MISSION 01 // AI-SUPPORTED BUSINESS JUDGMENT
Do not ask the AI for the answer first.
You are the analyst. Meridian Professional Advisory gives you four connected data files before telling you the management decision. Your job is to understand the evidence, define what it can support, and only then move toward a recommendation.
RRoleWho should the AI take?
CContextWhat data, decision, goal, and situation matter?
CConstraintsWhat boundaries, evidence rules, and assumptions must it respect?
OOutputWhat should the result look like: Excel, chart, dashboard, JSON, memo, CSV, HTML, or another format?
WORKFLOW
01Describe the data
02Bound the evidence
03Receive the goal
04Prioritize signal
05Plan analysis
06Analyze
07Challenge
08Recommend
02 // DATA INTAKE
Understand the files before you analyze them.
The data are synthetic, but designed to behave like a real professional-services dataset. Students need filtering, grouping, joining, and judgment rather than visual inspection.
180Clients
600Engagements
12,138Time entries
96Monthly cost rows
FULL FILES + SAMPLE PREVIEW
clients.csv 180 ROWS
Download full clients.csv
| client_id | industry | client_tenure_years | annual_revenue_band | services_used | referral_source | headquarters_state | satisfaction_score | strategic_client_flag |
|---|---|---|---|---|---|---|---|---|
| C0001 | Manufacturing | 9 | $100M+ | Audit, Client Accounting Services, Tax, Valuation | Partner | ID | 9.7 | Yes |
| C0002 | Construction | 15 | $100M+ | Audit, Client Accounting Services, Valuation | Bank Referral | CA | 6.9 | No |
| C0003 | Financial Services | 5 | $100M+ | Audit, Valuation | Attorney Referral | ID | 7.6 | No |
| C0004 | Technology | 3 | $50M-$100M | Audit, Valuation | Bank Referral | WA | 7 | No |
| C0005 | Technology | 14 | $25M-$50M | Audit, Tax, Valuation | Partner | CA | 8.6 | No |
| C0006 | Manufacturing | 5 | $50M-$100M | Audit, Tax, Valuation | Existing Client | CA | 8.4 | No |
| C0007 | Healthcare | 8 | $100M+ | Audit, Valuation | Website | WA | 7.2 | No |
| C0008 | Financial Services | 3 | $100M+ | Audit, Tax, Valuation | Existing Client | OR | 8.9 | No |
engagements.csv 600 ROWS
Download full engagements.csv
| engagement_id | client_id | service_line | engagement_type | completion_date | gross_billings | write_off | net_revenue | estimated_hours | actual_hours | realized_rate | office | manager | fee_type | payment_terms | meeting_mode |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| E10001 | C0074 | Audit | Annual Audit | 2024-07-24 | 89,300 | 3,617 | 85,683 | 365 | 327 | 262.03 | Portland | RB | Hourly | Net 30 | On-site |
| E10002 | C0133 | Audit | Review | 2024-09-04 | 132,194 | 6,630 | 125,564 | 534 | 501 | 250.63 | Spokane | AP | Fixed Fee | Net 60 | Virtual |
| E10003 | C0114 | Audit | Employee Benefit Plan Audit | 2024-03-16 | 96,369 | 4,714 | 91,655 | 390 | 404 | 226.87 | Spokane | JN | Fixed Fee | Net 60 | On-site |
| E10004 | C0088 | Audit | Employee Benefit Plan Audit | 2024-10-06 | 85,194 | 3,994 | 81,200 | 346 | 319 | 254.55 | Seattle | MH | Hourly | Net 45 | Hybrid |
| E10005 | C0164 | Audit | Employee Benefit Plan Audit | 2024-04-23 | 95,156 | 4,463 | 90,693 | 386 | 419 | 216.45 | Spokane | DR | Retainer | Net 30 | Hybrid |
| E10006 | C0140 | Audit | Annual Audit | 2024-05-27 | 68,307 | 2,283 | 66,024 | 281 | 284 | 232.48 | Spokane | MH | Hourly | Net 45 | On-site |
| E10007 | C0079 | Audit | Annual Audit | 2024-09-10 | 110,071 | 3,223 | 106,848 | 455 | 514 | 207.88 | Seattle | KT | Fixed Fee | Net 60 | Hybrid |
| E10008 | C0162 | Audit | Annual Audit | 2024-04-13 | 125,559 | 4,886 | 120,673 | 514 | 551 | 219.01 | Seattle | SC | Hourly | Net 30 | Hybrid |
time_entries.csv 12,138 ROWS
Download full time_entries.csv
| time_entry_id | engagement_id | employee_id | employee_level | hours | labor_cost | task_code | overtime | entry_date |
|---|---|---|---|---|---|---|---|---|
| T000001 | E10001 | S034 | Manager | 9.78 | 1,231.88 | MODEL | No | 2024-07-24 |
| T000002 | E10001 | S001 | Staff | 13.18 | 764.7 | MODEL | No | 2024-07-24 |
| T000003 | E10001 | S086 | Senior | 7.67 | 628.6 | MODEL | No | 2024-07-24 |
| T000004 | E10001 | S064 | Senior | 14.37 | 1,178.03 | PREP | No | 2024-07-24 |
| T000005 | E10001 | S005 | Senior | 27.47 | 2,252.57 | PREP | No | 2024-07-24 |
| T000006 | E10001 | S035 | Senior | 9.12 | 748 | REVIEW | No | 2024-07-24 |
| T000007 | E10001 | S063 | Manager | 20.4 | 2,570.44 | RESEARCH | No | 2024-07-24 |
| T000008 | E10001 | S072 | Manager | 11.32 | 1,426.94 | PREP | No | 2024-07-24 |
service_line_costs.csv 96 ROWS
Download full service_line_costs.csv
| month | service_line | software | dedicated_staff | contractors | marketing | allocated_IT | allocated_rent | allocated_admin |
|---|---|---|---|---|---|---|---|---|
| 2024-01 | Audit | 13,201.92 | 96,323.57 | 13,988.55 | 7,425.62 | 31,588.8 | 21,520.02 | 41,058.45 |
| 2024-02 | Audit | 10,627.7 | 112,277.92 | 13,592.11 | 7,234.13 | 32,638.98 | 21,948.02 | 38,008.59 |
| 2024-03 | Audit | 11,099.19 | 96,288.06 | 13,783.94 | 7,522.53 | 30,115.18 | 25,157.25 | 33,669.98 |
| 2024-04 | Audit | 12,177 | 94,178.54 | 12,938.16 | 7,644.6 | 28,366.43 | 21,102.38 | 34,926.31 |
| 2024-05 | Audit | 10,629.65 | 113,450.85 | 14,479.79 | 6,594.24 | 32,035.91 | 25,711.42 | 38,257.38 |
| 2024-06 | Audit | 11,325.71 | 92,881.65 | 14,283.06 | 7,174.93 | 28,784.78 | 25,495.58 | 34,335.66 |
| 2024-07 | Audit | 12,605.45 | 113,719.85 | 13,751.53 | 6,675.53 | 32,198.02 | 24,789.26 | 33,802.41 |
| 2024-08 | Audit | 13,036.4 | 102,868.79 | 13,727.59 | 6,703.75 | 32,531.92 | 23,602.95 | 36,794.6 |
First task: attach all four files to the AI product you choose. Ask it to describe the what one row represents in each file, the key fields, how the files connect, missing-value risks, duplicate risks, and unusual values. Do not reveal the management problem yet.
RCCO PROMPT CONSOLE
Build the instruction
R · C · C · O
This box is fully editable. The fields above are scaffolding, not a locked final prompt.
REVIEW THE AI RESPONSE
What did AI help you notice?
Instructor: Students should recognize clients as one row per client, engagements as one row per engagement, time entries as one row per labor entry, and service-line costs as one row per service line per month.
03 // ASK WHAT THE DATA CAN SUPPORT
Ask AI what this data can actually tell you.
You are not deciding the answer yourself yet. Ask AI to inspect the available fields and explain which questions the data can support, which questions need more evidence, and which conclusions would be unsafe. What AI says here should change your next move.
QUESTIONS TO ASK AI
Give AI the questions, not the answer categories
Use the four datasets and ask AI to respond to each question below. Require it to cite the fields it would use and identify missing evidence when necessary.
01
Which service line generates the most net revenue?Ask AI whether the files contain enough evidence and which fields it would use.
02
Are write-offs getting worse over time?Ask AI what comparison it would make and whether the available dates are sufficient.
03
Why are some clients leaving the firm?Ask AI whether the files actually contain evidence about client departures or reasons.
04
If Valuation is discontinued, will Audit or Tax clients leave too?Ask AI what the files can show and what would still be unknown.
05
Does low satisfaction cause lower retention?Ask AI to separate association from causation and identify missing retention evidence.
06
Which questions should management avoid answering with these files alone?Ask AI to identify the highest-risk unsupported conclusions.
RCCO PROMPT CONSOLE
Ask AI to test the evidence
R · C · C · O
Paste the six questions above into the prompt, or ask them one at a time. Use any AI product you choose.
WHAT DID AI SAY?
Use the AI response to decide what happens next
Do not score AI on whether it used the same words you would use. Evaluate whether it respected the evidence.
Instructor: Strong AI responses should recognize that revenue and write-off questions are directly supported, while client departure reasons, future client behavior, and causal retention claims require evidence that is not in the current files. The teaching point is that the AI response changes the analyst's next action.
04 // MANAGEMENT GOAL
Now the decision changes what matters.
Meridian's Valuation & Transaction Advisory service line reported a loss in 2025. Senior management is considering discontinuing the entire service line.
Primary goal: maximize firmwide profitability over the next 12 to 24 months while protecting strategically important client relationships.
Decision rule: only costs that would actually disappear if the service line were eliminated should initially be treated as avoidable.
05 // AI-DRIVEN PRIORITIZATION
Ask AI what matters. Then audit its judgment.
The AI does the first-pass prioritization. Your job is to evaluate whether its classifications are defensible for this management decision.
STEP 1Use the RCCO prompt below and ask your AI product to classify the variables.
STEP 2Record what the AI classified each variable as.
STEP 3For each variable, choose Accept AI, Challenge AI, or Need more evidence.
STEP 4Open the instructor comparison only after you have evaluated the AI.
Your role is not to sort the variables yourself first. You are reviewing an AI analyst's first-pass judgment.
RCCO PROMPT CONSOLE
Ask AI to prioritize the variables
R · C · C · O
Use any AI product of your choice. Then evaluate its judgment rather than treating its classification as the answer key.
AI OUTPUT REVIEW
Evaluate the AI's prioritization
Use the first dropdown to record what your AI said. Then evaluate that decision. Challenge the AI when its reasoning is too broad, too confident, or not tied to the management goal.
Variable
AI classification
Your evaluation
01
Revenue
02
Direct labor cost
03
Write-offs
04
Client ID
05
Services used
06
Office
07
Manager
08
Client tenure
09
Satisfaction score
10
Payment terms
11
Meeting mode
12
Headquarters state
Instructor comparison
| Variable | Suggested classification | Reasoning |
|---|---|---|
| Revenue | Critical | Directly affects whether the line adds economic value. |
| Direct labor cost | Critical | Needed to estimate contribution and avoidable cost. |
| Write-offs | Critical | Signals pricing, scope, and realization problems. |
| Client ID | Critical | Connects engagements to cross-service relationships. |
| Services used | Critical | Shows whether Valuation clients buy Audit, Tax, or CAS. |
| Office | Potentially useful | May reveal that losses are concentrated geographically. |
| Manager | Potentially useful | May reveal operational or pricing patterns. |
| Client tenure | Potentially useful | Could help interpret relationship risk. |
| Satisfaction score | Potentially useful | Potential relationship signal, but not causal proof. |
| Payment terms | Probably distracting | May matter for collection, but not the main discontinue decision. |
| Meeting mode | Probably distracting | Usually weakly related to this decision. |
| Headquarters state | Probably distracting | Not initially decision-relevant unless a regional pattern emerges. |
Instructor: The important outcome is not matching the suggested classification perfectly. Students should be able to identify where the AI's reasoning is strong, where it overreaches, and what additional evidence would justify reclassifying a variable.
06 // PLAN BEFORE CALCULATING
Make the AI propose the analysis before it produces the recommendation.
Connect every calculation to a management question and state what the result cannot prove.
RCCO PROMPT CONSOLE
Build the instruction
R · C · C · O
This box is fully editable. The fields above are scaffolding, not a locked final prompt.
Recommended analysis sequence
01Reported profit vs contribution
02Write-offs and realization
03Engagement concentration
04Cross-service client revenue
05Office and manager patterns
06Scenario if Valuation is removed
07 // ANALYZE
The reported loss is real. The interpretation is the problem.
The downloadable 2025 Valuation data reconcile to the figures below.
Net revenue$2,400,000
Direct labor$1,300,000
Avoidable fixed costs$580,000
Contribution$520,000
Reported profit-$230,000
AI VISUALIZATION TASK
Make AI show you the decision, not just describe it
Ask AI to create or specify visuals that make the management trade-offs visible. The visual should support the decision, not simply decorate the analysis.
Visual 1Reported profit vs contribution before allocated overhead
Visual 2Engagement profitability or write-off distribution
Visual 3Cross-service revenue tied to Valuation clients
Ask AI: choose the best chart type for each visual, identify the fields and calculations required, create the visual if the AI product supports it, and explain the management message in one sentence. Also state one way each chart could mislead the reader.
Open the evidence one layer at a time
Reveal 1 · Why is reported profit negative?
Valuation carries $750,000 of allocated IT, rent, and administrative overhead. If those costs remain after elimination, they are not avoidable.
Reveal 2 · What happens before allocated overhead?
2025 Valuation contribution is $520,000. Eliminating the service line could remove positive contribution while leaving allocated overhead behind.
Reveal 3 · Do Valuation clients matter elsewhere?
Exactly 29 of the 50 Valuation clients, 58%, use at least one additional service. In 2025 these clients generated $3,800,000 of net revenue from Audit, Tax, or Client Accounting Services.
Reveal 4 · Is the problem spread evenly?
Twelve of the sixty 2025 Valuation engagements were generated as high-complexity problem engagements with higher write-offs, more excess hours, and weaker realized rates.
Your preliminary decision
Suggested: Redesign.
08 // ADVERSARIAL REVIEW
Make the AI attack your own recommendation.
The purpose is to find assumptions that could materially reverse the decision.
RCCO PROMPT CONSOLE
Build the instruction
R · C · C · O
This box is fully editable. The fields above are scaffolding, not a locked final prompt.
Reversal conditions worth testing
AMost Valuation costs are actually avoidable
BCross-service revenue is not at risk
CPoor performance remains widespread after redesign
DCapacity can be redeployed to materially higher-return work
09 // MANAGEMENT DECISION
Give management a decision, not an AI transcript.
Your final answer should be short, evidence-based, explicit about uncertainty, and clear about what would change the recommendation.
FINAL SUBMISSION
Complete the report, then submit the Word file
01CompleteFinish every section of the Management Decision Report below.
02DownloadSelect Download Microsoft Word Report to create your .docx file.
03ReviewOpen the Word file and make sure your responses appear correctly.
04SubmitUpload the downloaded .docx file to the assignment in Canvas.
Canvas submission: upload the Microsoft Word (.docx) report generated by this lab. Do not submit a screenshot of this page or an AI chat transcript.
MANAGEMENT DECISION REPORT
Ready to submit?
The downloaded file will be named with your name when available.
Instructor: The Word export includes the student's recommendation, supporting evidence, evidence limitations, requested data, reversal condition, and AI judgment reflection. Students should upload the generated .docx file to Canvas.
10 // TRANSFER
The decision changes. RCCO stays.
Use the same workflow with a different strategic question. The goal is to learn how to manage an AI analyst.
Close an office?Shared clients, capacity, unavoidable costs
Accept a major contract?Incremental margin, capacity, concentration risk
Why is gross margin falling?Price, mix, cost, returns, data quality
Where should sales resources go?Margin, retention, growth, capacity
Reusable sequence: Describe → Bound Evidence → Goal → Prioritize → Plan → Analyze → Challenge → Recommend.