Most training requests aren't training problems. A course closes two of the six performance gaps. This guide is the 20-minute conversation that shows which ones you're facing.
L&D partners
HRBPs
Anyone who gets "we need training" from three levels up
You know the message: "The team needs training on X. Can we get something running by next month?" What you ask next decides whether you build a course to order or find the business problem behind it.
You'll be able to
turn a training request into a business problem in one conversation
run the first 20-minute CLUE conversation with any stakeholder
write an impact hypothesis a sponsor would sign
open a Case File that survives a steering committee
The guide in one picture
Every request arrives above the waterline. File 03 finds out which of the six gaps sits underneath. Four of them won't move, however good the course is.
Part one
Talk about learning so the business listens
File 02
File 02
The Detective Mindset
When to use this: before any conversation with a business leader about learning.
About 2 min
Most business leaders don't ask how people learn. They ask how people perform. Describe your work by the number it moves, and they hear a partner instead of a cost center.
The first interview
✕The order-taker trap
"My new managers are struggling. We need a leadership workshop ASAP."
The wrong reply asks about headcount and start dates, which locks you into a solution before anyone has named the problem.
✓The detective pivot
Say this instead:
I can help with that. Before we book anything, give me twenty minutes on what's happening on the ground. I want to know what this is costing us before I propose what fixes it.
Three rules for talking to executives
1
The Filter.
Does this make money, save money, or protect money? If you can't trace it to one of those, expect it to be treated as a nice-to-have, however good the design.
2
Ditch the education jargon.
Bloom's, scaffolding, andragogy: keep them for L&D colleagues. Say "phased support" instead of scaffolding, "real-world application" instead of andragogy.
3
Use "we", not "you".
You"Your team needs to learn X" sounds like blame.
We"We need ticket-handling speed down so we don't lose our best customers" makes you a co-owner.
The translation dictionary
The test for every "Try saying" phrase: does it point to making, saving or protecting money?
Instead of saying
Try saying
Why it works
Training needs analysis
Performance gap analysis
Where the business loses money
Learning objectives
Performance outcomes
What the team does differently on Monday
Upskilling / reskilling
Building capacity
Work the team couldn't take on yesterday
Soft skills
The behaviors: giving feedback, running meetings, writing clearly
Leaders buy behaviors, not categories
Seat time / hours
Time to proficiency
Every learning hour is a non-working hour
Updating content
Keeping company knowledge current
The right answer, first time
Compliance training
Risk mitigation
Insurance against lawsuits and fines
Satisfaction scores
Behavioral adoption
Whether people use it, not whether they liked the session
Completion rates
Operational readiness
Ready to run the next launch
Gamification
Practice with a scoreboard
Names the practice. "Gamification" sounds like games to a CFO
Adapt the words to your company.
The "So what?" reality check
The clue"We want to launch a communication skills path."
So what?"So project updates are clearer."
So what?"So PMs stop spending half their day chasing status."
The hook"So we free up 20% of PM time for billable client work."
Keep asking until you hit make, save or protect. Run whatever is top of your queue through it now. If you find no hook in four steps, better to know that before the meeting.
Part two
Investigate before you build
Files 03–05
File 03
The CLUE Method
When to use this: a training request just landed and you have one conversation before the scope gets locked.
About 6 min
Four moves, in order. C and L fit in the first 20-minute conversation. U and E come after it, once you have evidence.
Two sizes of investigation
Size one The quick read, 20 minutes.
The C questions plus one L question, inside the conversation you were already having. Covers most requests.
Size two The full investigation, one to two weeks.
Interviews, shadowing, a look at the data, a Case File. For a big audience, real budget, an executive sponsor, or a number leadership tracks every month.
Tell the sponsor which one you're running. The quick read is a first filter, not a verdict. Upgrade it the moment it turns up something expensive.
C
C · Clarify the business goal
Ask this
If you keep one questionWhat's happening in the business that made this urgent now?The why-now probe. Urgency names the real trigger, often the real sponsor.
Imagine it's 90 days from now and this worked perfectly. What are people doing differently on Monday morning, and which number moved?The miracle question, borrowed from solution-focused coaching. Turns a vague ask into a picture you can check.
What is this costing us today: time, money, risk, or customers?
Who's losing sleep over that metric?
What have you already tried?
Done when"KPI X is at A, needs B, the gap costs Y" and the stakeholder nods. (→ String Board)
Watch out"We need training" is a solution, not a goal. Don't argue with it. Keep asking about the problem behind it. Some requests are about visibility, not performance. Then ask: "What does your leadership need to see, and by when?" and plan a fix that gives the sponsor something to show.
You leave with the gap named. There are six kinds, adapted from Thomas Gilbert, and most cases have two or three at once.
Skillnever done under pressure
Knowledgenobody ever told them
Processbroken upstream
Toolsfail when used
Incentivespoint the other way
Expectationsnever said out loud
Ask this
If you keep one questionIf we trained everyone tomorrow and the training was perfect, what would still get in the way?The most useful question in this guide. The answer lists the gaps that sit in the environment, not in people.
Could your best performer do this today? What do they do differently?The bright-spot contrast. If one person can, it isn't purely skills, and you've found your expert witness.
On a scale of 0 to 10, how well does the team do this now? What makes it a 4 and not a 2?The scaling question. The second half names what works, which is where the fix hides.
What do we already have on this: tickets, error logs, CRM notes, survey verbatims?The paper trail. Cheaper than an interview, and it is what the sponsor will believe.
Do people know this is expected of them? And what happens when someone doesn't do it?
The field testCould they do it if their livelihood depended on it? If yes, you're not looking at a skill gap. (Mager and Pipe's classic.)
✓Training fixes this
KnowledgeNobody has shown them how
SkillTheory known, never done under pressure
KnowledgeA new system, process or policy landing next month
KnowledgeA product launch where the story is new
KnowledgeCompliance, where the record of training is the requirement
✕Training does not fix this
ProcessA step broken upstream
ToolsSystems that fail at the moment of use
IncentivesPay or targets that reward the wrong thing
IncentivesWork nobody is held to
ExpectationsNobody has said it's expected of them
ExpectationsManagers who don't manage
Outside the sixToo few people for the volume
Outside the sixA role that doesn't fit the person in it
Outside the sixA culture where naming the problem costs more than living with it
If the cause is on the "does not fix" list, no course will move the number. Those causes belong to people who can change them faster than you can build a course, and they go on your handover list (→ 07). Headcount, hiring and culture sit outside all six and are the hardest to hand over. At this stage all of it is still a hypothesis: U checks it.
Sort the evidence
Eight things a stakeholder might tell you. Tap a quote, then tap its box. With a mouse, you can drag it. On a keyboard, focus a quote and press 1 or 2.
1A course can fix this
Knowledge"Nobody has walked the new hires through the returns screen."
Skill"They know the escalation rule. They've just never done it with a customer shouting."
Knowledge"The new policy lands next month and nobody has seen it."
2Hand it over
Tools"The tracker asks for fifteen fields, so people update it at sprint end."
Process"Discount approval takes five days and two signatures."
Incentives"Handle time is 40% of the scorecard."
Expectations"Nobody ever said the cascade was the manager's job."
Tools"The best analyst has a template. Everyone else rebuilds the report by hand."
Three of eight belong to a course. In my intakes that mix is common.
Done whenthe gap type named out loud and agreed, or corrected with evidence.
Watch outDon't run this step from a meeting room. Leaders describe the workflow they designed. Watch the one people actually run. (→ Fingerprint Dust)
If you keep one questionTell me about the last time this went wrong. Walk me through it, step by step.The last-time question. One real episode beats ten opinions about "usually".
Say more about that.The probe. Then wait three seconds (→ Witness Interview).
If that were the cause, what would we expect to see in the data?
Who would disagree with this explanation?Triangulation. Now go talk to that person.
If you keep one questionWhat's the smallest pilot that would prove, or kill, this idea in six weeks?
Who has to reinforce this on Monday morning, and do they know it yet?
What do we measure, from what baseline, and who owns that number?
At the gate: what result means scale, what means adjust, what means stop?
What part of this will training not fix, and who owns that part?Say it in the recommendation. An unnamed cause becomes your failure.
Match the fix to the cause
The cause
What usually fixes it
Who usually owns it
Knowledge
Show them how: a short course, a job aid, a walkthrough
L&D
Skill
Practice with feedback on real cases, until it holds under pressure
L&D builds it, the manager keeps it going
Process
Fix the broken step
Whoever owns the process
Tools
Fix the system, or give everyone the tool the best performer built
IT, or the team that owns the system
Incentives
Change what the target or the scorecard rewards
Whoever sets the target
Expectations
Say it out loud, then write it into the role
The manager, and the leader above them
Outside the six
A staffing, hiring or culture decision
HR and leadership
Most cases need two or three rows at once. Put every owner's name in the recommendation, not only yours (→ 07).
Write the hypothesis before you build anything
The templateIf we intervention for audience, then metric will move from baseline to target within time, measured by source.
One filled inIf we pre-brief managers 48 hours before each announcement, then the share of teams hearing it from their own manager within a week moves from a third to 80% in one quarter, measured by a two-question pulse.
Done whenthe hypothesis written, threshold and gate signed.
Watch outScope creeping back toward "a full course". The targeted fix you designed in week one grows modules by week four.
When they push back
Sooner or later: "I don't have time for questions. Just build it." If you keep one line from this table, keep the last one. The parallel track costs the sponsor nothing, and it is the move that works when they outrank you by three levels.
They say
You say
Why it works
"Just build it."
Makes sense, you need this moving. What's driving the date?
A deadline you understand you can design around
"I don't have two days."
Two ways to go. We build exactly what you asked for, ready in three weeks. Or I spend two days with the team first and we build something narrower that's more likely to move your number. Your call.
People defend the plan they picked
"That's your job, not mine."
You own the date. I own whether this works. Give me two days and I'll own both.
Naming who owns what works better than defending the method
"We'll see how it goes."
Let's build it. Can we look at the number together in six weeks and decide if it needs a second round?
Turns a delivery into a case with a follow-up
"The date is fixed, full stop."
Then we run a parallel track. Design starts today, and in the same week I get three short conversations with the people who do the work.
You lose nothing and the build lands closer to the problem
And if they still say build it? Build it well, and write the hypothesis down anyway. If the number doesn't move, the written hypothesis opens the next conversation. If it moves, your diagnosis was wrong, and you learned something.
Run the intake
A request just landed? Fill this in during the conversation or straight after it. For a quick read, C and L are enough. If you can't put a number in the C line, that's your first question next time.
C
Put the number in: where it is, where it needs to be, what the gap costs.
L
U
What you saw, and the two sources that agree.
E
E, Execute: What's the smallest pilot that would prove, or kill, this idea in six weeks?
Write it as the hypothesis.
It stays in this browser. Nothing is sent anywhere.
Size two? Move it into the Case File template: your notes fill most of §1.
File 04
The Gadgets
When to use this: the conversation is drifting toward opinions and you need it back on evidence.
About 3 min
The 5 Whys Magnifying Glass
use during U
Reach for it when a symptom is on the table and everyone already "knows" the cause.
Ask
They answer
The symptom
We need communication training.
Why is communication the problem?
Status reports arrive late and leadership gets blindsided.
Why do reports arrive late?
PMs wait days for engineers' status data.
Why do they wait?
Engineers only update the tracker at sprint end.
Why only at sprint end?
The tracker demands 15 fields and duplicates the dev tool.
Why the duplication?
The two systems were never integrated after the reorg.
Verdict: a tools and process case. Training PMs to "communicate better" would have treated the symptom, not the cause.
Before you trust itWhen a why has two answers, follow both. If a why lands on a person, you asked the wrong why, so back up and ask what makes that mistake easy. Verify the final why against data: the 5 Whys finds a thread, not the whole truth.
The String Board
Stakeholder map · use during C and L
Four people, one question each, before you open any case.
The Sponsor
buys the solution
"What does success look like on your dashboard?"
The KPI Owner
feels the problem in their number
"Which number keeps you up at night, and where is it today?"
The Performer
does the work
"Walk me through where this gets hard."
The Reinforcer
the performer's direct manager
"What would you need to see to coach this every week?"
Red flagIf you can't name the Reinforcer, the case isn't ready to open.
The Fingerprint Dust
Job shadowing · use during L
Leaders design training around the workflow as it should be. Shadowing shows the one that actually runs.
1
Shadow two or three performers for an hour each.
Watch, don't demo.
2
Log every workaround.
A workaround is someone already solving a process failure, informally and invisibly.
3
Put the official process doc beside what you observed.
Mark every difference.
4
Note what the top performer skips.
What experts skip is usually what the process got wrong.
If the environment is broken, training people on the ideal process only adds frustration.
Shadow openlySay what you're looking for and why, check that now is a good moment, and show people what you noticed before you show anyone else. "I'm looking at the process, not at you" usually covers it.
The Witness Interview
use during C and U
The technique the rest of the toolkit depends on. Stakeholders are witnesses, not suspects. Ask for 30 minutes and say you're there to fix the process, not to grade anyone.
The move
Do this
Not this
Open, don't lead
What's getting in the way?
"Is it the new tool?" A leading question returns your hypothesis in their voice
Ask for the last time
Tell me about the last late report.
"Are reports usually late?" That gets a mood, not a fact
Listen for verbs
"They stopped raising risks in standup" is evidence
"The team is disengaged" is a verdict. Ask: "What did that look like?"
The three-second pause
After their answer, wait. Count it silently. What comes next is usually the real answer
Filling the silence yourself. "So would you say it's the tool?" hands them your answer to agree with
Mirror the last few words
"…so the handover just doesn't happen." "Doesn't happen?" They keep going. Borrowed from hostage negotiators
Paraphrasing at length. "So what you're saying is…" turns their story into yours, and they stop telling it
File 05
The 7 Rules of Evidence
When to use this: you're about to design, pitch or report on a solution and want it to hold up under cross-examination.
About 2 min
Seven rules borrowed from how product teams ship, grouped into three habits.
Habit one Prove it
01
Outcome over output.
Celebrate "onboarding time down 18%", never "10 modules delivered". (→ the Filter, File 02.)
Do this: write the outcome sentence before the kickoff.
02
Hypothesis-driven.
Not "we need leadership training" but a testable bet: if we give managers structured feedback practice, escalation cases drop 25% within three months, measured in the case tracker.
Do this: if you can't fill the template, you're not ready to build.
03
Data first.
Capture the baseline before anything ships. Screenshot the dashboard. That picture is your before-and-after story. A randomized control group is rarely possible in L&D, so use the realistic substitutes:
1a staggered rollout, pilot group against not-yet group
2a before-and-after on a stable metric
3a comparison against the trend line
Do this: no baseline, no launch.
Habit two Ship it small
04
Ship the smallest pilot.
Small enough to run fast, clear enough to give a verdict. Iterate on behavior, not on feedback forms.
Do this: one cohort, six weeks, one metric.
05
Give every asset a sunset date.
Treat learning like a product: discovery → validation → launch → growth → optimization → sunset. An outdated course costs credibility every day it stays open.
Do this: every asset gets a review date and an owner at birth.
Habit three Design the system, not the course
06
Design the reinforcement loop.
Training alone rarely survives the workplace. The manager's role, the tool changes and the workflow integration are part of the solution, or the solution fails.
Do this: name the Reinforcer in the plan (→ String Board). If that box is empty, stop.
07
User and buyer.
The user is the employee, the buyer is the business leader. If learners like it and no metric moves, you lose the budget. If the metric moves and learners hate it, you lose the users.
Do this: one sentence of value for each, side by side (→ Case File §2).
Part three
Write it up, then hand over what isn't yours
Files 06–07
File 06
The Master Case File
When to use this: the case needs the full investigation, and it's time to open a formal file.
About 2 min
Seven sections, each labeled with the CLUE letter it serves. Each answers a question a sponsor or steering committee will ask you.
You won't need all seven
Not every case faces a steering committee. The template is the full kit: keep the sections your case needs and delete the rest.
Every case
The Case · §1 The Motive · §7 Close the Case
When you build something
add §2 The Targets · §3 The Blueprint
Big casesreal budget, a pilot gate, several teams
add §4 The Investigation Plan · §5 The Chain of Command · §6 Risk & Resources
Strategy link · the gap and its KPI · root cause · ROI hypothesis · cost of inaction
C
§2
The TargetsAudience & success criteria
Persona · jobs to be done · value for employee and business · KPIs baseline → target → time · leading indicators · threshold
C, L
§3
The BlueprintSolution architecture
Components and format · reinforcement and practice loop · measurement · scalability and sunset plan
E
§4
The Investigation PlanRoadmap
Discovery → business case · Design → MVP · Pilot → the gate: scale / adjust / stop · Rollout · Optimization
U, E
§5
The Chain of CommandRACI
Approval, KPI definition, pilot launch, scale and sunset decisions × Sponsor, L&D product owner, HRBP, BU Lead, IT/Data, Finance
E
§6
Risk & ResourcesBudget & reporting
Investment · return · payback · risks with probability, impact, mitigation · reporting on KPI movement and adoption, never seat time
E
§7
Close the CaseThe one-pager you send the sponsor
What we found · what we did · what moved · what we recommend next
E
Close the Case takes ten minutes after the solution ships. It's why the business remembers you as "the people who fixed our discount problem", not "the training team".
File 07
Case Closed
When to use this: you want to see the whole method run once, then decide what you build and what you hand over.
About 4 min
The main case: the cascade that never landed
A composite of cases I've worked. The numbers show the shape, not one company's record.
Work it yourself first? The solved case stays folded until you're done.
Step
What it revealed
What changed
!The request
Middle managers can't communicate. We need communication training for all people managers. Leadership arrived with the verdict already written
Nothing yet
CClarify
The survey had just landed, "I understand where the company is going" at 41%, and the board asked why
A behavior instead of an adjective: each team hears the update from their own manager within a week
LLook
One department scores 78%. Its director pre-briefs her managers with a one-pager. Everyone else gets 60 slides at the same moment as their teams, no talk track, nowhere to ask, and nobody had said cascading was their job
Gap classified: process and expectations, plus a thin, real skill need
UUncover
Three managers on the June pivot all learned it in the all-hands with their teams. Two improvised in the hallway within the hour (Honestly, I know as much as you do.). Verbatims clustered on "my manager couldn't explain why"
Two independent sources, one cause
EExecute
One BU, one quarterly cycle: a pre-brief 48 hours before the announcement, a one-page talk track, a Q&A channel open for those 48 hours, one 90-minute practice session
Hypothesis signed: 80% of teams hold the conversation within a week, survey item +5 from 41%
✓Verdict
9 of 10 pilot teams held the conversation inside a week. The same survey item, asked again in a pulse, gained 6 points. The hallway questions stopped
Make the pre-brief permanent
The primary metric is a behavior you can count next month, not a survey you wait a year for. Closing the 37-point gap to the best department in one quarter isn't a realistic target.
Case closed
Found
Did
Moved
Next
Briefed last and armed with nothing
Four fixes, one of them training
9 of 10 teams, +6 points
Make the pre-brief permanent
Training was a quarter of the fix, and it worked because the other three-quarters shipped first.
The cold case files
In each case the fix sat upstream of skill. Take the question with you.
Guess the cause before you look.
Cold case 01
The negotiation course that was a signature
Asked for
Negotiation training; deals dying at the final stretch.
Question
"Show me the last deal we lost. Where exactly did it stall?"
Which gap?
Your verdict
Turned out
ProcessDiscount approval took five days and two signatures. Buyers cooled off.
Fix
A pre-approved discount matrix below a threshold. Cycle time down in a week.
Cold case 02
The empathy problem that was a scorecard
Asked for
Empathy training for support, after rough NPS.
Question
"What happens to an agent who spends four extra minutes with an upset customer?"
Which gap?
Your verdict
Turned out
IncentivesThe scorecard punished that. Handle time was 40% of the score.
Fix
Reweight it, add one quality criterion. Tone changed in two weeks.
Cold case 03
The Excel course that was a template
Asked for
Excel training for the finance team.
Question
"Could your best analyst do this fast? What is she doing differently?"
Which gap?
Your verdict
Turned out
ToolsShe had a formula-locked template. Everyone else rebuilt the report by hand, monthly.
Fix
Her template plus a 15-minute walkthrough video. Done that afternoon.
Cold case 04
The presentation training that was one slide
Asked for
Presentation skills for engineers; updates called unclear.
Question
"Think of an update that landed well. What did it look like?"
Which gap?
Your verdict
Turned out
ExpectationsNo agreed format. Every engineer invented their own.
Fix
A one-slide template, what changed / so what / the ask. Complaints stopped that quarter.
Cold case 05
The time-management course that was a calendar default
Asked for
Time-management training for new managers.
Question
"Walk me through your Tuesday."
Which gap?
Your verdict
Turned out
ExpectationsEvery invite defaulted to 60 minutes, and nobody believed they could decline.
Fix
25- and 50-minute calendar defaults, plus the director saying "decline freely" out loud, in public.
Cold case 06
The feedback training that was a permission setting
Asked for
Teach managers to discuss survey results.
Question
"What do you actually have in front of you when you sit down with the team?"
Which gap?
Your verdict
Turned out
ToolsManagers couldn't see their own team's dashboard. They were asked to discuss data they had never seen.
Fix
A permission change and a one-page talk track.
Cold case 07 · open for two quarters
The case we lost
Asked for
Sales enablement for a line nobody was selling.
Question
"What happens to a rep's payout when they sell the new line instead of the old one?"
Turned out
IncentivesThe new line paid less commission.
Fix
Not ours to make. Comp wasn't on the table, the training shipped as requested, and the number stayed flat, as the written hypothesis predicted. Two quarters later, when someone asked why the line still wasn't moving, that hypothesis was the only document in the room with an answer. Comp changed, and the training started working.
Who actually makes these fixes
The head of sales ops changed the approval rule. The support lead reweighted the scorecard. IT flipped a permission. L&D owned none of those fixes. The job was to find them and bring them to the right owner.
Bring the evidence, not the verdict.
"Here's what we saw in six tickets" beats "your scorecard is wrong."
Name one owner, not a department.
A fix assigned to Ops is a fix nobody makes (→ String Board).
Ask for the smallest change that tests the diagnosis.
For example, one field, one line in a policy or one calendar default.
Let them have the win.
The support lead changed the scorecard, so the credit is theirs.
When training is the answer
Build a course only if all three conditions below hold.
✓Build the course when
The bright-spot performer succeeds because she knows something the others don't
The environment would let everyone do the same
Nobody has tried teaching it properly
✕Don't when
This exact training has already run and the number didn't move
Everybody knows the process and nobody follows it
The managers who would reinforce it aren't reinforcing anything today
When all three hold, build it, and pilot it small anyway. By then the business has seen you turn down work that wouldn't help, so they trust you when you say a course will.
About the author
Hanna Petrasian
I've spent more than 12 years in learning, split between university lecturing and L&D management in software development and travel tech. I've built training strategies, launched a company-wide mentorship program from a blank page and run needs analysis with everyone from department heads to the C-suite. I run learning work the way product teams run products.
The method stands on older work: Thomas Gilbert's Behavior Engineering Model, Mager and Pipe's performance analysis, Dana and Jim Robinson's performance consulting, Geary Rummler's process view and Cathy Moore's action mapping.
What I care about most is the gap between "I want to learn this" and "this actually changed how we perform."
