school_compare/docs/superpowers/specs/2026-07-02-ux-audit-report.md

# SchoolCompare UX/UI Audit — 2026-07-02

## Method summary

Journey-led walk-through of the **live site** (schoolcompare.co.uk) via Playwright browser tools, at two viewports: **390×844 mobile (primary — 56% of traffic)** and **1440×900 desktop** (720×450 used as the 200%-zoom reflow proxy). Five journeys in traffic order — home→find-school (63% of entries), cold landing on a school page (SEO long tail), building a comparison (27% of views), rankings→shortlist (12%), admissions (5%, light pass) — followed by a cross-cutting cohesion pass comparing computed typography, colour, and component styles across all six page types, verified against `nextjs-app` source to distinguish "token exists but bypassed" from "no token exists".

Every page state visited was scanned with **axe-core 4.10.2** (WCAG 2.2 A/AA rule set) plus manual checks: touch targets ≥24px (2.5.8), keyboard order and focus visibility, and 200% reflow. A finding is included only if it cites a Nielsen heuristic, a WCAG 2.2 AA failure, a mobile-usability standard, or an observed task-flow obstruction — no taste-only findings. Priorities are traffic-weighted using the 30-day analytics baseline: entries `/` 63% / `/compare` 20% / `/rankings` 6%; exits `/` 46% / `/compare` 32% / `/rankings` 13%; 56% mobile. Uplift bands are honest small/moderate/large indications — there is no funnel instrumentation to support percentages.

Full evidence trail: `docs/superpowers/specs/2026-07-02-ux-audit-notes/` (six notes files; screenshots referenced by filename live in the session scratchpad, not committed).

## What works today — keep

These are strengths the fixes below must not regress:

- **Name search is genuinely excellent.** "Welland Primary" → the correct school as sole top result, in 3 taps with instant loads — comfortably under the 15s target. Result cards are information-rich (Ofsted grade+year, RWM % with trend arrow, "+12 pts vs national", pupil count, LA) and adapt correctly to phase (Attainment 8 for secondary, RWM for primary).
- **Full-postcode proximity search works, and works well** *(verified on 2026-07-02 recheck)* — typing a full postcode (e.g. `B91 3DL`) into the home search box yields "13 schools within 1.0 miles of B91 3DL", distance on every card, nearest-first ordering, a 0.5–5 mile radius selector, and a List/Map toggle; the "Schools near me" geolocation button offers the same without typing. Only partial input degrades (see P2.0).
- **Search is the unmistakable primary action** — above the fold on both viewports, 70px tap target, plain-English placeholder, content order puts the task first.
- **The primary (KS2) school page explains its data in plain English** — "End-of-primary-school tests taken by Year 6 pupils", the in-place "Why is combined lower?" explainer, legends on subject charts. This is the model the secondary template should copy.
- **Everything is anchored to the national average** ("+14 pts", "National avg 39.1") — a non-specialist can judge good/bad without leaving the page.
- **Graceful, honest missing-data handling** — COVID-cancellation notes, "contact the admissions authority" for absent cut-offs; no broken or empty UI anywhere data is missing.
- **Ofsted sections are clear, current, and sourced** — grade + inspection date, the post-Sept-2024 grading-change note, link to the real Ofsted report.
- **Compare's empty state is instructive and actionable** — "No schools selected" + "+ Add Schools to Compare" opening an in-page search modal; the 20% who enter cold can start immediately. The modal supports efficient multi-add (stays open, clears field) and accepts postcodes.
- **State persistence is strong across the site** — compare selections survive navigation (localStorage) and are URL-encoded/deep-linkable; rankings restores filter, metric, *and* scroll position on back-navigation (the classic mobile task-killer check, passed cleanly).
- **Compare's metric selector carries a plain-English caption** — jargon defined at point of use; the chart legend labels series by colour *and* name (WCAG 1.4.1 satisfied).
- **Accessibility fundamentals are partly in place** — working skip link, logical tab order everywhere, strong 2px orange focus outlines on links/buttons, real `<table>` semantics on rankings, properly labelled rankings filters, clean heading hierarchy on admissions, no horizontal scroll at 200% reflow on any page tested.
- **Admissions content is well-integrated** — reachable in one tap from anywhere, surfaced proactively on the homepage, with 12 stage-tailored cross-links into Search/Compare/Rankings that read as genuinely contextual.
- **Cohesion anchors:** header/footer pixel-identical on all six pages; coral=primary / teal=secondary held without role collisions; rankings and compare share one phase-tab treatment; the deadline chip is byte-for-byte reused between home and admissions.

## Findings

### P0 — Urgent

- **P0.1 — Colour-contrast failures on every main route, including primary data and primary CTAs** *(merges J1-F4, J2-F4, J3-F6, J4-F1, J5-F1)*
  - Evidence: axe `color-contrast` (serious) on **all five routes**: home 6 nodes (active nav tab, `.btn`, how-it-works text) + 3 on results (`.btn-primary`, Ofsted badge); school template **13 nodes** (Back button, "View on map", active metric tab); compare 5 nodes (`.btn-primary`, active phase/nav tabs); rankings **100 nodes on desktop** — the teal metric values (`.valueCell strong`) in every row plus "+ Compare" links; admissions 13 nodes including the **deadline countdown chips**. Screenshots `j4-rankings-desktop.png`, `j5-admissions-mobile.png`.
  - Criterion: WCAG 2.2 **1.4.3 Contrast (Minimum), AA** — failing on every main page.
  - Argument: this is not chrome-only. On rankings the failing text is the ranking score itself — the entire reason a parent is on the page; on admissions it is the days-until-deadline chip — the page's most load-bearing fact; on home/compare it is the primary CTA styling. The mobile-primary audience reads this outdoors on phones, where low contrast bites everyone, not just low-vision users.
  - Recommendation: adjust the failing token values once (`--accent-coral` on light text pairings, the teal-on-cream value colour, tint-background label colours) and re-run axe on all five routes. This is a token-level fix, not per-page work.
  - Uplift: **accessibility compliance — large improvement** (clears the single biggest violation class, present on 100% of audited pages); secondary small positive effect on all engagement metrics via legibility for the 56% mobile share.

- **P0.2 — School name (H1) is obscured by the map and the "+ Add to Compare" button on mobile — systemic to the school template** *(J2-F1)*
  - Evidence: measured bounding boxes on both audited schools — Castle: map bottom 280px / H1 top 272px / compare-button top 286px / H1 bottom 309px; Welland: same pattern — leaving a ~6px legible sliver of the name. Screenshots `j2-school-mobile-fold.png`, `j2-welland-mobile-full.png`. Desktop is unaffected.
  - Criterion: Nielsen #1 (visibility of system status) and #8; task-flow obstruction of cold-landing orientation; defeats the intent of WCAG 1.4.10 (content obscured at the primary viewport).
  - Argument: school pages are the SEO landing template. A parent arriving from Google asks one question first — "is this the school I searched for?" — and the element that answers it is illegible on the majority viewport. This is the exact first-screen failure that produces an immediate back-to-Google bounce, on the template whose entire job is converting search traffic.
  - Recommendation: restack the mobile hero so map, H1, and compare button occupy non-overlapping space (e.g. compare button below the identity block or docked, map height capped above the H1). One template, both school-detail components (see P2.6).
  - Uplift: **school-page bounce-back-to-Google — decrease, moderate** (fixes the first-glance orientation on every SEO landing; the page content behind it is already strong, so retention should follow).

### P1 — High

- **P1.1 — Filter and Sort selects on home results have no accessible name** *(J1-F3)*
  - Evidence: axe `select-name` (impact **critical**, 2 nodes) on `/?search=…` — the phase filter and sort selects. (Borderline P0 by the "clear WCAG failure on a main page" rule; held at P1 because it affects one page state and two controls.)
  - Criterion: WCAG 2.2 **4.1.2 Name, Role, Value, Level A**.
  - Argument: these are the controls a parent uses to narrow a mixed result set to "Primary" — a screen-reader user hears two anonymous comboboxes on the main journey's results view. Rankings labels its identical selects correctly, so this is an omission, not a pattern.
  - Recommendation: associate visible `<label>`s (or `aria-label`) exactly as `/rankings` already does.
  - Uplift: **accessibility compliance — moderate** (removes the only *critical*-impact violation found); minutes of effort.

- **P1.2 — School pages offer no "nearby / similar schools" path — the compare value proposition has no on-ramp** *(J2-F2, corroborated by J3 tap-count)*
  - Evidence: full-page snapshots of both school pages show no nearby/similar module (sections end Wellbeing → footer). Building a real 2-school comparison from a school page requires Compare → "+ Add School" → **manual name search** — 6 interactions total, and the parent must already know the competitor's name (`j3-compare-two-schools-mobile.png`, `j2-compare-oneschool-mobile.png`).
  - Criterion: Nielsen #7 (flexibility & efficiency); task-flow obstruction — the site's differentiator is unreachable from its SEO entry template without prior knowledge.
  - Argument: a parent cold-landing on one school has maximum comparison intent and zero alternatives on screen. The site sends them back to Google to discover the other local schools — Google then keeps them.
  - Recommendation: add a "Schools nearby" module to the detail template (nearest same-phase schools by lat/long — data already exists) with per-row "+ Compare"; the same list can power an in-modal "nearby" tab on `/compare`.
  - Uplift: **school-page bounce-back-to-Google — decrease, moderate**, and **compare 32% exit — decrease, small-to-moderate** (creates the missing bridge between the two biggest content surfaces).

- **P1.3 — Shared compare links are silently overridden by the recipient's own localStorage** *(J3-F4)*
  - Evidence: navigating to `/compare?urns=142161,113105` with a different stored selection resolved the page and URL back to the stored set; the shared schools never appeared. Reproduced; only clearing localStorage restored the link's intent.
  - Criterion: Nielsen #1 (visibility of system status); task-flow obstruction of an advertised primary feature (the Share button).
  - Argument: 20% of sessions *enter* on `/compare`, many via shared links. A parent shares a shortlist with a partner; if the partner has ever used compare, they silently see their own old schools with no notice — the collaboration breaks and neither party knows why.
  - Recommendation: explicit URL params must win over localStorage (persist *after* honouring the URL); optionally prompt "Replace your saved comparison with the shared one?".
  - Uplift: **compare 32% exit rate — decrease, small-to-moderate** (repairs a confusing dead end at a high-intent entry point; band limited by the shared-link share of entries being unmeasured).

- **P1.4 — Trend/results charts expose no text alternative to assistive tech** *(merges J2-F5, J3-F3)*
  - Evidence: axe `role-img-alt` (serious) on the `<canvas>` on both school pages and populated `/compare`; `aria-label` is null at both viewports.
  - Criterion: WCAG 2.2 **1.1.1 Non-text Content, Level A** — on the two main content templates.
  - Argument: the site's key visualisation — performance over time — announces nothing to a screen-reader parent. Compare's data table partially mitigates; the school-page trend chart has only a collapsed raw-data expander.
  - Recommendation: generate a descriptive `aria-label` per chart ("Attainment 8, 2019–2025: school 53.4 vs national 39.1, school above national every year") — the data is already in the component; one shared chart wrapper fixes all instances.
  - Uplift: **accessibility compliance — moderate** (clears a Level-A class on the highest-traffic templates).

- **P1.5 — Horizontally-scrollable regions are keyboard-inaccessible site-wide** *(merges J1-F5(a), J3-F2, J4-F4)*
  - Evidence: axe `scrollable-region-focusable` (serious) on the home countdown rail, the compare detailed-table wrapper, and (manually confirmed, same pattern) the rankings table wrapper — `overflow-x: auto` with no `tabindex="0"`, no role/label.
  - Criterion: WCAG 2.2 **2.1.1 Keyboard, Level A**.
  - Argument: keyboard-only users literally cannot reach the off-screen columns — which on compare and rankings contain the actual comparison data (see P2.1). Three instances of one missing pattern.
  - Recommendation: `tabindex="0"` + `role="region"` + `aria-label` on every overflow wrapper; one shared wrapper component prevents recurrence.
  - Uplift: **accessibility compliance — moderate**; unlocks the data for keyboard users wherever P2.1's layout still requires scrolling.

- **P1.6 — Search offers no autocomplete/typeahead** *(J1-F1)*
  - Evidence: typing "Welland Primary" (both viewports) produced no suggestions at any keystroke — raw input until submit. The compare modal's live-results search proves the capability exists in the codebase.
  - Criterion: Nielsen #6 (recognition rather than recall); NN-g site-search guidance; error prevention.
  - Argument: parents rarely know a school's exact registered name ("Welland" vs "Welland Church of England…"). With no suggestions, a misspelling means a zero-result dead end on the 63%-of-entries path, and every search costs full typing plus a blind submit.
  - Recommendation: reuse the compare modal's live-search behaviour on the home search box (debounced suggestions: school name + LA, and a "search this postcode" row for postcode-shaped input).
  - Uplift: **home 46% exit rate — decrease, moderate** and **share of sessions reaching a school page — increase, moderate** (assists the majority entry path at its first interaction).

- **P1.7 — The mobile hero omits the value proposition entirely** *(J1-F6)*
  - Evidence: desktop shows the "UPDATED WITH 2026/2027 ADMISSIONS RESULTS" trust badge and the "24,000+ schools… side by side, in one place" subheading; mobile renders only the poetic H1 ("Every school in England, *compared.*") and a bare search box (`j1-home-desktop-fold.png` vs `j1-home-mobile-fold.png`).
  - Criterion: mobile content parity; Nielsen #1 — first-visit orientation ("what is this, why trust it") absent on the primary viewport.
  - Argument: 63% of entries land here and 56% of traffic is mobile; a first-time visitor gets no statement of coverage, data source, or freshness above the fold. Weak value proposition at first glance is a classic bounce driver and plausibly a material slice of the 46% exit rate.
  - Recommendation: restore a compact version of the badge + one-line value prop under the mobile H1 (one text block; the fold has room above the deadline rail).
  - Uplift: **home 46% exit rate — decrease, small-to-moderate** (copy-only change aimed squarely at first-impression exits).

### P2 — Medium

- **P2.0 — Partial or malformed postcode input silently falls back to text search with no distances** *(J1-F2, corrected on 2026-07-02 recheck — the original P0 claim that postcode search lacks proximity was wrong: a full postcode works, see "What works today")*
  - Evidence: the search submit handler routes input through a strict full-postcode regex (`FilterBar.tsx` `isValidPostcode()`). `B91 3DL` → `/?postcode=…&radius=1`: "13 schools within 1.0 miles", distance on every card, nearest-first, radius selector, map toggle (`j1-fullpostcode-results-mobile.png`). But `B91 3` (an outcode+sector a parent plausibly types) → `/?search=B91+3`: text-matched, relevance-ordered, distance-less, mixed-phase results led by a secondary school (`j1-postcode-results-mobile.png`) — with no notice that a different, better mode exists.
  - Criterion: Nielsen #1 (visibility of system status — silent mode switch) and #9 (help users recognise and recover); error tolerance for the "or postcode" promise in the placeholder.
  - Argument: the two result pages look near-identical, so a parent who types "B91" or fat-fingers a character never learns that proximity search exists; they just get an apparently arbitrary list on the highest-traffic path.
  - Recommendation: detect postcode-*shaped* input that fails full validation (outcode/sector patterns, near-miss typos) and either geocode it anyway (postcodes.io supports outcodes) or show an inline nudge — "Enter a full postcode like B91 3DL to see schools by distance."
  - Uplift: **home 46% exit rate — decrease, small-to-moderate** (recovers the postcode-search subset who type partial input; the full-postcode path already works, which caps the band).

- **P2.1 — Mobile data tables hide the decision-critical columns behind unaffordanced horizontal scroll** *(merges J3-F1, J4-F3)*
  - Evidence: compare's "Detailed Comparison" table is 626px inside a 324px wrapper — only Year + the first school's column visible, no scroll cue (`j3-compare-two-schools-mobile.png`); rankings shows RANK/SCHOOL/AREA but the metric % (the ranking's justification) sits off-screen (scrollWidth 551 > 324) (`j4-rankings-mobile.png`).
  - Criterion: mobile-usability standard (primary content hidden off-viewport); Nielsen #6 — the user must memorise school A's numbers while scrolling to school B.
  - Argument: on compare the side-by-side promise is defeated on the majority viewport (a parent may believe the second school's data is missing); on rankings the list shows names without the scores that rank them. Softened by compare's stacked metric cards, which do show both schools — hence P2 not P1.
  - Recommendation: on ≤390px, prioritise the metric column next to the school name (rankings) and use a per-metric stacked layout or sticky first column + visible scroll affordance (compare).
  - Uplift: **compare 32% exit — decrease, small**; **rankings→school click-through — increase, small**.

- **P2.2 — Rankings on mobile removes the View/+Compare actions** *(J4-F2)*
  - Evidence: the Action column `td` computes `display:none` at 390px; shortlisting from rankings is desktop-only.
  - Criterion: task-flow obstruction (the journey is rankings → *shortlist*); Nielsen #7.
  - Argument: a parent on a phone scanning the good local schools cannot build a comparison set from the list — each school needs a page visit and a different route in, on the majority device class.
  - Recommendation: keep a compact "+" compare affordance per row on mobile (the bottom-tab badge already communicates basket state).
  - Uplift: **rankings→school click-through / compare usage — increase, small-to-moderate** (restores the journey's intended endpoint on mobile).

- **P2.3 — Metric jargon unexplained at the point of first contact** *(merges J2-F3, J4-F5)*
  - Evidence: secondary school pages show "Attainment 8 score 53.4", "EBacc average point score 4.72" with zero definitions/tooltips (the plain-English definition exists — but only on `/compare`); rankings *defaults* to "higher standard" (greater depth) without explaining it, so the top school's headline reads "63.0%" in the terms parents least understand (`j2-school-mobile-gcse.png`).
  - Criterion: Nielsen #2 and #10.
  - Argument: a non-specialist cannot judge whether 53.4 or 4.72 is good; a parent may misread 63% at higher standard as mediocre when it is exceptional. The site owns the explanations and withholds them at first contact. The primary KS2 page proves in-place explanation works.
  - Recommendation: surface the existing `/compare` metric definitions on school pages (caption or expander beside each figure); default rankings to "expected standard" with "higher standard" as an explained option.
  - Uplift: **school-page bounce-back-to-Google — decrease, small-to-moderate** (comprehension is the retention lever for the secondary template).

- **P2.4 — Add-school modal lacks dialog semantics and drops focus** *(J3-F5)*
  - Evidence: no `role="dialog"`/`aria-modal`; after a keyboard add, focus lands on `document.body`; after Escape-close, focus is on `body`, not the trigger.
  - Criterion: WCAG 2.2 **4.1.2 (A)** + **2.4.3 Focus Order (A)**.
  - Argument: screen-reader users aren't told a dialog opened; keyboard users re-tab from the page top after every add — compounding friction on the flow every comparison depends on.
  - Recommendation: dialog role + focus trap + focus restoration to the trigger on close; keep focus in the search field after an add (the multi-add UX is otherwise good).
  - Uplift: **accessibility compliance — moderate**.

- **P2.5 — School-page sticky section nav clips its last items on mobile** *(J2-F7)*
  - Evidence: at 390px, "History" and "Wellbeing" render off-screen; the 448px nav row is clipped, not scrollable (document scrollWidth stays 390) (`j2-school-mobile-gcse.png`).
  - Criterion: mobile usability — interactive content unreachable in-viewport; Nielsen #7.
  - Argument: 2 of 6 jump links are dead weight on the primary viewport; parents hunting SEN/Wellbeing data must scroll-hunt instead.
  - Recommendation: make the nav row horizontally scrollable with an overflow affordance (fade/chevron), or wrap to two lines.
  - Uplift: **school-page bounce-back-to-Google — decrease, small**.

- **P2.6 — Two parallel school-detail components have drifted; a duplicate `.btn` rule makes button geometry unpredictable** *(merges cohesion F1, F2)*
  - Evidence: `SchoolDetailView` vs `SecondarySchoolDetailView` hardcode divergent values for the same controls (`.btnAdd` radius 8/pad 12×20 vs radius 6/pad 8×16; badge radii 4 vs 3; tab padding differs). Separately `globals.css` defines `.btn` **twice** (lines 151 and 1558) with `.btn-sm` declared between them — source order makes the second definition override `.btn-sm`, so "small" buttons render at 12×24px padding (live-verified on rankings).
  - Criterion: Nielsen #4 (consistency & standards); maintenance hazard.
  - Argument: parents moving between primary and secondary school pages (the compare flow mixes phases) meet the same controls at subtly different sizes/corners; and any future button edit has a 50/50 chance of landing in the dead rule.
  - Recommendation: dedupe `.btn` to one definition (restoring `.btn-sm`); converge both school-detail modules on shared tokens (`--radius-md`), ideally one shared component.
  - Uplift: **school-page bounce-back-to-Google — decrease, small** (visual consistency is a trust signal on the template that must convert cold traffic); also removes a recurring source of future drift.

- **P2.7 — Design tokens exist but are bypassed system-wide** *(merges cohesion F3, F4, F5)*
  - Evidence: **15 distinct raw radius values** in module CSS against a 4-token scale (no pill token for the 21× `999px` uses); **four different H1 sizes** across five pages (36/44/48/52px), each a separate hardcoded `clamp()`; the "switch view/phase" control has **three different treatments** (rounded segmented on home, square tabs on rankings/compare, underline strip on admissions).
  - Criterion: Nielsen #4.
  - Argument: corner rounding, title scale, and segmented controls are the primary "one system" cues; the drift reads as many hands and erodes the data-source credibility the site trades on. (Rankings↔compare already share one tab treatment — the model to extend.)
  - Recommendation: token-enforcement pass (map raw radii to tokens; add `--radius-pill`), create hero/section title tokens, consolidate the segmented control on the rankings/compare pattern.
  - Uplift: **home 46% exit rate — decrease, small** (a visibly coherent system reads as a credible data source at first glance); primarily protects future velocity.

- **P2.8 — Two admissions cross-link CTA variants are below the 24px touch-target minimum** *(J5-F2)*
  - Evidence: "Build and compare your shortlist →" and "Look up your allocated school →" measure **21px** tall vs 42px for the other four CTA variants on the same page (getBoundingClientRect).
  - Criterion: WCAG 2.2 **2.5.8 Target Size (Minimum), AA** (rendered as standalone CTAs, so the inline-text exception is doubtful).
  - Argument: these are the conversion path the admissions content exists to support; a 3px-under-spec target inconsistent with siblings adds mis-tap friction for a stressed parent. (Admissions-only, but a WCAG failure, so it stays P2 per the prioritisation rules.)
  - Recommendation: apply the 42px CTA style to all twelve cross-links.
  - Uplift: **accessibility compliance — small**.

### P3 — Nice-to-have

- **P3.1 — Mixed-phase selection reports "Comparing 2 schools" while rendering one** *(J3-F7)* — the second school hides behind a "Secondary (1)" tab with no explanation (Nielsen #1/#2). Add one line of copy ("Shown separately — KS2 and GCSE metrics differ"). Uplift: compare exit, small.
- **P3.2 — Compare remove (×) is instant with no undo** *(J3-F8)* — accidental tap on the 28px target silently loses a searched-for school (Nielsen #3). Add a brief undo toast. Uplift: compare exit, small.
- **P3.3 — "Add to Compare" gives no toast/live-region confirmation** *(J2-F8)* — feedback is only a button relabel in the (currently overlapped) hero zone plus a small tab badge; no `role=status` fires (Nielsen #1). Largely mitigated once P0.2 lands. Uplift: compare usage, small.
- **P3.4 — Inconsistent focus indicators on form controls** *(merges J1 search-input note, J4-F8)* — filter selects suppress the outline for a faint 1.5px border hue shift; the home search input uses a 0.12-alpha ring — both far weaker than the site's standard 2px orange outline (WCAG 2.4.11 borderline). Apply the standard outline. Uplift: accessibility compliance, small.
- **P3.5 — Rankings table headers aren't sortable and lack `scope`** *(J4-F7)* — users expect ranked tables to sort on header click (Nielsen #4); `<th>` cells lack `scope`/`aria-sort` (WCAG 1.3.1 best practice). Uplift: rankings→school CTR, small.
- **P3.6 — 33-option metric dropdown is a flat jargon list** *(J4-F6)* — GPS/EAL/SEN abbreviations, ungrouped (Nielsen #6). Group by theme (Results / Progress / Cohort) with plain-language labels. Uplift: rankings→school CTR, small.
- **P3.7 — Home deadline rail defaults its scroll to the least-urgent card** *(J1-F5(b))* — the nearest deadline (121 days) hides behind less urgent ones (Nielsen #1). Order/scroll to soonest-first. Uplift: home exit, small.
- **P3.8 — School trend chart has no legend for the national series; badge overlaps a heading** *(J2-F6)* — the grey dashed line is unexplained (the primary page's charts *do* carry legends — inconsistent); "Nat avg 39.1" overlaps the section heading. Uplift: school-page bounce, small.
- **P3.9 — Residual colour/type token drift** *(cohesion F6, F7, F8)* — pressed-coral exists as three hexes (`#e07256`/`#c45a3f`/`#d4654a`), an off-token gold `#b8920e` on the school-page cross-link badge, H3s render in body font while H2s are Playfair, and school-detail tables use an 11/13px type scale vs 12/15px on rankings/compare. Fold into the P2.7 token pass. Uplift: school-page bounce-back-to-Google, decrease, small.

## Accessibility summary

Axe-core 4.10.2, WCAG 2.2 A/AA rule set. Desktop serves the same DOM/CSS, so mobile results apply at both viewports unless noted.

| Page / state | Viewport | Violations | Detail |
|---|---|---|---|
| `/` (initial) | 390×844 | 2 | `color-contrast` (serious, 6 nodes); `scrollable-region-focusable` (serious, 1) |
| `/?search=Welland+Primary` | 390×844 | 2 | `color-contrast` (serious, 3); **`select-name` (critical, 2)** |
| `/school/…castle` (secondary) | both | 2 | `color-contrast` (serious, 13); `role-img-alt` (serious, 1) |
| `/school/…welland` (primary) | 390×844 | 1 | `role-img-alt` (serious, 1) |
| `/compare` (empty) | 390×844 | 1 | `color-contrast` (serious, 2) |
| `/compare?urns=…` (2 schools) | 390×844 | 3 | `color-contrast` (serious, 5); `role-img-alt` (serious, 1); `scrollable-region-focusable` (serious, 1) |
| `/rankings` (default & filtered) | 390×844 | 1 | `color-contrast` (serious, 2) |
| `/rankings` | 1440×900 | 1 | `color-contrast` (serious, **100 nodes** — metric values in every row) |
| `/admissions` | 390×844 | 1 | `color-contrast` (serious, 13 — incl. deadline chips) |

**Manual checks:** Touch targets ≥24px — pass on all core paths (smallest: compare × at 28px, Back at 64×28) **except** two admissions CTAs at 21px (P2.8). Keyboard — order logical everywhere with a strong 2px outline on links/buttons; gaps: unlabelled home-results selects (P1.1), modal focus loss (P2.4), suppressed outlines on filter selects and the faint search-input ring (P3.4), keyboard-unreachable scroll regions (P1.5). 200% reflow — **pass on every page tested** (no horizontal scroll at 720px; tables retain contained scroll, a recognised 1.4.10 exception).

**Overall WCAG 2.2 AA posture:** not currently conformant. Four violation classes account for everything axe found — contrast (every route), missing chart text alternatives (Level A), keyboard-inaccessible scroll regions (Level A), unlabelled selects (Level A, critical) — plus manual findings on target size, focus appearance, and dialog semantics. All are pattern-level fixes; none require redesign. The foundations (skip links, tab order, semantics, reflow, labels on rankings) are solid, which is why the fix list is short relative to the violation node counts.

## Suggested implementation sequence

Batches ordered by uplift-per-effort; each is a plausible standalone follow-up project.

**Batch 1 — WCAG compliance sweep (small effort, large compliance uplift).**
P0.1 contrast token fixes; P1.1 label the two selects; P1.4 chart `aria-label`s via one shared wrapper; P1.5 `tabindex`/role on the three scroll wrappers; P2.4 dialog semantics + focus restoration; P2.8 42px CTAs on admissions; P3.4 standard focus outlines. Mostly CSS-variable and attribute changes; re-run the axe harness on all five routes as the acceptance check. Clears every axe violation class found.

**Batch 2 — School-template mobile fold (small effort, direct SEO-retention uplift).**
P0.2 un-overlap the H1/map/compare button; P2.5 scrollable section nav; P3.8 trend-chart legend + badge overlap; P3.3 add-confirmation live region. One template, all SEO landings.

**Batch 3 — Home search assistance (small-moderate effort, home-funnel lever).**
P1.6 autocomplete on the home search (reuse the compare modal's live search); P2.0 graceful handling of partial/malformed postcodes (geocode outcodes or nudge toward a full postcode — the full-postcode proximity search already works); P1.7 mobile value-prop line. Together these target the 46% home exit rate from three directions.

**Batch 4 — Discovery and sharing bridges (moderate effort, compare-funnel uplift).**
P1.2 "Schools nearby" module on school pages (feeds compare); P1.3 URL-over-localStorage precedence for shared links; P2.2 mobile +Compare on rankings rows; P3.1/P3.2 compare state-clarity nits.

**Batch 5 — Comprehension pass (small-moderate effort).**
P2.3 reuse `/compare` metric definitions on school pages and default rankings to "expected standard"; P2.1 mobile table layouts (sticky first column / stacked metrics); P3.5/P3.6 rankings sort affordance + grouped metric picker; P3.7 deadline-rail ordering.

**Batch 6 — Cohesion & token enforcement (housekeeping, protects velocity).**
P2.6 dedupe `.btn`, converge the two school-detail components; P2.7 radius/title/segmented-control tokenisation; P3.9 residual colour/type drift. Best done after Batches 1–2 so the new token values land once.