cancellation is already easy or the occupancy records are wrong, change the recommendation rather than adding features.

## Questions for your practice reviewer
- What evidence would make a reminder the better first option?
- What happens when the user retries during a service failure?
- Is cancellation-task completion the outcome we care about?
- Which requirement could be simplified without weakening correctness?

## Your contribution
Create a new case or revise an assumption and document your reasoning. Credit this starting example. Do not present its hypothetical research or test data as work you performed for a real organization.