3. Key screens

Five screens carry the product. Each is specified by purpose, primary user, layout, core interactions, states, and acceptance criteria. Supporting screens are listed at the end.

3.1 Methodology Studio

Purpose. Encode the firm's diagnostic once, with enough rigor that any engagement team produces the same score and the same interpretation from the same data.

Primary user. Practice Lead (author), Partner (approver), Senior Analyst (item work).

Layout.

┌ Pack header: name · version 3.2 · state [Draft] · owner · reviewers · [Request review] ┐
├───────────────┬──────────────────────────────────────────────┬────────────────────────┤
│ Framework     │ Canvas                                       │ Inspector              │
│ tree          │ Outcome/Dimension columns → Practice cards   │ Selected object        │
│ ▸ Direction   │ → Item chips. Weights shown on edges.        │ - text, scale, reverse │
│   ▸ Vision    │ Coverage warnings inline:                    │ - weight, required     │
│   ▸ Strategy  │ "Practice 'Role clarity' has 1 item.         │ - construct tags       │
│ ▸ Leadership  │  Minimum is 2 for a reportable score."       │ - reading level (FK)   │
│ ...           │                                              │ - translations         │
│               │                                              │ - validation status    │
├───────────────┴──────────────────────────────────────────────┴────────────────────────┤
│ Tabs: Framework · Items · Scoring · Bands & Interpretations · Recommendations ·        │
│       Open-ends & Video · Interview guide · Forms (census/short/frontline) · Preview   │
└───────────────────────────────────────────────────────────────────────────────────────┘

Core interactions.

• Build the hierarchy: Outcome → Dimension → Practice → Item. Drag items from the library; forked items keep lineage.
• Set scoring per level: weights, reverse coding, response scale, minimum items answered, required-for-index flags.
• Define band scheme: absolute cut points or norm-relative (quartiles against a named norm version).
• Write interpretations per band per dimension/practice. AI can draft; drafts are visibly marked and cannot be locked until a named partner approves each one.
• Map recommendations to (practice, band) pairs and to service-catalog offerings.
• Generate forms: census, executive short form, frontline. Studio shows estimated time per form and which practices lose reportability in each.
• Generate interview guides (exec / middle / frontline) from the same constructs, with the coding scheme attached.
• Simulate: upload a past dataset or synthetic responses; see scores, bands, reliability (Cronbach's α per practice), and item-total correlations before locking.

States. Draft → In review → Locked for fieldwork → Retired. A locked version is immutable. Edits create a new minor (copy, translations) or major (items, weights, bands) version. Major versions break comparability; Studio says so and lists which items survive.

Acceptance criteria.

• A practice with fewer than 2 items, or an index whose required dimensions are missing from a form, cannot be locked.
• Every band interpretation shows author, approver, and date.
• Version diff renders item, weight, band, interpretation, and recommendation changes side by side.
• Time estimate for each form within ±15% of observed median after 3 engagements.

3.2 Launch Command

Purpose. Get fieldwork live fast and drive completion like an engagement workstream, without breaking anonymity.

Primary user. Engagement Manager. Secondary: Client Admin, Champions.

Layout.

┌ Readiness bar: Population ✓ · Hierarchy ✓ · k-simulation ⚠ 3 cuts · Comms ✓ · Legal ✓ · Translations 7/8 ┐
├──────────────────────────────────────────────────┬──────────────────────────────────────┤
│ Completion map                                   │ Today                                │
│ Hierarchy treemap coloured by response rate.     │ 61% overall (2,562 / 4,200)          │
│ Units below target outlined. Units whose         │ Target 65% by Fri                     │
│ population < k shown rolled into parent.         │ Pace: needs +128/day                 │
│                                                  │ Top risk: Plant C night shift 31%    │
├──────────────────────────────────────────────────┼──────────────────────────────────────┤
│ Breakdowns: channel · language · device ·        │ Comms sequence                       │
│ time-to-complete · drop-off item                 │ Mon sponsor ✓ Tue managers ✓         │
│                                                  │ Wed all-staff ✓ Fri nudge ◷          │
│ Drop-off: item 41 ("decision rights") 6.2% exits │ [Add nudge] [Pause] [Preview]        │
├──────────────────────────────────────────────────┴──────────────────────────────────────┤
│ Champions: named coordinators, units, completion only. No scores. No names of non-responders. │
└──────────────────────────────────────────────────────────────────────────────────────────┘

Core interactions.

• Readiness checklist blocks launch until population hygiene, k-simulation acknowledgement, legal pack, and language review are complete. Each item is one click to fix.
• Cell-size simulator: "At 60% response, 14 of 212 reportable units will fall below k=5. Finance / Band 4 will roll up to Finance." EM can merge cuts before launch.
• Comms sequencer: channel × audience × send time, per language. Previews render in the firm brand.
• Nudge rules target units, not people: "If unit completion < 40% at T+5 days, send reminder to all non-completers in unit and notify champion." Never reveals who has or has not responded to managers or champions.
• Kiosk / QR mode for plants and stores: shared device, per-session token, auto-reset after submit or 3 minutes idle.
• Close controls: extend by unit, close by unit, final 48h warning.

States. Not ready → Ready → Live → Closing (48h) → Closed → Reopened (audited).

Acceptance criteria.

• EM can launch a 4,000-person census within 1 hour of a clean hierarchy file.
• No screen on this surface shows response status for any group smaller than k.
• Drop-off item analysis updates within 15 minutes of live data.

3.3 Diagnostic Room

Purpose. Run the steering-committee conversation. The product is not the dashboard; it is the 45-minute meeting.

Primary user. Partner (presenting). Secondary: Client Sponsor and exec team (viewing, live or after).

Layout. Full-screen, paper background, one idea per view.

┌───────────────────────────────────────────────────────────────────────────────────────┐
│ 2 / 7   Supervisors are the constraint, not the shop floor.                           │
│                                                                                       │
│ [Chart that argues the sentence: practice scores by layer, supervisor row highlighted]│
│                                                                                       │
│ So what: Layer-3 manager practices explain 38% of the explained variance in the index.│
│                                                                                       │
│ Evidence ▸ n=2,562 · k=5 · RWA, bootstrap 95% CI [31%, 44%]                           │
│          ▸ "My supervisor spends the shift on spreadsheets." — Operator, Plant C      │
│          ▸ "We promote the best welder, then never train them." — Supervisor, Plant B │
│          ▸ ▶ 0:42 clip · Team lead, Plant D · consent: shareable                      │
│                                                                                       │
│ Monday move: 90-day supervisor cadence (Pack rec. MGR-B2) → Offering: Frontline       │
│ Leadership Sprint                                                                     │
└───────────────────────────────────────────────────────────────────────────────────────┘
  ← → navigate · E expand evidence · D drill (index → dimension → practice → quote) · N presenter notes

Core interactions.

• Spine navigation: 5–7 findings, then "If we only do three things," then the ask.
• Drill: index → dimension → practice → item → quotes. Motion used only to show the drill path or a wave-over-wave change.
• Live cut: partner can re-cut a chart by any pre-approved k-safe cut during the meeting. Unapproved cuts are unavailable in presenter mode.
• Evidence expand: n, method, CI, quotes, clip, all with provenance.
• Role views: Partner (so-what + commercial next step, notes hidden from client), Sponsor (decisions requested), EM (workplan).
• Capture: decisions and questions logged against the finding; they flow into Actions.

States. Rehearsal (firm only, notes visible) → Live (presenter) → Published (read-only in Client Portal).

Acceptance criteria.

• Only partner-approved Insight Objects appear.
• A partner new to the engagement can read the spine end to end in 8 minutes (tested with timed sessions).
• Every number on screen resolves to a Scorecard ID and cut on click.
• Works offline for the session once loaded (cached, encrypted, expires at session end).

3.4 Client Portal

Purpose. Give the client a durable, branded, safe home for results and actions. Keep the firm present between waves.

Primary user. Client Sponsor. Secondary: exec team, Client Admin, unit managers.

Layout.

┌ <Firm logo> × <Client logo>      Organizational Health Diagnostic · Wave 2 · Published 14 Mar ┐
├──────────────────────────────────────────────────────────────────────────────────────────────┤
│ The answer in one page                                                                       │
│ Health index 58 (▲4 vs Wave 1, significant) · Book median 63 · 3rd quartile                  │
│ 1. ... 2. ... 3. ...   [each expands to evidence]                                            │
├───────────────────────────────┬──────────────────────────────────────────────────────────────┤
│ Scores · Wave compare · Actions · My area · Documents · Share-back kit                        │
└───────────────────────────────┴──────────────────────────────────────────────────────────────┘

Core interactions.

• Read the spine; expand evidence (quotes limited to "client-shareable" permission tier).
• Compare waves on locked items.
• Track actions: owner, due date, status, leading indicator, linked items.
• Managers see "My area" only if their unit clears k. Otherwise: "Your team had fewer than 5 responses, so results are shown for Operations – North (n=48)."
• Download deck, PDF, k-safe appendix.

Never available: row-level responses, unapproved insights, comments for cuts below text-k, cross-client data, vendor branding.

Acceptance criteria.

• Zero vendor strings in DOM, email headers, PDF metadata, or file names.
• Every view passes the anonymity engine at render time, not just at publish time.
• Custom domain with firm-controlled TLS.

3.5 Wave Compare

Purpose. Prove movement. Show what changed, whether it is real, and whether it is where the actions were aimed.

Primary user. EM and Analyst (build), Partner and Sponsor (read).

Layout.

┌ Wave 1 (Mar 2025, n=2,562)  →  Wave 2 (Mar 2026, n=2,781) · Comparable items 58/62 · Hierarchy mapped 97% ┐
├────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ Slope chart: dimensions, W1 → W2. Solid = significant and ≥ 3 pts. Dashed = not significant.           │
│ Initiative markers on practices linked to actions: "Supervisor cadence → MGR-B2 ▲9 (sig.)"            │
├──────────────────────────────────────────┬─────────────────────────────────────────────────────────────┤
│ Movers: top 5 up / top 5 down            │ Mapping panel: units merged/split/closed since W1,          │
│ Drift alert: 'External orientation'      │ mapping rule, % of W1 respondents mapped, excluded units    │
│ was top quartile, now mid (-6)           │                                                             │
└──────────────────────────────────────────┴─────────────────────────────────────────────────────────────┘

Core interactions.

• Choose waves (any two, or a series of 3+ for trend lines).
• Comparability gate: only item versions marked comparable are compared. Changed items appear greyed with the reason.
• Hierarchy mapping: map W1 units to W2 units (1:1, merge, split with weight, retire). Unmapped units excluded with a visible count.
• Initiative linkage: every action's linked items are highlighted; movement is shown against the action's leading indicator target.
• Composition check: flags if the respondent mix shifted enough (e.g., frontline share ±10 pts) to explain movement; offers a composition-adjusted view.

Acceptance criteria.

• A delta is labeled "significant" only under the rules in the scoring spec (§5.6).
• Mapping coverage below 90% triggers a banner on every chart in the view and in exports.
• Legacy waves imported from other tools are marked "imported · mapped items only" wherever shown.

3.6 Supporting screens (list and purpose)