Month three of a six-month cohort of twenty-four supervisors across three plants. On a Thursday the plant head announces an eleven-week shutdown of Line 2 for retooling, and the release letter covering two days a month is quietly withdrawn. HR’s message is reassuring: we will reschedule. Four months later, nine of the twenty-four have finished. The other fifteen did not resign and were not removed. Nobody asked them to do anything again, and nobody noticed for six weeks.
Cancelled dates do not kill a leadership cohort. The room does. Most programmes keep their whole evidence trail inside a session: attendance sheets, an exercise the facilitator observes, a discussion the cohort has together. Remove the room and there is no evidence, no contact point, no reason for the sponsor to look. The programme does not fail because the business changed. It fails because it had no existence outside a calendar slot.
Why the calendar carries more weight than it should
A leadership programme is designed for stable conditions. Same cohort, same venue, release letters signed in advance, a facilitator watching people practise. Every assumption there is a bet that nothing urgent happens for six months. In an Indian manufacturing or services business, that bet loses roughly once a year.
The evaluation literature explains why the loss is so complete. In a review of programmes at several dozen business schools, 70 percent of directors settled for evidence of positive participant reactions, and 63 percent for evidence of knowledge gained in the short term. None linked their programming to career trajectories, follower attitudes or organisation outcomes. If your only evidence is a reaction at the end of a session, the evidence exists only where the session happened. Cancel the session and you cancel the measurement with the teaching.
This is not only a provider problem. One longitudinal study of a leadership initiative across eight organisations started with 191 participants. It kept post-test scores for 147, but only 85, 44 percent, had a pre-test score and scores at both the three and six month points. Attrition at those measures was over 50 percent. That programme was never cancelled. A shutdown does to a cohort what ordinary life already does to its data, only faster and all at once.
So the design question changes. It is not how to protect the calendar. It is how to make the programme produce evidence in weeks when there is no room to meet in.
Three artefacts that replace the room
Each thing a session does well can be done by a piece of paper or a short meeting, if you write it that way in advance. Build the substitution list in month one, while everything is calm, not in the week the shutdown is announced.
| What the session does | Artefact that replaces it | Owner | How it fails |
|---|---|---|---|
| Facilitator observes a skill being practised | Fortnightly evidence note: one page on one real situation, what the participant did, what happened, what they would change. Line manager countersigns with one line on what they saw | Participant, with manager | Notes become a diary of feelings with no situation named |
| Case study and peer practice | Live task drawn from the disruption itself: the bridge roster, the redeployment list, the message to the affected crew. Deliverable goes to the plant head | Cohort, in pairs across sites | Task is invented for the programme instead of taken from the business |
| Cohort rhythm and peer pressure | Fixed 30-minute sponsor slot every fortnight, same day and time, two participants presenting their notes | Sponsor or named deputy | Slot moves whenever the sponsor is busy, then disappears |
| Attendance record | Completion ledger counting artefacts, not hours | Programme manager | Ledger measures submission, not quality |
Three rules make the artefacts hold. First, the evidence note names a situation that happened, not a skill that was learned. “On Tuesday I told the shift the line would run at 60 percent for three weeks, and two operators asked for transfer” is evidence. “I practised difficult conversations” is not. Second, the live task needs a real recipient who can reject it. A task marked by the facilitator is a classroom exercise with a longer walk. Third, the sponsor slot never moves: thirty minutes, fortnightly, chaired by a named deputy when the sponsor cannot attend. Once that slot starts sliding, the cohort has lost its only witness.
The order the work happens in
Before the disruption, in month one. Write the three artefacts and test them with one full fortnight and no session. If the cohort cannot produce evidence in a fortnight without a room, the artefacts do not work yet and you have found out cheaply. This is also when you take the programme design decision about what is core and what is optional, since you cannot make that call while a plant is closing a line.
The trigger week, within 72 hours. One meeting with the sponsor, one page out of it. Name what is suspended (content modules), what is replaced (observation becomes evidence notes, case work becomes the live task), what is protected (the sponsor slot and the ledger). Give the suspension a date, even provisional. A programme paused “until things settle” has been cancelled in polite words, and the cohort knows it.
The bridge, six to eight weeks. Keep the fortnightly rhythm at low intensity and change the delivery plan, not the learning outcomes. Where a participant’s manager has also been pulled into the disruption, a short coaching arrangement for that pair does more than a catch-up workshop, because it needs no room and no release letter. Do not compress the lost sessions into a two-day catch-up when the plant reopens. A 2025 meta-analysis of applied classroom studies, 22 reports and 31 effect sizes, found distributed practice beat massed practice at an effect size of 0.54, with the advantage growing at longer retention intervals. Two days of catch-up in week twelve will feel productive and will not survive to week twenty. Four short fortnights will.
The restart test. Before the workshop calendar resumes, check two conditions. Every remaining participant has at least three countersigned evidence notes, and the live task produced a decision the plant used. If both hold, the restart continues at the level the cohort reached. If either fails, you are starting again and should say so, and price it as a fresh start rather than pretending the first three months carried over.
A worked example
Illustrative example. An auto-components group runs a six-month leadership programme for 24 supervisors across three plants, priced at about Rs 18 lakh. In week eleven, one plant announces a ten-week shutdown of a line and pulls its release letters.
Under the old design the reschedule would have pushed the cohort into the next financial year, and honest completion would have been nine or ten of 24.
Under the artefact design, 21 of 24 stay on the ledger (two transfers and one exit are genuine losses). The sponsor slot holds eight of eight fortnights, twice chaired by the plant HR head while the plant head sat in a customer audit. Nineteen participants finish with three or more countersigned notes. The live task, a redeployment and retraining plan for 34 operators, is adopted with two changes by the plant head, which buys the sponsor’s attention for the rest of the year. The restart runs two sessions instead of four and ends at month seven, six weeks late. Extra facilitation for the notes and live task comes to about Rs 90,000, roughly 5 percent of programme value, against a cohort that would otherwise have been written off.
These figures are constructed for illustration, not a client result. What generalises is the ratio: a small extra facilitation cost, spent before the disruption, protects most of a programme’s value.
The honest limit
This works when the sponsor survives the disruption and the disruption has an end date. A shutdown, a line transfer, a relocation, an attrition wave all leave a sponsor in place and a date on the horizon. A merger with an owner who arrives with their own framework does not. There the artefact set does not save the cohort. Pause honestly, hand the evidence ledger over in the transition pack, and renegotiate with the incoming leadership rather than defend a calendar they did not sign.
There is a second limit worth naming. A four-year study across an organisation of 9,471 people found that as the proportion of leaders completing a development programme rose across 19 service units, staff absenteeism and formal grievances rose too. New expectations create new demands, and a workforce facing redeployment may experience a supervisor’s fresh approach as more pressure, not less. So the sponsor slot needs one more standing question: what is the new behaviour costing the team this fortnight. A cohort that keeps practising difficult conversations through a shutdown without asking that is not resilient, it is oblivious.
The same reasoning applies to ordinary transfer, which we set out in why behavioural training fades. A multi-site programme meets a version of this without any shutdown, which is what running one programme across twenty locations covers.
What to do on Monday
Open the cohort file and answer three questions in writing: what evidence exists outside a session, who countersigns it, which recurring slot belongs to the sponsor. Book the sponsor slot for the next six fortnights at a fixed time with a named deputy. Then schedule one fortnight with no session at all and see whether the cohort can still evidence its work.
If they cannot, you have four weeks of warning and a cheap fix. If you wait for the shutdown notice, you will be defending dates instead of teaching people, and the dates will lose.
Sources
- Leroy, Anisman-Razin and Detert, Leadership Development Is Failing Us, MIT Sloan Management Review, Winter 2024: 70 percent of programme directors settle for reaction data, 63 percent for knowledge gain, none linked to career, follower or organisation outcomes: https://sloanreview.mit.edu/article/leadership-development-is-failing-us-heres-how-to-fix-it
- Packard and Jones (2015), An outcomes evaluation of a leadership development initiative, Journal of Management Development 34(2): 191 participants, 85 (44 percent) with pre-test and three and six month scores, over 50 percent attrition at those measures: https://theacademy.sdsu.edu/wp-content/uploads/2015/04/outcomes-eval-lia-packard-loring-2015.pdf
- Mawson and Kang (2025), The Distributed Practice Effect on Classroom Learning, Behavioral Sciences 15(6):771: distributed over massed practice, d = 0.54, 22 reports and 31 effect sizes, larger effects with longer retention intervals: https://pmc.ncbi.nlm.nih.gov/articles/PMC12189222
- The perils of leadership development (2024), Work and Stress: programme spread across 19 units of a 9,471-person organisation predicted higher absenteeism and grievances: https://www.tandfonline.com/doi/full/10.1080/02678373.2024.2332174
