At 09:10 on a Monday, a production supervisor named Ravi stands in a meeting room at his own plant with sixteen people in front of him. He is the internal trainer for the shift handover module, and he has a 94-page facilitator guide open on the table. Page one is a learning objective. Page eleven is a slide-by-slide script. Page forty-one is the exercise he is supposed to be running now. He has not looked at the guide since the certification workshop four months ago, and he will not look past page forty-five today.
The cohort still learns something. It is just not the thing the design intended, and nobody in the room, including Ravi, can tell you which parts of the course actually happened.
That is the handover problem. Not the content, not the trainer’s capability. The document is the wrong document.
Why the binder exists at all
A facilitator guide is usually written for one reader: the person who signs off on quality. It has to look complete, defensible and replicable, because a client or an internal sponsor will page through it before approving the spend. So it accumulates learning objectives, session plans, theory notes, answer keys, slide thumbnails, timings, assessment rubrics and a reading list. Every addition makes it a better record of the design.
None of that helps a man standing up with sixteen people who want to finish before the tea break.
Two things go wrong. First, attention. A guide that contains everything gives the trainer no priority order, so in the room he falls back on what he already knew as a supervisor. The course drifts back toward the job, which is comfortable and familiar and roughly half right. Second, timing. A 94-page document takes a full day to read, and no plant releases a supervisor for a day to re-read something he was already certified on. The guide is therefore never read again after certification. The trainer delivers from memory plus the slide deck, and the deck was never a plan.
This is why the handover is the part of a train-the-trainer programme that decides whether the programme worked. Certification proves the trainer can deliver once, in front of a trainer. The handover decides what gets delivered in week nine, alone.
What the trainer actually needs, in the order he needs it
Four artefacts, four pages, and one two-hour dry run. Nothing else survives contact with a live cohort.
| Artefact | Length | Used at | Job |
|---|---|---|---|
| Run sheet | 1 page | In the room, on the table | Tells the trainer where he is and what happens next |
| Failure scripts | 1 page | In the room, when it stalls | Gives a rehearsed response to the three predictable derailments |
| Observation checklist | 1 page | During practice, and after | Records whether the behaviour actually appeared |
| Cohort return | 1 form | Within 48 hours of the session | Tells the designer what really ran |
The dry run is the fifth item and it is not a document. It is the trainer delivering the first forty minutes to two colleagues, one of whom plays the difficult participant. An hour and a half of watching plus a half hour of correction.
The run sheet
One page, landscape, columns: clock time, minutes, what the trainer says or does, what participants do, and the one judgement call for that block. Not the content. The content lives in the exercise sheets and the deck. The run sheet is a map with a clock on it.
“The one judgement call” is the column that earns its place. For each block the design names a single decision the trainer will have to make live, and the default if he freezes. For the shift handover module it reads: block 3, 22 minutes, participants list the exceptions from yesterday’s shift on a card, trainer sorts the cards into “named owner” and “vague”. Judgement call: if fewer than half the cards have an owner, slow down and take two examples from the floor before moving on. Do not lecture the exceptions framework.
That single line replaces about six pages of facilitation notes. It also encodes the failure the design is trying to prevent, which brings us to the scripts.
The three failure scripts
Every cohort produces the same three objections, and an unprepared internal trainer handles all three badly, because he is a colleague and he wants to be liked. Write the line, write why it works, write the trigger that tells you it is happening. One page, three blocks.
The sceptic. Trigger: “we already do this on our shift.” Line: “You probably do. Show me the last handover note where an open exception had a name and a time on it.” Then wait. The script works because it converts an opinion into an artefact request, and because the trainer is not defending the course, only the standard.
The defence of the old method. Trigger: “our shift is different.” Line: “It might be. What is the exception that got lost last month?” The script works because it accepts the premise and moves to evidence. Most shifts do have that exception, and the group names it themselves.
The clock collapse. Trigger: the exercise is running twelve minutes late and the trainer starts talking over the pairs to catch up. Line, to himself: “cut block five, keep block three.” The design should name which block gets cut. A trainer who improvises the cut usually cuts the practice, because practice is the part that looks like wasted time.
That last one is the most valuable line in the handover, and it is almost never written down. It tells the trainer, in advance and with the designer’s authority, that the discussion block is expendable and the rehearsal is not.
The observation checklist
Six items maximum, each scored yes or no, each anchored in something a person could see from the back of the room. This is the same discipline as writing behavioural anchors a manager can rate, and the calibration matters more than the wording.
The checklist is used during the practice block, by the trainer or a co-trainer, on two or three participants, not all sixteen. Its purpose is not to grade people. It is to answer one question at the end of the session: did the behaviour the course exists for actually appear in the room?
Calibration is where programmes quietly fail. Two trainers watching the same role-play will score it differently unless they have agreed on what counts. Research on frame-of-reference training for raters is relevant here. Tsai, Wee and Koh, across five studies with 1,143 participants, found that a restructured frame-of-reference training produced an average improvement in rating accuracy at least twice as large as typical frame-of-reference training, and that reducing the anchoring pull of a rater’s own self-rating was the mechanism. The practical translation is short: before the cohort, the trainers score the same five-minute recorded segment independently, then argue until the disagreements are resolved by pointing at the anchor, not at impressions. One hour. Do it once per cohort, not once per year.
The cohort return
One form, filled by the trainer within 48 hours: which blocks ran to time, which were cut, which script was used, how many participants passed the checklist, and the one thing that did not work. Five questions, three minutes.
This is the artefact that makes the handover a loop rather than a handover. Without it the designer never learns that block five has been cut in eleven of twelve cohorts, and keeps writing a guide for a course that stopped existing in month two.
A worked example
Illustrative example. An auto-components manufacturer ran a train-the-trainer programme for twelve internal trainers across three plants, delivering a supervisory module to about 300 people a year. The original handover was a 96-page guide plus a two-day certification workshop.
Nine months later, an audit of ten sessions found that the practice block had run in four of them, and the exception-naming standard appeared in the participants’ own notes in three. Trainers reported reading an average of about forty pages of the guide, all of it during certification.
The redesign kept the certification workshop and replaced the handover with the four pages above and a two-hour dry run per trainer. Cost: roughly 30 hours of trainer time across the twelve, plus a day of design time. In the next six months, the practice block ran in eleven of twelve observed sessions, and the cohort return came back from ten of twelve trainers, mostly within three days.
These figures are illustrative, not a client result. The transferable point is that the fix was subtraction plus one artefact that forces a reply, not more training for the trainers.
The honest limit
A good handover cannot fix a badly chosen trainer. If the internal trainer has no standing with the group, or is the supervisor everyone avoids, a one-page run sheet will not create credibility. Selection comes first, and the training solutions conversation should start there rather than with the document.
It also does not fix the manager. The research on transfer is blunt about this: supervisor support, coaching and feedback sit consistently on the positive side of the evidence, while post-training interventions such as goal-setting after the event have not shown a reliable independent effect in the meta-analytic work summarised by the Quality Improvement Center for Workforce Development. A handover document governs what happens inside the session. What happens in the following three weeks is decided by whoever reviews the work, which is a different conversation, and one we have made before in why behavioural training fades.
One more limit worth stating: the evidence base on transfer is thinner than the confidence of most vendor decks. Thalheimer’s 2020 review of the transfer research estimates that well over 80% of studies measure learners’ perceptions of transfer rather than transfer itself. Treat any claim that this or that design “guarantees” transfer with suspicion, including claims in this article. What the four pages give you is a way to see whether the course ran as designed. That is a precondition, not a result.
What to do on Monday
Take the guide for whichever module your internal trainers deliver most often. Count its pages. Then open a blank document and write the one-page run sheet from it: clock time, minutes, what the trainer does, what participants do, one judgement call per block. Where you cannot fill the judgement call column, you have found a design gap, not a documentation gap.
Then write the three scripts. Do it by asking one of your trainers what actually goes wrong in the room, not by imagining it. Add the observation checklist of six yes/no items and the five-question cohort return. Run the next cohort with only those pages in the room, and read the returns for the two cohorts after that.
If you want a second pair of eyes on the trainer selection and the dry run, coaching for the internal trainers is usually the cheaper half of the fix. The skill of rating a behaviour and the skill of naming a standard out loud are both learnable in a day, and both are worth more than another fifty pages of facilitator notes.
Sources
- Tsai, Wee and Koh (2019), Journal of Organizational Behavior, on restructured frame-of-reference training and rating accuracy across five studies, N=1,143: https://ink.library.smu.edu.sg/cgi/viewcontent.cgi?article=4081&context=soss_research
- Quality Improvement Center for Workforce Development, umbrella summary of the training transfer evidence, including supervisor support and the weak result for post-training interventions: https://www.qic-wd.org/umbrella-summary/training-transfer
- Thalheimer (2020), Factors That Support Training Transfer, on the weak research base and the reliance on learner perceptions: https://www.worklearning.com/wp-content/uploads/2020/01/Transfer-of-Training-Quick-Research-Review-2020-v1a.pdf
- SessionLab, on what facilitator guides contain and how they are used at scale: https://www.sessionlab.com/blog/how-to-create-more-effective-facilitator-guides
- Blume, Ford, Baldwin and Huang (2010) transfer meta-analysis, as summarised in the QIC-WD summary above.