They could run the shift, but not the hard conversation
Newly promoted first-line managers were leaving fast, so before I designed anything I coded their exit interviews and found the real gap, not competence, but the nerve for a hard conversation.
The problem
Newly promoted first-line managers were leaving the role, or being performance-managed out of it, within a year of stepping up. That is the design baseline this case study runs on, and it is treated as a starting assumption throughout, never a measured result.
The organisation had already had one guess at why. The previous management course taught rostering, KPI tracking and process compliance, built from a generic competency framework rather than anything specific to this workforce. It failed, and it failed on skills nobody was short of. Nobody had checked the diagnosis against the evidence sitting in the exit interviews, which had been coded by nobody, for years.
So I read them properly. Coded thematically, across a twelve-month window, one cause dominated everything else: difficult conversations and performance reviews. Not rostering. Not KPIs. People who could run a shift perfectly well, and could not sit a direct report down and say something hard to hear.
The learner and the constraints
These are people freshly promoted from the shop floor or the warehouse, often still covering operational shifts alongside the new role. Mixed prior management experience, some none at all. There is no version of this cohort with spare capacity to sit through another classroom block on interpersonal skills it does not yet know it is missing.
Whatever I built had to live inside their working week, not beside it. And it had to answer for the previous course's failure, not repeat its shape with different slides.
The question I set myself: how do I build something that tells a manager the truth about their own readiness, twice, without either administration letting them hide behind self-image?
The approach, and why
I did not run another competency audit. I built the programme from what the exit interviews actually said.
- Code the evidence before designing against it. Twelve months of exit interviews, coded thematically, sorted into technical/operational versus interpersonal/managerial. Every design decision after this traces back to a coded theme, not a hunch, and that traceability is the artefact's spine.
- Sequence against the moment, not a term calendar. I placed each touchpoint against the 5 Moments of Need instead of a single workshop block. The readiness diagnostic sits at the New moment, before the manager needs the skill. The branching scenario sits at Apply, close to when a real difficult conversation is about to happen, not weeks removed from it in a classroom.
- Scaffold the coaching skill on GROW. Goal, Reality, Options, Will. Chosen because this audience will recognise the structure without a glossary, and a portfolio artefact's credibility depends partly on that.
- Teach the decision, not the definition. The difficult-conversation scenario is a branching exercise built on GROW, with a full decision-consequence-reflection cycle. A weak choice does not get flagged wrong. It plays out, the direct report gets defensive or disengages, and the learner has to work out why.
- Make one instrument do two jobs. The manager-readiness diagnostic is the same instrument, pre and post. Before the programme it asks what a manager believes about their own readiness. After it, the manager scores the same items against a real conversation they have just handled and builds an action plan from the gap. One tool, asked twice, carrying both the readiness read and the Kirkpatrick Level 3 and Level 4 evaluation, with no sixth artefact bolted on to prove it.
Artefacts
A findings-to-design traceability document tracing every coded exit-interview theme to a design response, a cohort curriculum map plotting each touchpoint against the 5 Moments of Need, the full branch map and one complete decision-consequence-reflection cycle for the difficult-conversation scenario, and the manager-readiness diagnostic in both its pre- and post-programme forms. All available to walk through on request.
The outcome
The honest number here is the one we started with. Forty percent of newly promoted managers gone within a year is the design baseline, illustrative and amber, not a confirmed statistic and never presented as one. Everything the programme is built to move against that baseline is a design target, marked in amber until a real cohort runs it.
What I took from it
The previous course did not fail because it was badly delivered. It failed because it answered a question nobody had asked, and the real question was sitting uncoded in a filing cabinet the whole time. You do not fix a manager who freezes in a hard conversation by teaching them KPIs better. You fix it by rehearsing the conversation itself, and by asking them, honestly, twice, whether they are ready. That is the difference between a course that ticks a box and a programme that changes what a manager does the next time someone needs to hear something hard.
Further reading: measuring behaviour, not completion, action mapping for regulated learning.