← Back to the exerciseThe full worked example: Answer "is training the problem?" with your onboarding and turnover data
campo.
Field guide / L&D / UC-001 · SC-06
The result

"Is training the problem?" Probably not the main one.

An L&D manager in an 11-plant business unit puts onboarding, turnover and exit data side by side to answer his president's question honestly, with any AI chat his company allows.

The resultWhat Marcus sends the BU president. Fictional company.
QUILLON · ROOFING & SIDING BUSINESS UNIT
New hires: what's really going on
For: Tom Hargrove, BU President
Ops review, Dec 2 · From: Marcus Bell

THE SHORT ANSWER

Probably not the main one. So far 36% of new hires have left within 90 days, and the top exit reasons are overtime and schedule (41%) and pay (22%). We don't measure readiness yet, so I can't rule training out. · Input 3

WHAT I NEED FROM YOU

THIS WEEK
Put Plant 4's two lead trainers back on training
They've covered the line since September. 18% of leavers say they felt unprepared or had no buddy on the floor. · Inputs 3, 4
DECEMBER
Close Plant 4's 38-operator gap and end mandatory overtime
74 of Plant 4's 140 hires (53%) left within 90 days. · Inputs 3, 4
Q1
Have Operations and HR review pay and shifts for new operators
Schedule and pay are 63% of exit reasons. Training can't fix them. · Input 3

WHAT THE NUMBERS SAYInputs 2, 3, 5

Onboarding 96%Misleading
Videos only; leavers removed
Early leavers36% so far
187 of 520 hires
Plant 453%
74 of 140 hires
Plant 9Unclear
4 leavers, 6 hires: not one group
ReadinessUnknown
Week-4 checks in binders

PLANT 4'S CLASSROOM REQUEST

Not now. Two weeks off the floor would deepen a 38-person gap. Trainers go back as buddies first, and I'll report week-4 skills checks from all 11 plants in January before we decide. · Input 4
Caught on review
1
"96% onboarding complete" is videos, with leavers removed
2
"Only 4 early leavers" is a count, not a rate
3
"New hires aren't ready" is a staffing gap first
UC-001 · SC-06 · Rev B
FOR
Business-unit L&D Manager
SECTOR
Building products
COMPANY
9,000 in the business unit (of 35,000) · 11 plants · US
TASK
Answer "is training the problem?" for the BU president
WRITTEN FOR
ChatGPT · Claude · Gemini · Copilot
AI TEST
Claude, Nov 13, 2026 · 6 of 6 found
Step 0 · The mess

What landed on Marcus's desk

A president's question, an onboarding report at 96%, turnover data, and two plants telling very different stories. The question is about training. Most of the answer is somewhere else.

EMAIL · BU PRESIDENT[1]
New hires
“Every plant manager tells me new hires aren't ready… Is training the problem?”
1
A yes-or-no question. Answer it.
LMS EXPORT[2]
Onboarding completion: 96%
“Completion = pre-start online modules… Learners who leave during the period are removed.”
2
Videos, and only the people who stayed.
HR DATA[3]
Hires and early leavers
520 hires, 187 left within 90 days. Exit reasons: overtime 41%, pay 22%, no buddy 18%.
3
Most reasons aren't training.
NOTE · PLANT 4 HRBP[4]
Two-week classroom request
“38 operators short… Our two lead trainers have been working on the line since September.”
4
A staffing gap asking for a training fix.
NOTE · PLANT 9 HRBP[5]
Good month!
“Only 4 early leavers in October. (We hired 6 in October.)”
5
A count, not a rate.
You're not alone
31%
hourly turnover across US manufacturing in both 2023 and 2024, with the first 90 days the critical window for early mismatches.
Manufacturers Alliance with Aon, research released July 2026 (survey of 22 enterprise-tier manufacturers plus Aon benchmarking). Source
Steps 1–3 · How it works

Three steps, one chat

Here the AI separates what training can fix from what it can't. The prompt asks it to.

1
Open a new chat

Use the AI tool your company approves: ChatGPT, Claude, Gemini or Copilot.

2
Paste the prompt and the data

Copy the prompt and the five fictional inputs from the appendix into one message.

3
Answer the question first

Write the one-sentence answer yourself. Then decide which asks only your leader can grant.

The prompt, part by part
Copy-ready version in the appendix

I'm the L&D manager for a business unit of about 9,000 people in 11 manufacturing plants. Our BU president asked whether new hires aren't ready because of training, and wants one page for his operations review on December 2. Below are his email, our LMS onboarding export, HR hiring and turnover data, and notes from two plant HR business partners.

1
The question your leader asked, in his words. The AI's job is to answer it, not to describe the data.

Please give me:

  1. The answer to his question in two sentences.
  2. What the numbers really show: onboarding, early turnover and readiness, with the input each point comes from.
  3. Plant 4's request: should we approve it, and why or why not.
  4. What I need from the president: no more than three asks.
2
Four sections, starting with a two-sentence answer. The limit of three asks forces priorities.

Be skeptical. Check what each number counts and who it leaves out, compare small counts with rates, and separate training problems from staffing, pay or schedule problems. If something is missing, say it's missing. Don't add facts that aren't in the inputs.

3
The heavy lifting: who each number leaves out, counts versus rates, and training versus staffing, pay or schedule. It found all six problems.
Where to paste
Every tool has one message box. Paste the prompt, add a blank line, paste the data and notes, send. Exports can be attached as files if your tool allows it.
Safe to try
Practise with the fictional data first. Real turnover and exit data is personal: use only your company's approved tool and follow its rules on employee data.
The AI's first answer

What Claude actually wrote

Exactly as it came back, not edited. Read it the way Marcus did: check each point against the source it cites.

UNEDITED
Claude · fresh session, no other context · Nov 13, 2026 · one run

I wrote the one-page answer below from your five inputs only. A: cut

1. The answer in two sentences B: plainer

The data doesn't show that training is the main problem: the top exit reasons are mandatory overtime or schedule (41%) and pay (22%), and we have no data at all on whether new hires can do the job. Only 18% of leavers said they felt unprepared or had no buddy on the floor, and at Plant 4 the trainers who would be that buddy have been working on the line since September to cover staffing gaps.

2. What the numbers show

Onboarding

  • The 96% completion rate overstates things. It counts only people still employed. The export removes anyone who left during the period (input [2]). found
    • The base is 429 people. HR shows 520 hires (input [3]), so 91 hires are missing from the LMS figure.
    • The 187 early leavers are probably among those dropped. The inputs don't let us reconcile the two counts.
  • "Completion" means 2.5 hours of safety videos watched before day one (input [2]). It shows people finished the videos. It doesn't show they can work on the floor. found
  • There is no readiness data. The week-4 on-the-job skills check is kept in paper binders at each plant, not in the LMS (input [2]). found
    • So we have nothing that measures whether new hires are "ready," which is the question the president asked.
    • The plant managers' view (input [1]) is a judgment. It isn't a measurement.

Early turnover

  • All plants: 187 of 520 hires left within 90 days, or 36% (input [3]).
  • The 90-day window isn't complete yet. People hired in September and October haven't reached 90 days. So 36% will probably rise.
  • Plant 4 Ridgeway: 74 of 140 hires left, or 53% (input [3]).
    • Without Plant 4, the other 10 plants are at 113 of 380, or about 30%.
    • Plant 4 made 27% of the hires but accounts for 40% of the early leavers.
  • Exit survey reasons, all plants combined (input [3]):
    • Mandatory overtime or schedule: 41%
    • Pay: 22%
    • Felt unprepared or no buddy: 18%
    • Other: 19%
  • Three things the survey doesn't tell us:
    • How many people answered it.
    • The reasons broken down by plant.
    • How many said "unprepared" versus "no buddy." Those are two different problems combined into one answer option.

Plant 9 Lakeview: the "good month" doesn't hold up (input [5])

  • The 100% completion uses the same active-learners-only count, so it has the same gap.
  • "Only 4 early leavers" is a count, not a rate. Against 6 October hires, that would be 67%. found
  • It isn't clear whether the 4 leavers were October hires.
  • No earlier figure is given, so "way down" can't be checked. A hiring freeze would lower the count anyway.

Staffing, pay and schedule versus training

  • Staffing and schedule: Plant 4 is 38 operators short of plan. It has run 12-hour shifts with mandatory overtime since August, and its two lead trainers have been on the line since September (input [4]). found
  • Pay: 22% of exit reasons (input [3]).
  • Training: at most 18% of exit reasons (input [3]). Part of that may be "no buddy," which is a trainer availability problem rather than a gap in course content.

3. Plant 4's request: don't approve it as written D: softened

The plant manager wants a two-week classroom programme before anyone works on a line (input [4]). It doesn't address what the inputs point to:

  • It needs trainers the plant doesn't have. Both lead trainers are covering on the line. found
  • It adds to the staffing gap. Every new hire would spend two weeks off the floor at a plant that is already 38 operators short. That likely means more overtime for current staff, and overtime is the top exit reason.
  • It doesn't touch the two biggest exit reasons. Schedule and pay together account for 63%.
  • We can't measure whether it works. There's no readiness baseline, because the week-4 checks aren't collected centrally.

What would help more: put the two lead trainers back on training and buddying, and close the 38-operator gap. Revisit classroom training once week-4 skills data shows a real skills gap.

4. What I need from the president

  1. A staffing decision for Plant 4: relief on the 38-operator gap and the mandatory overtime, so the two lead trainers can go back to training and buddying new hires.
  2. A requirement that plants report the week-4 skills check centrally, and that the LMS onboarding report include people who left. That gives us a real measure of readiness. C: moved
  3. A review of pay and shift schedules for new operators, owned by operations and HR. Those are the top two exit reasons and training can't fix them.

Still missing: readiness results of any kind, the exit survey response count and plant-level breakdown, early turnover for the 90-day period once it closes, and a Plant 9 rate based on a clear denominator.

foundcaught a planted problemA–DMarcus changes it; see the next section
The human part

What Marcus changed, and why

Claude found every planted problem, and noticed that the 90-day window hasn't closed for recent hires. Marcus's edits are judgment calls about a president who asked a simple question.

What Claude found: 6 of 6 planted problems
One run
✓“Complete” means videos only
✓Leavers dropped from the report
✓Plant 4 is short 38 operators
✓Trainers pulled onto the line
✓Plant 9: a count, not a rate
✓No readiness data
Four edits, all judgment calls
A
CLAUDE WROTE
About two pages: an answer, four sections of findings, a recommendation and three asks.
MARCUS'S VERSION
One page.
WHY · LENGTH
An ops review gives this five minutes.
B
CLAUDE WROTE
“The data doesn't show that training is the main problem: the top exit reasons are…”
MARCUS'S VERSION
“Probably not the main one.” Then the evidence, and what we can't yet rule out.
WHY · AUDIENCE
He asked a yes-or-no question, so the answer goes first. Without readiness data, it can't be a flat no.
C
CLAUDE WROTE
Ask 2: “A requirement that plants report the week-4 skills check centrally.”
MARCUS'S VERSION
Ask 1: “Put Plant 4's two lead trainers back on training.” Marcus handles the reporting himself.
WHY · PRIORITY
Only the president can pull people off a short-staffed line, and it's the fastest fix. Reporting is Marcus's own job.
D
CLAUDE WROTE
“Plant 4's request: don't approve it as written.”
MARCUS'S VERSION
“Not now. Trainers go back as buddies first, and we decide with week-4 data in January.”
WHY · TONE
The plant manager is right that new hires struggle. Giving him a path keeps him on side.
What only you can do
Say plainly what training can't fix, ask for what only your leader can give, and keep a frustrated plant manager on side. The AI separated the causes. You make the case.
Your turn

Make it yours

Use it whenever a leader asks whether training is the answer: onboarding, quality problems, customer complaints, safety incidents.

Your prompt: fill in the brackets

I'm the [your role] for [your organization or unit] of about [number] people. [Your leader] asked whether [the problem] is because of training, and wants [one page / a short answer] for [the meeting]. Below are [their question, your training data, related HR or operations data, and notes from the sites].

Please give me: 1. The answer to the question in two sentences; 2. What the numbers really show, with the input each point comes from; 3. [Any specific request on the table]: should we approve it, and why or why not; 4. What I need from [your leader]: no more than three asks.

Be skeptical. Check what each number counts and who it leaves out, compare small counts with rates, and separate training problems from staffing, pay or schedule problems. If something is missing, say it's missing. Don't add facts that aren't in the inputs.

Five checks before you send
1
Your first sentence answers the question that was asked.
2
Each "completion" number says what was completed, and by whom.
3
Rates are shown next to counts, especially for small groups.
4
Every ask is something only your leader can give.
5
The people asking for training get a path, not just a no.
Make it repeatable (optional)
Save the prompt once and reuse it whenever the question comes up: a custom GPT in ChatGPT, a project in Claude, a Gem in Gemini, or an agent in Copilot. Some companies switch these off; copy and paste always works.
With real HR data
Turnover and exit data is personal. Use only the AI tool your company approves, and follow its rules on employee data.
What this example shows
One run of this prompt in Claude found all six problems planted in fictional inputs and invented nothing.
What it doesn't show
That every tool, setting or run will do the same, or how much time it saves you. Quillon Building Products and its people are fictional.
Appendix

Copy-ready: the prompt and the five inputs

Copy the prompt, then all five inputs, into one message in a new chat. Everything here is fictional and safe to paste.

The prompt

I'm the L&D manager for a business unit of about 9,000 people in 11 manufacturing plants. Our BU president asked whether new hires aren't ready because of training, and wants one page for his operations review on December 2. Below are his email, our LMS onboarding export, HR hiring and turnover data, and notes from two plant HR business partners.

Please give me:

  1. The answer to his question in two sentences.
  2. What the numbers really show: onboarding, early turnover and readiness, with the input each point comes from.
  3. Plant 4's request: should we approve it, and why or why not.
  4. What I need from the president: no more than three asks.

Be skeptical. Check what each number counts and who it leaves out, compare small counts with rates, and separate training problems from staffing, pay or schedule problems. If something is missing, say it's missing. Don't add facts that aren't in the inputs.

[1] Email from Tom Hargrove, BU President

Subject: New hires

Marcus, every plant manager tells me new hires aren't ready when they hit the floor. Is training the problem? Give me one page for the Dec 2 ops review: what's going on and what you need from me.

[2] LMS export: new-hire onboarding, July to October

New-hire onboarding completion: 96% (412 of 429) Completion = pre-start online modules (safety orientation videos, 2.5 hours), assigned at offer acceptance. Report shows active learners only. Learners who leave during the period are removed. Week-4 on-the-job skills check: recorded in plant binders, not in the LMS.

[3] HR data: hires and early leavers, July to October
Hires Left within 90 days %
All 11 plants 520 187 36%
Plant 4 Ridgeway 140 74 53%

Exit survey, top reasons (all plants): mandatory overtime or schedule 41% · pay 22% · "felt unprepared / no buddy on the floor" 18% · other 19%.

[4] Note from the Plant 4 Ridgeway HR business partner

Plant manager wants a new two-week classroom onboarding programme before anyone touches a line, because "new hires aren't ready". For context: we've run 12-hour shifts with mandatory overtime since August because we're 38 operators short of the staffing plan. Our two lead trainers have been working on the line since September to cover gaps.

[5] Note from the Plant 9 Lakeview HR business partner

Good month! Onboarding completion 100%, and turnover is way down: only 4 early leavers in October. (We hired 6 in October under the hiring freeze.)