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

Published: 10 September 2025 · Editorially reviewed: 26 August 2026

Evidence basis: original UK repair-shop operating method; current consumer-service, data-protection, health-and-safety and workplace guidance. No revenue, margin, throughput, deposit, staffing or productivity target is presented as an industry benchmark.

A repair shop is under control when every device has one current owner, one truthful status, one authorised scope, one physical location and one next action. The job record must explain how the device entered custody, what was approved, what changed, which parts moved, what tests passed and how the device left.

Busy is not the same as controlled. More tickets, notifications or dashboards amplify bad definitions if staff can still work outside the record.

!Repair shop job-control chain from custody and authorisation through diagnosis, parts and repair to final test, handover and linked exception

Make one job record authoritative

The minimum record contains:

  • repair reference and exact customer/device identity;
  • custody start, physical location and authorised contacts;
  • customer-reported symptom in their words;
  • pre-existing condition and intake evidence;
  • diagnostic scope and charge already agreed;
  • quote version, part description, total price and approval;
  • current status, reason, owner and next action;
  • part reservations, fitment, returns and credits;
  • work performed, technician and procedure version;
  • final-test results and limitations;
  • messages, delivery states and customer decisions;
  • payment, collection/delivery and custody end; and
  • linked warranty, RMA, complaint or refund cases.

Download the repair-shop job-control register. It follows one repair from intake to closed custody and keeps changed scope, parts, final test and exceptions linked without overwriting history.

Use states that describe evidence

A status should tell staff what is true and what happens next. A compact model is:

StateEntry evidenceExit evidence
Bookedcustomer, device request and intake route recordeddevice received or booking closed with reason
Receivedcustody, identity, condition and contact authority recordeddiagnostic or approved repair scope assigned
Diagnosingdiagnostic authority and owner presentfinding, quote or no-fault result recorded
Awaiting customercomplete decision request sentapproval, decline or escalation recorded
Awaiting partexact part and order/reservation linkedreceived/verified part allocated or plan changed
In repairapproved scope, procedure and technician setwork note and final-test entry created
Final testrequired test version and repaired scope recordedpass, not-testable or rework decision recorded
Readyprice, work, test and collection/delivery route completepayment/custody handover or overdue exception
Closedcustody ended and financial outcome recordedreopened only through a linked case

Do not create “urgent”, “problem” or “waiting” as unexplained statuses. Urgency is a priority; a problem needs a cause and owner; waiting needs a dependency and next review.

Gate custody and authorisation at intake

Do not begin technical work until the shop can answer:

  • Is this the exact device recorded?
  • Who is authorised to approve work and collect it?
  • What condition and accessories entered custody?
  • What can be tested before opening?
  • What data access is necessary, if any?
  • What diagnosis or repair scope is currently authorised?
  • What price, timing and material limitations were communicated?
  • What event stops the job?

The customer communications playbook owns the messages. The work-order guide owns detailed approval evidence. Operations must refuse progress when either is incomplete.

Separate diagnosis from promised repair

A symptom is not a diagnosis. Record observed evidence, test result, uncertainty and alternative causes. If work is needed to complete diagnosis, quote that scope rather than promising a component replacement from the customer's description.

Consumer Rights Act 2015, section 49 requires a trader to perform a consumer service with reasonable care and skill. That does not create a universal technical procedure; it makes controlled scope, competence, evidence and escalation commercially important.

When the finding changes scope or price:

  1. stop the affected work;
  2. preserve the original quote and approval;
  3. record the new evidence;
  4. issue the changed scope, total price and timing;
  5. wait for the required customer decision; and
  6. record approval, decline or return path.

Silence is not approval for additional chargeable work.

Control parts by job and outcome

Reserve the exact part to the exact job. “In stock” is insufficient if the unit is allocated, quarantined or already fitted elsewhere.

At receipt, preserve supplier, part number, described tier, cost, batch where relevant and inspection result. At use, record fitment and the technician. At failure, open the RMA workflow without deleting the original movement. The inventory-control guide owns reorder, count, obsolescence and waste rules.

Do not release a supplier credit into margin reporting until the accounting treatment and actual credit state are reconciled.

Make final test a release gate

Define the test by device and repair scope. Record each required result as pass, fail, skipped with reason or not testable. A blank result is not a pass.

Release only when:

  • approved work and parts are recorded;
  • every required final-test field has an allowed result;
  • safety and customer-data steps are complete;
  • remaining limitations are stated accurately;
  • price and payment state agree; and
  • collection or delivery authority is confirmed.

If final test fails, return the job to an owned rework state. Do not create a new successful repair that disconnects the failure from the original cohort.

Use the warranty policy guide for commercial coverage and the KPI definitions for first-pass and reopened-repair measures.

Run the shop from exception queues

The normal path should progress with minimal management. Owners need the exceptions:

  • received with missing intake evidence;
  • awaiting customer beyond the promised review;
  • part overdue or received but quarantined;
  • current status with no owner or next action;
  • promised-ready time at risk;
  • final-test failure or not-testable result;
  • ready but delivery/contact/payment blocked;
  • reopened repair, complaint, refund or supplier RMA; and
  • record correction requiring review.

Each exception needs reason, owner, opened time, next action and review time. A red dashboard without an owner is decoration.

Use a daily control routine

Opening

  • reconcile every device physically present with the job/location record;
  • review overnight replies, failed messages and bookings;
  • assign unowned exceptions;
  • confirm parts expected and promised-ready risks; and
  • stop unsafe or unauthorised work.

Midday

  • review awaiting-customer and awaiting-part cases;
  • compare technician work with approved scope;
  • move completed work through final test rather than straight to ready; and
  • contact customers only from verified events.

Closing

  • reconcile devices, keys/accessories, payments and custody handovers;
  • record every unfinished job's owner and next action;
  • resolve or escalate stock movements outside jobs;
  • inspect failed tests, reopenings and complaints; and
  • record the next opening's risks.

The exact cadence depends on volume and risk. The invariant is that no device becomes an overnight unknown.

Build SOPs beneath the state model

The operating model states what must be true. An SOP explains how a competent person performs a bounded task. Create SOPs for intake, diagnosis, quote change, parts receipt, chosen repairs, final test, handover, returns, battery/waste handling and data incidents.

Use the repair-shop SOP guide for versioning, competence and audit. Do not write a 200-page manual before the common failure paths are controlled.

Protect people and customer data

The HSE's risk-assessment guidance requires employers to identify hazards, assess risks and control them. The operation must stop when safety-critical equipment, competence or conditions are missing.

Apply the ICO's data-protection principles to intake data, passcodes, photographs, device contents, messages and retention. Give staff only the access needed for the task, and record when exceptional access is authorised.

The GDPR guide owns the wider data system. Operations owns the point at which a job must stop because that system cannot be followed.

Reconcile operating and financial truth

Closed job count, collected repair revenue, parts movement, refunds, supplier credits and rework cost must agree with approved definitions. Do not mix quotes with sales, ready jobs with paid jobs or list price with collected value.

Use the accounting guide for formal records and the earnings model for contribution and overhead. This playbook owns job events, not accounting policy.

How Cellbot fits

Cellbot can connect customer, device, job, quote, parts, payment and message records and support controlled statuses. The shop still defines its state model, approvals, procedures, permissions, exception ownership and release gate. Confirm current plan and feature scope on the features page.

Test one complete repair plus a changed quote, failed message, failed final test and RMA before treating any automation as operationally safe.

Frequently asked questions

What is the most important repair-shop process?

Custody and authorisation. If the shop cannot prove which device it holds, what condition arrived and what work the customer approved, every downstream process is fragile.

How many repair statuses should a shop use?

Use the smallest set that creates distinct ownership or exit evidence. Add a state only when staff take a different action; keep reasons and priorities as separate fields.

Should technicians update customers directly?

Only through the controlled communication process. A technician can supply the verified event and explanation, but material price, timing and approval messages must remain attached to the job.

How should urgent jobs be handled?

Record a priority with a reason and approved commitment. Do not skip intake, authorisation, safety or final test. If the promised priority cannot be met, open an exception and update the customer.

What should be reviewed first each morning?

Devices or jobs whose physical location, owner, authorisation, promise or next action is unknown. Resolve control gaps before optimising throughput.

Bottom line

Run the repair shop as a chain of evidence: custody, approved scope, diagnosis, part movement, work, final test and handover. Make normal work easy, make exceptions visible and never let a plausible status replace a recorded fact.

Implement the state model with the job-control register, then write the SOPs beneath it.

What changed on 26 August 2026

The former market-size, deposit, stock-loss, pay, bonus, daily-revenue and claim-value targets were removed. The replacement adds one authoritative job record, evidence-based states, custody and approval gates, parts linkage, final-test release, exception queues, a daily routine, legal/safety/data boundaries and a downloadable control register.

Sources and method

The state model, exception queues, daily routine and register are original Cellbot editorial operating methods, not industry benchmarks. Obtain qualified UK legal, safety, accounting and data-protection review before publication.