The Tool Request and Approval SOP

Purpose

To define how a request for a tool the shop does not currently own is made, checked, decided, recorded, and closed out. The goal is not to slow purchases down. It is to make sure every request gets an answer inside a known window, that the answer is recorded with its reason, and that a tool with a training or licensing requirement never reaches a job before the person using it is qualified.

Shops without this procedure fail in two directions at once: techs stop asking because nothing ever comes back, so they buy quietly and the shop loses control of what it owns; and the requests that do get made are decided on whoever asked loudest that week.

Scope

Applies to any request for a tool, instrument, or piece of equipment the shop does not already own in an available condition. Includes requests to rent, and requests to subcontract a task because the shop lacks the tool.

Does not apply to: consumables and stock parts (ordered against normal stock procedure), replacement of a like-for-like tool that has failed and is inside the standard kit (handled under the replacement cycle), or the reissue of an existing tool from the crib (handled under the issue and return procedure).

Roles and responsibilities

Procedure

1. The requester submits a complete request.

A request is not a comment in a truck. It is a record with six fields, and an incomplete request is returned same day rather than sitting:

• What the tool is, in canonical naming (type, size or capacity, power source), not a brand or model.
• What task it does that is not currently possible or is currently slow.
• How often that task comes up, expressed as a count over a stated window ("about 3 times a month over the last 6 months"), not "all the time."
• What the shop does today instead, and how long that takes.
• When it is needed by, and what happens if the date is missed.
• Whether a rental would serve, and whether the requester knows of a rental source.

The frequency field is the one that carries the decision. A request that cannot supply a count over a window is a request based on one bad afternoon, and it should go back for the count rather than be declined.

2. The coordinator runs the duplicate and availability check within 1 working day.

Search the tool list on canonical name and on the requester's own words. Confirm three things: the shop does not own it, or owns it and it is unavailable; if unavailable, why (in use, broken, lost); and whether a comparable item already in the crib would do the task.

If the shop owns it and it is available, this is not a purchase, it is a location problem. Close the request by telling the requester where it is, and log the near-miss - a run of these is the signal that the tool list is not findable.

3. Route by tier.

The tier is set by consequence and commitment, not by price alone. A modest-cost tool that commits the shop to a new battery platform or a recurring calibration cycle is tier 3, because the purchase is the smallest part of what you are agreeing to.

4. Test the request against the earn-back standard.

The decider compares the frequency and time-saved fields against the shop's standing hurdle for a tool purchase. The shop should have one hurdle expressed in hours, not a per-request debate: for example, a tool must return its cost, expressed as the equivalent number of shop labor hours, inside 12 months of expected use. Working the arithmetic here is the whole point of the frequency field. See related: How to Decide What a Specialty Tool Has to Earn.

5. Decide one of four outcomes, and only these four.

• Approve purchase. The frequency clears the hurdle and the tool is a fit for the crew.
• Approve rental. The task is real but infrequent, or the tool class changes fast enough that owning it is a bet. Rental is also the correct first answer for anything the shop has never used, because it doubles as an evaluation.
• Subcontract the task. The work is real, the tool is not the shop's business to own, and a sub does it faster.
• Decline, with a reason. A decline is a legitimate outcome and it must carry a written reason, because an unexplained decline reads as "the owner does not care" and stops the next request from ever arriving.

6. Record the decision the same day it is made.

The record carries: request date, decision date, decision, reason, and who decided. Declines are recorded with the same care as approvals. This record is what makes the quarterly review in step 9 possible, and it is what stops the same tool being requested and declined four times in two years by four different people who never saw the earlier answer.

7. For any hazard-class or certification-gated tool, block issue until training is complete.

This step has no discretion in it. A tool that arrives on a Friday does not go out on a Monday because the schedule is tight.

Specific triggers to check before issue:

• A powder-actuated tool may be operated only by an employee who has been trained in the operation of that particular tool in use, and the operator must carry the card or certificate showing that qualification (29 CFR 1926.302(e)(1) and (e)(2)).
• A powered industrial truck may not be operated by an employee until the employer has trained that operator and evaluated their performance, and the employer must certify the training and evaluation with the operator's name, training date, evaluation date, and the evaluator's identity, with re-evaluation at least once every three years (29 CFR 1910.178(l)(1), (l)(4)(iii), (l)(6)).
• Where the new tool changes how energy is isolated on equipment the shop services, the energy control procedure has to be updated before the tool is used on that equipment (29 CFR 1910.147(c)(4)), and that procedure's periodic inspection is required at least annually (1910.147(c)(6)). Two different subsections and two different duties: (c)(4) is the procedure itself, (c)(6) is the annual check that it is still right.

If training cannot be scheduled before the need date, the correct answer is rental with a trained operator, or subcontracting, not a fast verbal handover in the yard.

8. Close the loop with the requester within the tier's decision window.

Every requester hears an answer, including a decline, inside the window. This is the single mechanism that keeps the procedure alive. A shop that lets three requests go unanswered has trained its crew to buy quietly, and every control in steps 2 through 4 becomes decorative.

9. Review the request log quarterly.

Read the whole quarter's requests in one sitting and look for four patterns:

• Repeats. The same tool requested by different people is a real need being under-weighted.
• Declines that came true. A declined tool followed by subcontract invoices for the same task is a decision that should be revisited with the new evidence.
• Approvals that went idle. A purchased tool with no usage in its first quarter needs a look at training, not a shrug.
• Near-misses from step 2. Requests for tools the shop already owned and available. More than a couple in a quarter means the findability problem is back.

Worked example: a tier 3 request in a nine-tech shop

A lead submits a request for a powered cutting machine. The frequency field says the task comes up about 3 times a month over the last 6 months, so roughly 18 occurrences in the window. Current method takes about 2.6 hours per occurrence; the lead's estimate for the machine is about 1.5 hours.

The check. The coordinator confirms the shop does not own it and that no crib item covers the task. Routed tier 3, because the machine needs a dedicated accessory line and a competency check-out before use.

The arithmetic. Estimated saving is 1.1 hours per occurrence. At 3 occurrences a month that is 3.3 hours a month, and 39.6 hours over 12 months. The shop's standing hurdle is that a tool must return its cost, expressed as equivalent shop labor hours, inside 12 months of expected use, and this machine's cost converts to about 22 hours. At 3.3 hours a month, the hurdle is reached in about 6.7 months, comfortably inside 12.

The discount the decider applied. The 1.5-hour estimate is the lead's, from having used one at a previous employer, and it does not include the learning curve for eight other techs. The decider haircuts the saving to 0.8 hour per occurrence for the first six months. That gives 2.4 hours a month for months 1 through 6 (14.4 hours) and 3.3 hours a month after, so the 22-hour hurdle clears during month 9. Still inside 12 months, so the decision holds even on the pessimistic case. That is the test worth running: does the approval survive the conservative version of its own numbers.

The training gate. Purchase approved on a Tuesday, machine arrives the following week. Two jobs of the type are already on the board. The tool does not go out on either, because no check-out has happened. The trainer runs a 45-minute check-out for the lead plus three techs on the Thursday, and the machine goes out on the next job of the type. Those two jobs ran the old way at 2.6 hours each. That is the cost of the gate, and it is the correct cost to pay.

Recorded outcome. Approved, reason recorded, hurdle arithmetic attached to the record. At the quarterly review, usage showed 8 occurrences with the machine used on 7, which is the pattern an approval should produce. The one exception was a job where the access would not take the machine, which is useful information for the next request of this kind.

References

• OSHA 29 CFR 1926.302(e), powder-actuated tools: operator training and qualification
• OSHA 29 CFR 1910.178(l), powered industrial truck operator training, evaluation, and certification
• OSHA 29 CFR 1910.147(c)(4), energy control procedures, and (c)(6), their periodic inspection
• See related: How to Decide What a Specialty Tool Has to Earn, How to Decide Whether to Buy, Rent, or Subcontract a Tool, The Tool Issue and Return SOP, How to Stop Buying Duplicates You Already Own