Six annotated excerpts from a leadership training discovery call: what the client said, the wrong turn, and the exact words that redirect it.
Every article about discovery calls gives you a list of questions. Almost none of them show you the part that is actually hard, which is what you say after the client answers.
What follows are six excerpts from a discovery call for a leadership programme. They are composite reconstructions, not recordings — assembled from the pattern of conversations I have run and sat in on, with nothing traceable to a client. I have kept them in dialogue because the useful information is in the wording, and wording does not survive being summarised into a bullet.
At each excerpt there are three notes: what the client is actually telling you, the wrong turn I have taken myself, and the sentence that redirects without turning the call into an argument.
The client here is a VP of Operations at a mid-size business services firm. The request, as it arrived: “We need a leadership programme for our new managers.”
VP: So we’ve got about forty first-line managers, most of them promoted in the last eighteen months, and they’re struggling. We want a leadership programme. Six months, cohort-based ideally.
Me: That’s useful, thank you. Before we get to shape — when you say struggling, what does that look like on a Tuesday? What are you seeing or hearing that made this a priority now?
VP: Honestly? My inbox. Everything comes to me. Decisions that shouldn’t need me.
What he actually said. He did not describe a capability gap. He described his own workload. That is not a criticism of him. It is the most common way a leadership brief arrives, because the person who feels the pain is rarely the person the pain is about.
The wrong turn. Writing down “escalation, decision-making confidence” and moving on to logistics. The brief now looks specific, so it feels like progress. It is not: I have accepted his causal claim without testing it, and every module I design after this point is downstream of an assumption I never examined.
The redirect. “What does that look like on a Tuesday?” Ask for the observable instance, not the category. Categories are conclusions. Instances are evidence. If a client cannot produce an instance, the problem may be real but nobody has yet seen it happen, which is worth knowing before you price six months of work.
Me: Of the forty, are there any who don’t escalate? Same role, same conditions?
VP: Sure. Four or five are excellent. Priya’s team barely comes to me at all.
Me: What’s different about Priya?
VP: She was here before the restructure. She knows where the bodies are buried, so she’s not scared of getting it wrong.
What he actually said. He answered the diagnostic question and then, unprompted, gave me the diagnosis. Not scared of getting it wrong. The others are. That is a statement about consequences, not about skill.
This is the fork Robert Mager and Peter Pipe built their whole performance-analysis flowchart around. The chart runs the skill question first — is it a skill deficiency?, and if so, did they used to do it? (arrange review) or not (arrange formal skills training). But it does not stop there. It carries on into a second set of nodes that have nothing to do with instruction: is non-performance rewarding? Is performance punishing? Are there obstacles? — with the corresponding actions being to remove the punishment, arrange a positive consequence, or remove the obstacle (Analyzing Performance Problems, 2nd ed., 1984, Mager & Pipe).
“Performance punishing” is the branch this conversation just landed on, and there is no training solution at the end of it.
The fork the second excerpt turns on
Only the left-hand branch ends in a curriculum. Robert Mager and Peter Pipe’s flowchart runs the skill question first, then carries on into a second set of nodes that have nothing to do with instruction. Most leadership briefs arrive having assumed the left branch and never tested it. Structure after Mager & Pipe, Analyzing Performance Problems (2nd ed., 1984).
The wrong turn. Hearing “she’s not scared” and designing confidence-building modules. This is the subtlest error in the whole call, because it produces a programme that is coherent, well-reviewed and useless. You cannot build individual confidence against a live organisational consequence. The manager is not miscalibrated. They are correctly calibrated to an environment you did not change.
The redirect. “Has anyone here been visibly corrected for a decision that went the wrong way?” Ask about the instance, in public, with a name attached to the memory. If the answer arrives fast and specific, you have your diagnosis and it is not a curriculum.
Me: Suppose this works perfectly. All forty start deciding at their level. Whose job gets harder?
VP: (pause) Mine, I suppose. I’d have to live with decisions I’d have made differently.
Me: How would you feel about the first one that goes badly?
VP: (laughs) Ask me in six months.
What he actually said. He named the cost, he located it in himself, and he told me he has not decided to pay it. That pause is the single most valuable four seconds in the call.
Ronald Heifetz and Donald Laurie’s distinction is the right frame here. Adaptive work, they write, “is required when our deeply held beliefs are challenged, when the values that made us successful become less relevant, and when legitimate yet competing perspectives emerge” — and, crucially for the design, “solutions to adaptive challenges reside not in the executive suite but in the collective intelligence of employees at all levels” (The Work of Leadership, Harvard Business Review; reprint carrying the 2001 HBSP copyright).
A programme that trains forty managers to decide, while the person who has to tolerate their decisions is not part of it, has intervened at the wrong level. That is not a scheduling detail. It is the difference between the programme working and the programme producing forty people who now decide, get overruled twice, and stop.
The wrong turn. Letting the laugh close the topic. It is a friendly moment and it invites you to move on. Moving on means the trade-off never gets named, which means it never gets owned, which means it surfaces in month five as “the programme didn’t stick.”
The redirect. “I want to put that on the record, because it changes what we build. Would you be willing to be in the room for part of this?” Asking the sponsor to participate is the cheapest test of whether the organisation has actually decided to change. A yes reshapes the design. A no is information you need before you price it, not after.
Me: Six months after this finishes, what would you see or hear that would tell you it worked?
VP: We’d have run all four cohorts, everyone would be through it.
Me: That’s the plan working. What’s the result working?
VP: (pause) Fewer things in my inbox, I suppose. And I’d want the regional managers to stop saying their teams can’t cover for each other.
What he actually said. His first answer was a completion metric. His second — the one after the pause — contained two observable things and a stakeholder group he had not mentioned once until that moment.
Ask this question at the end of the call and you get the first answer. Ask it in the first twenty minutes and you get the second, because there is still time for him to correct himself and no proposal on the table for him to be polite about.
The wrong turn. Accepting “fewer things in my inbox” as the measure. It is directional, it is his, and it is unmeasurable in any way that survives contact with a CFO. Volume in one person’s inbox moves for a dozen reasons that have nothing to do with forty managers.
The redirect. “Which of those two would you be willing to have counted before we start?” The word before is doing the work. Any measure that cannot be baselined is a story you will be telling from memory in nine months. Once you have the baseline you have the beginning of a chain that runs to a business case, which is where this eventually has to land.
Me: If we build this well and the behaviour still doesn’t shift, what will the reason have been?
VP: The regional structure. Two of the four regions run their own P&L and their MDs won’t be told anything by this office.
Me: Are those regions in scope?
VP: They have to be. It’d look political otherwise.
What he actually said. Half the population sits under leaders who will not reinforce any of this, he knows it, and the reason for including them anyway is optics rather than design.
Most clients answer this question immediately and accurately. They have simply never been asked, because most vendors do not want the answer written down where it can be produced later.
The wrong turn. Either of the two obvious ones. Refusing the work is self-indulgent — a programme reaching two of four regions properly is still worth building. Accepting it silently is worse, because you have now agreed to be measured against an outcome you have private evidence you cannot produce.
The redirect. “Then I’d like to write that into the design as a stated assumption, and propose we report those two regions separately.” Segmenting the measurement is the move. It costs nothing at design time, it is completely defensible, and if the split shows up in the results you have the strongest possible evidence for the structural conversation he has been avoiding — evidence he can use, produced by his own programme.
Me: Based on this, I don’t think a six-month cohort programme is the first thing I’d build. I think about half of what you’re describing is a consequence problem, and a leadership programme will not touch it.
VP: So you’re saying don’t do the training.
Me: I’m saying do less of it, and do one other thing alongside it. Roughly eight weeks with the forty on decision rights and delegation, and a separate short piece of work with you and the four regional MDs on what happens when one of those decisions goes wrong. The second one is smaller and it’s the one that determines whether the first one holds.
What he actually said — in the words that came next, which are not in the excerpt — was yes. Not because the recommendation was clever, but because it was the first time in the procurement process that anyone had proposed spending less of his money.
The wrong turn. Delivering this as a critique rather than as a design. “Training won’t fix this” is true and it is useless. It leaves the client with a problem, no plan, and the impression that you have talked yourself out of the work. Always arrive at the smaller recommendation with the alternative already attached.
The questions are not the hard part. These are:
| When they… | Say |
|---|---|
| Describe a category (“accountability”, “ownership”) | “What does that look like on a Tuesday?” |
| Name a good performer | “What’s different about them?” — then stop talking |
| Explain away the good performer | Write down that explanation verbatim. It is usually the diagnosis. |
| Answer the success question with a plan | “That’s the plan working. What’s the result working?” |
| Name a measure you can’t baseline | “Which of those would you be willing to have counted before we start?” |
| Name a barrier they can’t move | “Then I’d like to write that in as a stated assumption, and report that group separately.” |
| Laugh off the cost to themselves | “I want to put that on the record, because it changes what we build.” |
| Ask you to proceed anyway | “That’s your call to make. I’d just like it written down that we made it.” |
None of these contradict the client. Every one of them moves the conversation from a conclusion to an instance, which is the only move that matters in this call.
It would be convenient to tell you that running a conversation like this produces better training outcomes. The strongest available evidence does not say that, and you should know it before you build a practice on it.
Arthur, Bennett, Edens and Bell’s meta-analysis of training effectiveness found that only 6% of data points (22 of 397) reported conducting a needs assessment at all. That figure gets quoted a lot. What gets quoted far less often is what they found next: “Contrary to what we expected — that implementation of more comprehensive needs assessments… would result in more effective training — there was no clear pattern of results for the needs assessment analyses.” The authors add their own caution that these particular analyses rested on four or fewer data points and “should be cautiously interpreted” (Journal of Applied Psychology, 88(2), 234–245, p. 242).
So: the practice is rare, and the evidence that doing more of it improves training outcomes is thin and unresolved.
I still think the conversation is the highest-value forty minutes in the engagement, and here is the honest version of why. It is not that more analysis makes a course work better. It is that this conversation changes what gets built, who is in the room, and what gets counted — and two of those three are decisions made before any training exists to be measured. Excerpt 3 changes the participant list. Excerpt 5 changes the measurement design. Excerpt 6 changes the intervention itself. A study comparing training-with-needs-assessment against training-without would not capture any of that, because in the cases where it works best there is less training to compare.
A client who checks this will find the null result. Better that they find you said it first.
Those five lines are the diagnosis. Everything after them is design — including which framework the symptom actually points to, which is a separate decision with its own failure modes, and the measurement plan the baseline in line four eventually feeds.
If you leave a discovery call without line three, you have not run a diagnostic conversation. You have run a briefing, politely.
Try it on a live brief. The Diagnostic Conversation Coach is free, needs no signup, and runs this conversation against a gap you describe in your own words — including the branch points above and what each one commits you to measuring. If it doesn’t sharpen the way you run a discovery call, nothing else here will interest you.