An apprentice, a first-year teacher, a new hire out of a training programme. The institution that recruited them has done its part, and what happens next is decided by whoever they report to.
A substantial share of people who leave a trade or a profession in the first two years do so because of the workplace rather than the work, and the decisive period is short.
The supervisor was chosen for something else
The person assigned is generally the most technically capable available, which is the promotions error and produces a predictable set of difficulties.
Somebody excellent at the work frequently cannot describe how they do it, because the knowledge has become automatic. They will demonstrate and expect observation to suffice, which works for the physical component and not for the judgment.
They may also experience the apprentice as a drag on their own output, which in a firm that measures individual productivity is a rational reading and is the single most common reason supervision is thin.
Which means the organization has to make the development part of the supervisor’s actual job rather than something added to it, or the supervision will be as good as somebody’s goodwill permits.
Say the reasoning out loud
The single practice that most improves the transfer, and it costs no additional time.
An experienced person makes dozens of small judgments in a day that a learner cannot see: why this approach rather than that one, what they noticed that changed the plan, what they were checking for.
Narrating that while working converts observation into instruction. I am doing it this way because of what happened on the last one of these. I am checking this first because when it fails it fails here.
Without it, the apprentice learns the procedure and not the judgment, which is precisely the distinction the mentorship chapter identifies as the thing that cannot be transferred by documentation.
Let them decide something
Judgment forms only from making calls and living with the results, which is the stretch-assignment principle at the smallest possible scale.
An apprentice who executes instructions for two years has two years of experience at following instructions. One who is given a bounded decision — you choose the approach on this one and tell me why — begins accumulating the thing that distinguishes a tradesman from a pair of hands.
The bound has to be real and stated. Where safety, cost, or consequence makes independent judgment inappropriate, say so plainly rather than leaving somebody to discover the limit by crossing it.
Correcting without discouraging
Learners make mistakes continuously and how those are handled determines whether somebody stays.
The version that works is specific, immediate, about the work rather than the person, and delivered without an audience where an audience would humiliate. That is the feedback standard and it applies here with more force because the person has no accumulated confidence to absorb a poor delivery.
What does no good is the correction that carries an implication about whether they belong in the trade. Somebody hearing that in their fourth month will believe it, and the ones who believe it soonest are frequently the ones who would have been best.
Where a mistake was genuinely serious, the useful addition is what you did the first time you got this wrong, which is nearly always available and changes the register entirely.
The conduct question
Worth naming directly, because it drives departures that get attributed to unsuitability.
Some workplaces carry a tradition that new people are given a hard time, and the people administering it generally regard it as harmless and as something they went through.
A firm that tolerates it is running a filter that selects for tolerance of that treatment rather than for capability, and it is losing people at a rate it has never measured because those people do not explain why they left.
The standard is set by what a supervisor lets pass in front of them, per the culture chapter, and a single instance permitted establishes what the place is.
Ninety days decides it
What determines whether somebody stays is settled early and is mostly not about the work.
Whether anybody learned their name in the first week. Whether they had something to do rather than standing available. Whether one person was clearly responsible for them. Whether anybody asked how it was going and waited for the answer.
Each costs nothing and each is routinely absent, because everybody is busy and nobody owns the question.
For an employer in a place with a genuine shortage of trained people, losing an apprentice in month four is the most expensive avoidable outcome available, and it is generally caused by something that would have taken ten minutes to prevent.
Edited by Patrick J. Wolf, PhD