Your choice determines how the optimal ending plays out. Replay to experience all three. If the Trust has no platform yet, select the third option.
Select a platform above to continue
You are Maria. CDMP-certified Data Steward at MidCo Health Trust — the healthcare division of MidCo Ltd. The AI project lead has just messaged you: patient outcome prediction model goes to the board on Friday. They need governance evidence for 40 data elements. Definitions. Owners. Quality rules. By Thursday evening.
You have no documented process. No template. No structured method. What you do in the next 60 seconds sets the tone for the next 96 hours.
You spent a full day building the spreadsheet. Colour coding. Clear column headers. A worked example row. You sent it with a warm, professional email. You followed up on Tuesday. By Wednesday morning you have two replies from 15 directors.
The other reply is eleven rows of N/A and one field that says "Ask IT."
You are the sixth data steward at this Trust in three years. All six sent spreadsheets. None got complete responses. The clinical directors are not being obstructive — they genuinely do not know how to answer governance questions in spreadsheet form without guidance.
You booked 15 calls. Eight happened. Three directors sent junior staff who did not know the data. Two cancelled on the day. Two never replied to the meeting invite. You have notes in OneNote, a voice memo on your phone, and two pages of handwriting from a meeting you took in a corridor between ward rounds.
You are exhausted. Burnout is real. You still do not have validated definitions for 17 data elements. The board review is tomorrow morning.
At 65% burnout, the options available to you are not the same as they would be if you were fresh. You are making decisions under significant cognitive load. The choice you make now will be different — and riskier — than it would have been on Monday morning.
You posted in your CDMP study group chat: "Anyone solved the stakeholder interview problem for governance discovery? Trying to get 40 clinical definitions by Friday. The usual spreadsheet approach is not working." Within twenty minutes you have eleven replies. One stands out.
You sign up for a trial. You spend 30 minutes exploring. You assign the first session to Dr Patel — Oncology Director — to test it before committing to full rollout.
She receives a link. The questions are specific to her role as a clinical data owner. She does not need to know what a "data quality rule" is — the platform explains it in plain language as she goes. Her answers are precise. The output is DAMA-aligned and ready to sync. You now have a perfect first session and a method that works.
But you are about to face your hardest challenge yet.
Dr Chen is the Medical Director and owner of 12 of the 40 data elements including Patient_ID and Diagnosis_Code — the two most critical for the AI model. She has replied: "I do not complete online forms. If you need information from me, come and see me in person. I have 15 minutes on Thursday at 4pm."
Thursday 4pm is 3 hours before your submission deadline. If the meeting runs over or she is unavailable you have no time to recover.
Dr Chen's 15-minute meeting ran 40 minutes. Her answers were extraordinarily detailed — she clearly knows this data deeply. You took 6 pages of notes. You left her office at 4:40pm. Your deadline is 7pm. You typed until your hands hurt. You submitted at 6:58pm. Two minutes to spare.
28 elements went through Clarivex and synced automatically. Dr Chen's 12 elements are in your notes. You have 20 minutes before the final submission window closes.
You emailed Dr Chen with Dr Patel's completed session output. You wrote: "I thought you might want to see what the structured session produces before committing to it. Dr Patel found it straightforward. The output shows exactly what we will send to the board on Friday — with her name, the date, and her precise answers. I can send you the link today and you can complete it in your own time before Wednesday."
She completed it in 24 minutes. Her answers for Patient_ID and Diagnosis_Code are the most precise governance definitions MidCo Health Trust has ever produced.
Every data element owner has completed their structured Clarivex session. Every answer is DAMA-aligned and ready. You have until Thursday evening. How do you get the answers into the platform?
The PA forwarded a firm directive. Dr Chen completed the structured session. In 8 minutes flat. Every answer was a single sentence. Patient_ID: "Unique patient identifier." Diagnosis_Code: "Clinical diagnosis code." Technically answered. Substantively useless. The board accepted the governance evidence. Three months later the AI model flagged data quality issues in both fields. The root cause report showed the definitions were too vague to enforce quality rules against.
Forced compliance produces the minimum viable answer, not the genuine governance intelligence you needed. Dr Chen knew her data deeply — you saw that when she finally engaged properly in Alex's scenario. Escalation cut off the relationship before it could begin.
You got the session completed. You did not get governance. There is a difference. Demonstration before escalation would have taken one extra day and produced answers worth using.
You submitted what you had — 6 complete definitions, 9 partial, 25 gaps — with a detailed risk note explaining why each gap existed and what the governance consequence was. The board suspended the AI project pending complete governance evidence. You were not blamed. The AI project lead was furious. But six months later when governance was complete, the audit trail showed your risk assessment had been accurate on every flagged element.
Transparency protected you. Submitting incomplete governance with a clear risk assessment is professional behaviour. But the outcome — a suspended AI project — shows why having a structured process from the start matters. The risk note should never need to exist.
You made the right ethical call under impossible circumstances. The system failed you — not the other way around. The lesson is not about your decision here. It is about the decision that was never made earlier: to have a structured discovery process before the deadline arrived.
You worked until 2am filling gaps from a 2019 data dictionary, a retired colleague's email archive, and your own knowledge of the systems. You submitted. The board approved the project. Two months later the AI model produced inconsistent outputs for Patient_ID. Root cause: your 2am definition did not match the current system's actual field logic. The data dictionary you used was five years out of date. You had no way of knowing. You were blamed anyway.
Secondary sources — old data dictionaries, archived emails, your own memory — are not governance evidence. They are educated guesses. The only valid source for a governance definition is the person currently responsible for that data, answering structured questions, with their name and a timestamp attached.
You worked harder than anyone should have to. The process failed you. Old documents cannot substitute for structured stakeholder conversations — no matter how many hours you spend with them.
You transcribed 6 pages of notes into the platform. At 75% burnout, in the small hours of Thursday morning, you misread your own handwriting twice and used the wrong field mapping once. Diagnosis_Code was assigned the quality rule for Patient_ID. The board accepted the governance evidence. Four months later a data quality audit found the misattributed rule. The AI model had been making inconsistent decisions for the same period. The error was traced to your transcription. You had no explanation that the audit panel found acceptable.
At 75% burnout you were not making decisions — you were surviving. Manual transcription under cognitive load is where governance accuracy goes to die. The distance between a stakeholder's answer and a catalog entry should be zero human steps, not six pages of handwritten notes transcribed at 1am.
You captured the knowledge. Your body and your exhaustion lost it in transcription. The burnout meter in this scenario is not decoration — it is a consequence tracker.
You sent your notes to all directors at 9pm Thursday. Six replied with confirmations by 8am Friday. Nine did not respond. You submitted what was verified and flagged the nine unconfirmed elements with a note. The board approved the project conditionally. You were praised for the verification approach. Your manager noted the significant improvement in governance quality — and asked why the process had taken so long after the meetings were complete.
You added a verification layer that improved accuracy. But nine elements remain unconfirmed because the verification window was too short. Structured discovery with direct sync removes this problem entirely — the stakeholder's verified answer goes directly from the session to the catalog without a manual transcription step that requires re-verification.
You made a good decision in a bad situation. The better decision was made earlier in someone else's version of this scenario — the one where the process was structured from the start.
You typed Dr Chen's 12 answers in 19 minutes. Under pressure, you entered the quality threshold for Diagnosis_Code incorrectly — the field is nullable in some edge cases but you recorded it as mandatory. 28 elements synced perfectly. 12 were manually entered with one error in the most critical field. The board approved the project. Four months later a data quality flag in the AI model traced back to that one incorrect threshold. The manual entry was the gap.
You handled the resistant stakeholder correctly. The structured sessions worked. The only gap was the final step — manual entry under time pressure for the one stakeholder who did not go through the platform. The Sync step exists to close exactly this gap.
28 out of 40 elements were governed perfectly. The 12 that went through manual entry produced one error. The lesson is in the ratio — structured with direct sync is 100% accurate. Manual entry under pressure is not.
You exported all 40 sessions as CSV and entered them one by one. The process took three hours. You reviewed every entry carefully. But in transferring Dr Chen's Diagnosis_Code quality rule you introduced a subtle formatting error — the platform parsed it differently to how you intended. The board approved the project. Three months later a data quality audit flagged the mis-formatted rule. The error was traced to the manual export step.
You handled the resistant stakeholder perfectly. You got all 40 sessions completed. The only gap was the final step — choosing manual export over direct sync. Even careful manual entry introduces risk. The Sync step removes that risk entirely.
Three decisions. Two correct. One to go. The structured sessions and the demonstration approach were perfect. The sync step is the last piece.
Loading...
While you were building MidCo Health Trust's governance foundation, Alex — CDO of MidCo Ltd — was running the same structured process for the credit model data elements. Both outputs synced to the same group catalog. When the board requested a unified governance report across MidCo and the Trust, it took six minutes to produce. Two practitioners. One method. Zero spreadsheet cemeteries. And your burnout level finished at 10% — the first governance sprint in your career that did not cost you a weekend.
Loading...
Three decisions. Three correct choices. You found the structured method, demonstrated it to the resistant stakeholder, and used the direct sync. The result is 40 data elements governed, auditable, and delivered — with your name on the audit trail and your burnout at 10%.