The Shadowing Week SOP
Purpose
To define a repeatable one-week block of paired field work used whenever a person takes on a scope they have not held before, ending in a written competency decision rather than an impression. The week exists to move someone from watching to leading on a defined scope, with evidence at each transition, so the shop can say what a person is cleared to do instead of guessing.
The governing principle: the five days are the container, the evidence is the gate. A person advances from observing to assisting to leading because they demonstrated the prior mode, not because it is Wednesday. A week that ends with someone still in assist mode is a successful week that produced accurate information.
Scope
Applies to any existing employee taking on a scope they do not currently hold:
- A new service line or equipment class the shop has added.
- A promotion to lead, crew chief, or first solo dispatch on a job type.
- A tech moving from residential to commercial work or the reverse.
- A return after an absence of six months or more.
- A remediation block after a quality or safety issue, where the cause was capability rather than conduct.
Does not apply to new-hire orientation, which has its own program covering compliance, paperwork, culture, and truck setup before any shadowing begins. See related: Technician Week 1 Field Shadowing Program. A new hire who completes orientation and later takes on an additional scope comes back through this SOP.
Does not apply to conduct problems. If the issue is willingness rather than capability, a shadowing week will produce a week of good behavior and no information.
Roles and responsibilities
| Role | Responsibility |
|---|---|
| Shop lead or owner | Confirms the trigger, defines the scope in writing, names the host, approves the schedule impact, signs the final competency decision |
| Host tech | Runs the week, controls mode transitions, logs evidence daily, raises a stop by end of day 3 if the week is not viable |
| Shadowing tech | Does the work, prepares each morning, requests the next mode when they believe they are ready |
| Dispatcher | Protects the pairing, sources jobs matching the scope, does not split the pair for coverage without lead approval |
| Second signer | Any qualified tech other than the host who observes the day 5 demonstration for a high-consequence scope |
Procedure
1. Confirm the trigger and write the scope
Before scheduling anything, write one sentence naming what the person will be cleared to do at the end: the job types, the equipment class, and whether the clearance is solo or paired. Vague scope produces a vague sign-off, which is worse than no sign-off because it will be relied on.
Attach the objective list: six to twelve observable actions inside that scope. Anything that cannot be watched and scored pass or not-yet does not belong on the list.
2. Select the host and protect their week
Pick on teaching behavior, not seniority. The host must be someone who will let the shadowing tech struggle for a minute before taking the tool. A host who does the work to keep the day moving produces an entertained observer and no evidence.
Set the host's expected output for the week at roughly 60 to 70 percent of normal and say so out loud to the whole crew. An unstated expectation that the host will teach and also produce at full rate guarantees the teaching loses. See related: How to Train the Trainer in a Shop of Five.
3. Brief both people before day 1
Fifteen minutes, both present, covering: the scope sentence, the objective list, the three modes and what evidence moves between them, and the fact that ending in assist mode is a legitimate result and not a failure of either person. Naming that in advance removes the pressure that causes hosts to sign off out of politeness.
4. Run mode one, observation
The shadowing tech watches and narrates. The rule is that they call out what they expect to happen next, before it happens. Silent watching produces almost nothing, because the person cannot tell what they missed. A prediction that is wrong is the most useful event of the day, for both of them.
Evidence to leave observation: the shadowing tech can predict the next step correctly on the routine sequence, and can name the safety-critical steps unprompted.
Where the scope involves a real hazard, the safety action is taught first, in isolation, before any of the work sequence: de-energize and verify dead with a meter proven on a known live source, shut off and relieve pressure and verify at zero, set and check the fall anchor before the first step up. That teaching does not wait for the natural point in the job.
5. Run mode two, assist
The shadowing tech does defined portions with the host present and hands-off. The host's job is to intervene only for safety, for customer damage, or when a genuine dead end has lasted long enough to be unproductive. Everything else, let it run.
Assign the portions deliberately rather than handing over whatever comes up. Start with the parts that have the most reps available and the least consequence, and add the higher-consequence portions once the low ones are clean.
Evidence to leave assist: the shadowing tech completed the full sequence on their assigned portions across at least two separate jobs with no safety intervention and no more than minor procedural correction.
6. Run mode three, lead
The shadowing tech runs the whole job including the customer conversation. The host is present, silent, and taking notes rather than watching for a chance to help. The host writes down every moment they wanted to intervene and did not, which becomes the debrief content and is far more useful than a general impression.
Evidence to close the week: one complete job at standard with zero interventions, plus the customer interaction handled without the host stepping in.
7. Make the competency decision in writing
One of three outcomes, recorded the same day with the host's and the lead's names on it:
- Cleared for the written scope. Effective immediately, recorded in the skills matrix with the date.
- Cleared paired only. Can be dispatched on the scope with a second person, with a named condition for full clearance and a re-check date.
- Not yet. Names the specific objectives not met and what the next block will be. This is not a disciplinary outcome and should never be recorded as one.
8. Handle a week that is not working
The host raises a stop by end of day 3 if the week is not viable, for any of three causes:
- The calendar did not deliver the work. Not a capability finding. Reschedule and give dispatch a tighter sourcing rule.
- A prerequisite is missing. The shadowing tech is missing an underlying skill the scope assumed. Stop, close that gap, and rerun the week.
- The pairing is wrong. Some pairs do not work, and it is nobody's failure. Swap the host and rerun. Do not draw a conclusion about the shadowing tech from a bad pairing.
Worked example: adding a service line in a six-person shop
The shop is adding an equipment class it has not carried. One tech attended outside training and has run four of these jobs; three others have run none. The lead schedules three shadowing weeks over six weeks, using the trained tech as host.
Week 1. Scope written as: cleared to run standard service and one common repair on this equipment class, solo, on residential sites. Objective list has nine items. Dispatch sources six matching jobs across the week.
Days 1 and 2 in observation across three jobs. The shadowing tech's predictions are correct on about 6 of 9 objectives by end of day 2, which clears observation on the routine sequence but not on two diagnostic steps, so those two stay in observation into day 3 while the rest move to assist.
Days 3 and 4 in assist across two jobs. One safety intervention on day 3, on the isolation step, which resets that objective to observation for one more rep. Clean on the second job.
Day 5 in lead on one job. Two interventions the host recorded but did not act on, both on sequence efficiency rather than correctness. Outcome: cleared for the written scope, with a note that speed will come with reps.
Weeks 3 and 5. The other two techs run the same block with the same host. Tech two clears. Tech three ends at cleared-paired-only, with two diagnostic objectives unmet, and gets a half-week block four weeks later that closes them.
The arithmetic. Three weeks of host time at roughly 65 percent of normal output is about 1.05 weeks of production given up, spread over six calendar weeks. Against that, the shop went from one name schedulable on the new line to three cleared solo and one cleared paired, inside six weeks. Before the blocks, every job on that line had to route through one person's calendar, which capped the line at whatever that one tech could absorb. After, dispatch has three names, which is roughly a threefold increase in schedulable capacity on the line for about one week of lost production.
Note what the honest accounting includes: tech three did not clear on the first pass. Counting them as cleared would have shown four names instead of three and would have produced a callback on the first unpaired diagnostic job. The value of the SOP is entirely in that distinction.
Records
- Scope sentence and objective list, one page, retained with the employee record.
- Daily evidence log kept by the host: jobs, objectives touched, interventions with cause.
- Signed competency decision with date, host name, lead name, and second signer where required.
- Skills matrix updated the same day. A clearance that lives only in someone's memory is not a clearance.
- Re-check date entered on the calendar for any paired-only outcome.
Common failure modes
- The host takes the tool. Most common by a wide margin, usually under schedule pressure. Catch it by reading the evidence log: a week with no logged struggles did not happen as designed.
- Mode transitions driven by the calendar. Everyone advances to lead on Friday because it is Friday. This produces a sign-off that means nothing, and the first solo job proves it.
- Sign-off out of politeness. The host does not want to fail a colleague. Mitigate with the step-3 briefing, the second signer on high-consequence scopes, and by treating not-yet outcomes as normal.
- Dispatch splitting the pair. One busy Wednesday and the week is gone. This is why the pairing protection sits with dispatch as a named responsibility rather than as a request.
- Scope drift. The week ends up covering three job types instead of one because interesting work came in. The person is then cleared for nothing in particular. Hold the written scope.
References
- U.S. Department of Labor, registered apprenticeship on-the-job learning and supervision structure
- OSHA, competent-person and training requirements for hazardous work
- See related: The Mentor Pairing SOP, The Ride-Along Debrief SOP, The Cross-Training Rotation SOP
- See related: Build a Skills Matrix: Who Can Do What