Appendix D. Five stages, survey import and Microsoft 365

The workspace is organized around the five stages every engagement moves through. Each existing screen lives in exactly one stage, so a consultant always knows where they are and what comes next.

D.1 The five stages

Stage Question it answers Pages
1. Intake What decision will this inform, for whom, and what do we expect to find? Brief (purpose, decisions, audiences, scope, hypotheses, constraints, deliverables, success)
2. Design What will we ask? Survey builder; Framework (Methodology Studio)
3. Develop What will people see, and who is invited? Preview; Look & feel; Population & privacy; Launch kit
4. Deploy How do people hear about it, and how is fieldwork going? Channels & schedule; Live fieldwork
5. Diagnose What did we find, and how does it reach the client? Results; Evidence; Insights; Wave compare; Report; Present & publish; Team notes
  • Navigation. A stage bar under the project title shows all five stages with their state (done, current, to do) and a one-line status read from the engagement itself. The pages inside a stage are a row of tabs. The sidebar repeats the five stages. Each stage returns you to the page you last used in it.
  • Continuity. Every stage page ends with "Continue to ".
  • Firm level. Industry Learnings, Methodology, Visualization library, Brand kits and Microsoft 365 sit outside any one project.
  • Overview. The dashboard: four headline figures with real trend lines (health index and response as they built up during fieldwork, the insight approval pipeline, projects by stage), what needs you (approvals by role, drift alerts, sponsor comments, open translations, at-risk fieldwork elsewhere in the portfolio), project flow with stage, progress and fee, activity, and a pipeline summary. A wave switch changes the figures between Wave 1 and the month-9 pulse. The bell in the header lists the same attention items from any page.
  • Search and Ask (the search field in the header, ⌘K, "/" or Ask in the sidebar). Searches pages, insights, learnings, charts, notes, framework and library questions, and Microsoft 365 apps. Questions about drivers, suppression, change since Wave 1, comment themes, hypotheses, response, benchmark position, survey length and strongest or weakest dimensions are answered from the engine, with a link to the page that shows the evidence. Answers are computed, never generated.

D.2 Survey builder (Design)

  • Forms. Census (63 framework items), executive short form (30) and frontline form (16). Switching form updates the timing estimate immediately.
  • Framework items are locked. Wording changes go through the Methodology Studio, so scores stay comparable with the benchmark and across waves.
  • Additive questions. Consultants add sections and questions of nine types: agreement, frequency, 1–5 rating, 0–10 likelihood, single choice, multiple choice, open text, number and video. Added questions are reported beside the framework and never change the index.
  • Question library. Validated items from the firm's other packs and modules (change readiness, speaking up, integration, hybrid work, wellbeing, inclusion, advocacy, safety).
  • Logic. Follow-up questions shown only when an answer meets a condition. The four routed probes that ship with the pack are shown and locked; consultants add their own.
  • Timing. Median minutes on a phone from the assessment catalog's timing model (§7.1), including follow-ups weighted by how often they fire.
  • Export to Word. The whole instrument as an editable document for client review.

D.3 Import

Import accepts Word (.docx), Excel (.xlsx), CSV, a Microsoft Forms response export, and pasted text. Files are read in the browser and never uploaded.

Source How it is read
Word Title → survey title; headings → sections; numbered paragraphs → questions; bullets and indented list items → answer options; "(Yes / No)" after a question → options; a table with scale labels across the top → one statement per row on that scale. Instructions are skipped and listed.
Excel / CSV The firm template (Section, Question, Type, Options, Required), or any sheet with a question column; otherwise a plain list in the first column. The delimiter is detected.
Microsoft Forms Metadata columns are dropped; each remaining column is a question. Type is inferred from the answers (1–5 → rating, 0–10 → likelihood, ";"-separated → multiple choice, a few repeated values → single choice, otherwise open text). No respondent answer is imported.
Pasted text One question per line; lines ending in a colon, in capitals or starting "Section" become sections; indented bullets become options.

Type inference. From the options first (agreement or frequency labels, 0–10, 1–5, a list), then the wording ("select all", "how many", "how often", "how likely … recommend"), then punctuation (a question mark with no options is open text; a statement is an agreement item).

Matching to the methodology. Each question is compared with the framework items after removing stop words and words too common in surveys to carry meaning.

Result Rule Default
Already in the framework A scaled statement that shares most of its meaningful words with a framework item, both ways Left out, with the matching item shown, so the benchmark stays comparable
Close to a practice Shares at least two meaningful words with a framework item Tagged to that practice so it sits beside it in results
New Anything else Added as is

Choice, number and open questions can never be duplicates; at most they are tagged. Every row can be edited, retyped, included or excluded before adding. The firm's Excel template is downloadable from the import screen.

D.4 Develop and Deploy

  • Preview. Phone or desktop, in the chosen brand kit and language list, walking every screen: welcome, pre-filled "about you", framework blocks, added sections, follow-ups that fire on low answers, open questions, video, submit and thank-you.
  • Look & feel. Brand kit (firm, co-brand or white label), welcome and privacy text, progress style, one question per screen, save and finish later, languages with review status. The privacy promise is checked against the reporting threshold.
  • Population & privacy. Directory source (file from HR, Microsoft Entra ID, HR system connector), hierarchy by site and level, anonymity mode (k=5, or k=8 for works councils), cell-size simulator, data-protection checklist.
  • Launch kit. Eight readiness checks linked to the page that fixes each one; launch note, manager talking points, reminders and kiosk poster in the client's voice; export to Word; save to SharePoint.
  • Channels & schedule. Email (through Outlook when connected), Teams, text message, QR and kiosk, personal link, WhatsApp (opt-in only). A day-by-day schedule of sends. Reminders reach non-completers only; units below k receive site-level messages only.

D.5 Microsoft 365

App Used in Permission (Microsoft Graph) Never
Entra ID Population User.Read.All, GroupMember.Read.All Change the directory; read mail, chats or files
Outlook Channels Mail.Send, one shared mailbox Read any mailbox; reveal who has responded
Teams Channels TeamsActivity.Send Read chats; post in channels
SharePoint & OneDrive Every export Sites.Selected, the project site only Browse other sites
Power BI Report Dataset.ReadWrite.All, Workspace.Read.All Publish respondent-level rows
Excel, Word, PowerPoint, Forms Import and export None: files are read and written on the device Upload the file

Permissions are granted once by the client's Microsoft 365 administrator and are revocable from Entra ID. Each connection shows what OrgDiagnostic will and will never do before consent.

Prototype note. In the clickable prototype, the Entra ID, Outlook, Teams, SharePoint and Power BI connections are simulated and labelled as such. Word, Excel, PowerPoint, Power BI package, PNG and SVG files are real.