# Experiment Assignment

## Objective

Put each user in one experiment bucket and keep them there.

Deterministic, stable assignment of users to experiment variants, with eligibility rules and protection against overlapping tests.

## Before You Begin

This feature is being added to an application that already exists and already
works. Do not scaffold a new project, and do not assume a blank slate.

Inspect the codebase first and establish:

- The existing application structure and where code of this kind already lives.
- The framework and version in use.
- The existing design system — colours, spacing, typography, and component conventions.
- Existing UI components you can reuse instead of writing new ones.
- The existing database structure, if this feature needs to persist anything.
- The existing authentication and authorization system, if this feature is user-scoped.
- Dependencies already installed, so you don't add a library that duplicates one.
- The existing test setup and conventions.

Only start writing code once you understand the above. If the application
already implements part of this feature, extend it rather than replacing it.

## Implementation Instructions

1. Assign by hashing a stable identity key together with the experiment key. Assignment must be computed, not stored and looked up, so the same user always lands in the same bucket.
2. Pick the identity key deliberately: an account id where one exists, a durable anonymous id otherwise. Never assign on session id — a new session becomes a new user.
3. Define eligibility as an explicit filter applied before assignment: exclude staff, known bots, and plans the variant cannot apply to.
4. Support mutual exclusion groups so two experiments touching the same surface cannot both run on one user.
5. Do NOT build a flag store, admin toggle UI, server-side evaluation, caching, or sticky percentage rollout. All of that is owned by Feature Flags — assignment is a layer on top of it, not a replacement.

## UI and UX Requirements

Match the application's existing design system exactly. Reuse its components,
spacing, and typography. This feature should look like it was always there.

## Responsive Requirements

Works on mobile, tablet, and desktop. Touch targets are large enough to hit on a
phone, and nothing overflows horizontally at 320px.

## Accessibility Requirements

- Fully keyboard navigable.
- Correct semantic elements and ARIA roles.
- Visible focus states.
- Meets WCAG AA contrast.
- Dynamic changes are announced to screen readers.
- Respects prefers-reduced-motion.

## Edge Cases

- An anonymous user who signs in must not flip variants mid-journey. Decide the rule up front — carry the anonymous assignment onto the account, or exclude pre-login exposure from the analysis — and apply it consistently.
- A user who signs in on a second device will hash differently unless the account id takes over. Reassignment must be logged so results can account for it.
- Excluded users must get the control experience without an exposure event. Recording exposure for someone who was never eligible poisons the readout.
- Changing the traffic split mid-experiment reshuffles buckets. Either forbid it or restart the experiment; do not quietly rebalance live users.
- Bots and preview crawlers hitting the site at scale will skew allocation. Filter them before assignment, not after.
- Assignment must be cheap and must never block a page render. If the identity key is unavailable, serve control rather than waiting.

## Testing

Exercise the feature end to end in the running application. Cover every edge case
above, then run the existing test suite and confirm nothing regressed.

## Acceptance Criteria

- [ ] The same identity key and experiment key always produce the same variant.
- [ ] Assignment uses a durable identity, never a session identifier.
- [ ] Staff, bots, and ineligible plans are excluded before assignment, with no exposure recorded.
- [ ] The anonymous-to-authenticated transition follows one documented rule and is logged when it changes a variant.
- [ ] Mutually exclusive experiments never both apply to the same user.
- [ ] Assignment builds on the existing feature flag system rather than a parallel one.
- [ ] The feature matches the existing design system.
- [ ] No existing functionality is broken.

## Adaptation Rules

- Match the existing design system. Do not introduce a new colour palette,
  spacing scale, or component library.
- Reuse existing components and utilities wherever they fit.
- Follow the naming, file layout, and code style already present.
- Do not upgrade, replace, or remove existing dependencies to make this
  feature fit. Adapt the feature to the app, not the app to the feature.
- Do not break existing functionality. If a change is genuinely required in
  existing code, make the smallest one that works and say so.
- If something in these instructions conflicts with how the application is
  built, follow the application and explain the deviation.

## Final Verification

Before you report the work as done:

1. Re-read the acceptance criteria above and check each one against what you
   actually built.
2. Run the application and exercise the feature end to end.
3. Run the existing test suite and confirm you have broken nothing.
4. Check the feature on mobile, tablet, and desktop widths.
5. Check keyboard navigation and focus handling.
6. Summarize what changed: files added, files modified, and anything you
   deliberately did differently because of how this application is built.

If any acceptance criterion is unmet, fix it before reporting completion.