By Sajad, Co-founder at cellbot — 25 years in the tech repair industry

Published: 23 September 2025 · Editorially reviewed: 26 August 2026

Evidence basis: original repair-shop procedure-control method; current HSE, ICO and Consumer Rights Act sources. No onboarding-speed, productivity, loss, diagnostic-fee or software-enforcement claim is presented as an industry fact.

A repair-shop standard operating procedure is a versioned instruction for one bounded task, performed by an authorised role, with required evidence and a clear stop rule. It is not a policy, a training promise or a list of ideal outcomes.

A useful SOP lets a competent person perform the same safe process, record what happened and escalate when the real device falls outside the documented path.

!Repair shop SOP control cycle from a bounded task and hazards through versioned steps, demonstrated competence and audit to correction or retirement

Separate policy, SOP, checklist and record

DocumentJobExample
Policystates the business rule and authoritycustomer passcodes are requested only when an approved test requires them
SOPexplains how an authorised role performs one taskreceive and document a customer device
Checklistprevents omission during executionsix-sided condition photographs complete
Recordproves what happened for this jobintake result, operator, timestamp and exceptions

Do not hide a commercial rule inside a technical SOP. Do not treat a completed checklist as proof the operator understood or safely performed the work.

Download the repair-shop SOP control register. It captures owner, role, scope, prerequisites, hazards, PPE/control, evidence, stop rules, exceptions, training, approval, review and retirement in one version history.

Use one SOP anatomy

Every procedure should contain:

  1. Purpose: the outcome this procedure controls.
  2. Scope: exact devices, tasks, locations and situations included.
  3. Out of scope: conditions that require another procedure or escalation.
  4. Authorised roles: who may perform, supervise and approve the task.
  5. Prerequisites: competence, tools, environment, data and customer authority.
  6. Hazards and controls: what can harm people, devices, data or property.
  7. Inputs: job record, device, part, quote, evidence and procedure version.
  8. Steps: observable actions in execution order.
  9. Quality gates: result required before moving forward.
  10. Evidence: fields, photographs, measurements, signatures and timestamps.
  11. Stop and escalation: unsafe, ambiguous or failed conditions.
  12. Exceptions: permitted deviation, approving role and additional evidence.
  13. Outputs: status, record, customer message and physical location.
  14. Version control: author, approver, effective date and prior version.

If the procedure needs several unrelated outcomes, split it. “Run the whole shop” is an operating model, not an SOP.

Write from the real task

Observe the work before drafting:

  • follow one clean example;
  • follow one known exception;
  • record decisions, handoffs and evidence;
  • identify where staff use memory or off-record messages;
  • compare the actual path with policy, safety and customer terms; and
  • ask a second competent person to reproduce it.

Write verbs and observable results. “Check the phone carefully” is not testable. “Record the display, touch, camera, microphone, speaker, charging, network and biometric result as pass, fail or not testable” is.

Do not copy a manufacturer, supplier or internet procedure without verifying the exact model, tool, environment, permissions and licence. Link the controlled source and preserve the version used.

Start with the failure-sensitive procedures

Device intake and custody

Control identity, authorised contact, pre-existing condition, accessories, testability, data access, scope, physical location and custody handover. The operations playbook owns the states; this SOP owns the exact intake task.

Diagnosis and quote change

Require observed evidence, uncertainty, authorised diagnostic work, changed scope, revised total price and customer decision. Stop chargeable work until the required approval is attached to the job.

Part receipt and fitment

Verify identity, described tier, supplier, batch, physical condition and job reservation. Preserve fitment, removal, quarantine, RMA and credit evidence using the inventory guide.

Model-specific repair

State prerequisites, battery and power conditions, ESD and fume controls, disassembly source, fastener/part control, calibration, reassembly and stop conditions. Do not make a generic screen SOP responsible for every model.

Final test and release

Define required functions by device and scope. Blank, skipped and not-testable must remain visible. Release only when the approved work, part history, test, price and custody route agree.

Return, warranty and complaint

Open a linked case; preserve the original repair; record reported issue, inspection, causal uncertainty, remedy decision and supplier path. Use the warranty guide and disputes guide for their respective ownership.

Put safety controls before speed

The HSE's risk-assessment guidance says employers must identify hazards, decide how likely and serious harm could be, control risk, record significant findings where required and review controls.

An SOP should link to the relevant risk assessment and state the operational stop. Examples include:

  • damaged, swollen, leaking or overheating battery;
  • missing extraction or unsuitable workspace for a fume-producing task;
  • untrained operator or unapproved repair scope;
  • unknown power or stored-energy condition;
  • missing model-specific procedure;
  • tool or measurement equipment outside its approved condition; and
  • customer device or data state inconsistent with the job record.

Never write “use appropriate PPE” without naming the approved control for the assessed task. Obtain competent safety advice for the premises and work.

Treat customer data as a task input with limits

State whether the procedure needs customer-device access, what function is tested, who may see the data, how credentials are handled, and when access ends.

The ICO's data-protection principles include purpose limitation, data minimisation, accuracy, storage limitation and security. An SOP should make the minimum safe path the default, not depend on a technician remembering a privacy policy.

Use a manufacturer test mode or customer-present test where appropriate. Never retain a passcode in free text merely because it is convenient.

Train by demonstration, not acknowledgement

Reading and signing the SOP proves access, not competence. Use four states:

  1. Read and discuss: operator explains purpose, hazards and stop rules.
  2. Observed demonstration: trainer performs the task and points to evidence.
  3. Supervised execution: operator performs representative clean and exception cases.
  4. Authorised scope: approver records the precise task/device scope the operator may perform.

Record attempts, feedback, failed gates and retraining. Re-authorise after a material procedure, tool, model or risk change.

The phone-repair learning guide owns competence stages; the SOP provides the task against which competence is demonstrated.

Control versions without destroying history

Give every SOP a stable identifier and immutable version. Record:

  • version and effective date;
  • author and accountable owner;
  • reviewers and approver;
  • reason for change;
  • affected tools, devices, roles and forms;
  • training or acknowledgement required;
  • superseded version and retirement date; and
  • open jobs that remain on an older approved version.

Do not silently edit a live procedure. A historical repair must be judged against the version in force when the work occurred.

Audit the work, not the existence of the PDF

Sample recent jobs and compare evidence with the SOP:

  • correct version used;
  • authorised operator;
  • prerequisites present;
  • required steps and gates recorded;
  • deviations approved;
  • stop rules followed;
  • final output and job status agree; and
  • customer and data controls complete.

Classify the finding:

FindingMeaningAction
Procedure followed; outcome failedmethod, part or assumption may be wronginvestigate and revise if evidence supports it
Procedure not followedtraining, access, usability or supervision gapstop/limit scope and correct the control
Procedure could not be followedreal work exceeds documented pathopen an exception and update scope safely
Evidence missingoutcome cannot be reproducedtreat as control failure, not an assumed pass
Procedure obsoletesource, device, tool, law or policy changedretire or replace before further use

Use the KPI method to measure repeated failures with stable definitions. Do not reward checklist completion when rework or missing evidence increases.

How Cellbot fits

Cellbot can support job checklists, roles, records, statuses and audit history. It does not write a safe technical procedure, assess the premises or authorise technician competence. Confirm current product scope on the features page.

Test that a required failure blocks the next state and creates an owned exception. A checklist that can be skipped without evidence is a suggestion, not an enforced gate.

Frequently asked questions

How long should a repair-shop SOP be?

As short as the bounded task permits while preserving hazards, steps, evidence, stop rules and exceptions. Split unrelated tasks rather than compressing them into vague language.

Should every repair model have its own SOP?

Use shared controls for intake, custody, authorisation and records, but link model-specific technical procedures where disassembly, battery, calibration, adhesives, fasteners or tests differ.

How often should SOPs be reviewed?

Use a fixed review date and event triggers: incident, complaint, rework pattern, source update, new device/tool, policy or law change. An unchanged date does not prove the procedure remains safe.

Who should approve an SOP?

The accountable operational owner plus competent reviewers for the task's technical, safety, data, legal or financial risks. One approver may not cover every domain.

Can AI write repair SOPs?

It can organise verified notes, but it cannot observe the shop, validate the exact device, assess risk or prove competence. Every step and source needs human verification before use.

Bottom line

Write one bounded task, name its hazards and prerequisites, make each step observable, preserve evidence, define when to stop and train through supervised execution. Then audit the work against the version used and correct the procedure when reality proves it incomplete.

Start with the SOP control register and align it with the job-control playbook.

What changed on 26 August 2026

The former loss anecdotes, onboarding percentage, diagnostic-fee result and automatic-enforcement claims were removed. The replacement separates policy/SOP/checklist/record, adds a 14-part anatomy, safety and data gates, competence states, immutable versions, audit findings, stop rules and a downloadable control register.

Sources and method

The SOP anatomy, competence states, audit table and register are original Cellbot editorial operating methods, not industry benchmarks. Obtain competent technical, UK legal, safety and data-protection review before operational use or publication.