Your company has deployed Copilot, but how should you use it for a performance review or project planning? Start with questions you already face each month: why have results weakened, and what should the team do first?
These two worked examples from iTalent show how to brief Copilot, supply data, question the response and prepare a recommendation for your manager. Each includes copyable inputs and checked calculations, so you can practise with the example before using work material approved by your organisation.
To try the examples, paste the sample material and prompt into your organisation’s Copilot chat, then work through the follow-up questions. For actual documents, identify the version to use and check that the response cites it. File and work-data access depend on your account setup; these exercises can begin with pasted text.
Scenario one: more enquiries, fewer sales
Your manager asks whether to increase campaign spending. Start by locating the deterioration in the funnel and identifying the evidence still needed.
Copyable example data
Both cohorts are grouped by enquiry-entry month and observed for a complete 30 days afterwards. Duplicates are removed and definitions are consistent.
Cohort A: 200 enquiries; 100 qualified; 60 quoted; 24 sales; sales revenue HK$240,000.
Cohort B: 240 enquiries; 96 qualified; 48 quoted; 12 sales; sales revenue HK$120,000.
Median first-follow-up time: A 4 hours; B 10 hours.
Unknown: channel and customer mix, loss reasons, campaign expenditure and individual follow-up histories.
Decision: improve follow-up next month or increase campaign spending?
Analysis prompt
Check the funnel definitions, then calculate enquiry-to-qualified, qualified-to-quote, quote-to-sale and overall conversion, plus revenue per sale. Show each formula, cohort result and difference. Distinguish percentage-point changes from relative changes.
Identify where performance weakened without presenting simultaneous changes as causal. Give three possible explanations, the evidence needed to support or challenge them and the first records to inspect. Explain whether the evidence supports more spending.
Results to check
- Enquiries increased 20%, from 200 to 240. Sales count and revenue each fell 50%.
- Enquiry-to-qualified: 50% to 40%, down 10 percentage points.
- Qualified-to-quote: 60% to 50%, down 10 percentage points.
- Quote-to-sale: 40% to 25%, down 15 percentage points.
- Overall conversion: 12% to 5%, down 7 percentage points.
- Revenue per sale remains HK$10,000 for both cohorts.
The deterioration spans three stages. Unchanged revenue per sale also means the revenue decline does not establish a fall in average order value.
Slower first follow-up deserves investigation but is not a proven cause. Inspect channel mix, customer quality and loss records. Without expenditure, you cannot assess the return on additional spending.
Follow-up prompt: plan the investigation
Create a verification card for each hypothesis: changed customer mix, delayed follow-up and weaker quote competitiveness. Specify case fields, comparisons, confounding factors and results that would support or weaken each explanation.
Then propose one follow-up process adjustment to trial, identifying the responsible function, measures and conditions to confirm first. Do not estimate revenue uplift. Leave unagreed dates and thresholds blank.
For example, compare follow-up times and outcomes within similar channels and customer groups, then inspect loss reasons. A new assignment process needs the same outcome-observation period and workload monitoring. A simple before-and-after comparison still cannot rule out changes in customer mix.
Example management update
Enquiries rose 20%, but overall conversion fell from 12% to 5%, with deterioration across three stages. Without expenditure and channel-level data, we cannot recommend additional budget. I suggest checking follow-up and loss records for comparable enquiries before deciding on an assignment-process trial. I will then propose its timing, owner and review measures.
Scenario two: 130 hours of projects and 80 hours available
Identify non-deferrable work before comparing optional combinations. A weighted score created by AI is not a substitute for real constraints.
Copyable example data
Available project capacity this period: 80 hours, after routine work. Estimates include setup and review.
A | Committed client delivery: 35 hours; mandatory this period.
B | Recurring operations report: 20 hours; mandatory this period.
C | Reporting automation trial: 25 hours; estimated future saving 8 hours per period, unverified. Requires the same specialist as B.
D | CRM data cleanup: 30 hours; originally planned this period but deferrable with manager approval. Benefit unquantified.
E | Non-urgent presentation redesign: 20 hours; deferrable.
The specialist has only 30 hours available for B and C together; no substitute is currently available.
Goal: honour commitments and retain room for unexpected work.
Options prompt
Calculate total demand, the capacity gap and capacity remaining after mandatory work. Check individual skill constraints as well as team hours.
Compare mandatory work only; mandatory work plus all of C; and mandatory work plus a preliminary validation stage for C. The third option’s scope and estimate are unconfirmed: identify what must be agreed rather than assuming C can be split.
Show deferred work, required approvals, dependencies and estimation risks. Do not assume overtime or extra staff.
Results to check
- Total demand is 130 hours against 80 available: a 50-hour gap.
- A and B require 55 hours, leaving 25 team hours.
- A+B+C totals 80 hours, but B+C requires 45 specialist hours against 30 available: a 15-hour overload. The plan is infeasible as stated.
- After B, the specialist has 10 hours. A preliminary stage for C would need an agreed deliverable within that capacity; it cannot be treated as completed automation.
If an eight-hour validation stage is confirmed—for example, checking data inputs and process requirements—the total is 63 hours, leaving 17 team hours. Specialist demand is 28 hours, leaving two. This is a conditional planning example; the person doing the work must first agree the scope.
Management message template
We have 80 hours this period; A and B require 55. Full delivery of C fits the team total but overloads the shared specialist by 15 hours. I recommend securing A and B, then confirming whether C can start with an eight-hour validation stage. If feasible, that leaves 17 team hours for changes. Please approve deferring D; E will also move to the next period. If C cannot be split, we should complete A and B only and reschedule C separately.
Dividing C’s 25-hour investment by its estimated eight-hour recurring saving gives about 3.125 periods. This is only a rough labour-time payback comparison if the savings materialise and there are no additional maintenance or other costs. It is not a verified return.
Keep a record of your judgement
These examples show how to identify missing evidence, compare like with like and detect skill constraints hidden by a team total. Keep the brief, corrections and decision so you can explain your AI use in an interview beyond naming a tool.
Use the four-pass prompt framework to build your own template. When considering your next move, explore current iTalent opportunities.