A warehouse SOP for a new system should be rewritten from the system as configured, not edited from the old binder. Write one procedure per process in a fixed structure: purpose and scope, roles, safety, equipment, numbered steps, exceptions, checks and records. Write each system step with the exact screen name, scan prompt and confirmation message the user will see. Draft the SOPs during user acceptance testing from the test scripts, and test them with people new to the task. Then tie every revision to a software release, so the system cannot change without the procedures that describe it.
That approach applies whether the new system is a warehouse management system (WMS), a tracking app on the dock, or an ERP release that changes receiving or counting. This note covers the written procedures. How people are trained on them, and how super users support each shift, is in training warehouse teams on a new system with super users.
Why existing warehouse SOPs fail after a system change
Old SOPs describe the old system’s sequence, prompts and controls. After go-live, the triggers, the order of scans, the exception paths and the approval points all change. A procedure that was accurate last month now sends people down steps that no longer exist.
The US Environmental Protection Agency’s guidance on SOPs puts the stakes plainly: “If not written correctly, SOPs are of limited value”, and “the best written SOPs will fail if they are not followed” (EPA QA/G-6, April 2007). An SOP that contradicts the handheld will not be followed. People will follow the screen, or a coworker’s habit, and the document becomes shelf decoration.
What actually changes
| Element | Old system or paper process | New system |
|---|---|---|
| Trigger | Supervisor hands out a paper receiving sheet | A task appears in the receiver’s queue when the trailer is checked in |
| Sequence | Count, then key quantities at a desk | Scan pallet, confirm quantity, scan location, in that order |
| Identifiers | Item number written on the sheet | SSCC on the pallet label, lot from the case label |
| Exceptions | “Tell the supervisor” | A defined message, a reason code and a queue |
| Controls | Supervisor signs the receiving sheet | Approval rights assigned to user roles; tolerance rules in configuration |
| Records | Paper sheet filed by date | Transaction history in the system; paper only for exceptions |
Every row in that table is a reason to rewrite rather than edit. A procedure built on a paper trigger and a desk-entry sequence cannot be patched into a directed, scan-driven flow.
The binder and the floor
Many sites keep two sets of procedures: the approved binder and the way work is actually done. A system change is the moment to close that gap, because both versions are about to be wrong. Start by collecting the existing SOPs anyway. They show what people were told, which explains a lot about what they do.
What a warehouse SOP should contain
A warehouse SOP needs a control block, a purpose and scope, roles, safety warnings, equipment, numbered steps with exception paths, checks, and records. EPA’s QA/G-6 guidance gives a tested structure for this, written for laboratories and field work but general enough to adapt.
EPA defines an SOP as “a set of written instructions that document a routine or repetitive activity followed by an organization” (EPA QA/G-6). The April 2007 edition (EPA/600/B-07/001) replaced the March 2001 guidance. It lists the topics a technical SOP may include and notes that not every topic applies to every procedure.
| EPA QA/G-6 element | Warehouse SOP section | What it holds |
|---|---|---|
| Title page | Control block | SOP ID, title, revision, effective date, system release, owner and approver |
| Scope and applicability | Purpose and scope | The process, the sites and item types it covers, and what is out of scope |
| Summary of method | Summary | Three to five sentences describing the flow |
| Definitions | Terms | The floor term next to the system term, such as “pallet tag (LPN)” |
| Health and safety warnings | Safety | Dock plates, trailer restraints, forklift traffic, at the steps where they apply |
| Cautions | Cautions | Actions that damage the record, such as closing a receipt before every pallet is scanned |
| Interferences | Known conditions | Damaged labels, mixed pallets, Wi-Fi dead spots |
| Personnel qualifications and responsibilities | Roles | Who performs each step, who approves, what training is required |
| Equipment and supplies | Equipment | Handheld model, printer, label stock, scale |
| Procedure, including troubleshooting | Steps and exceptions | Numbered steps with prompts, and an exception table |
| Data and records management | Records | What the system records, what paper is kept and where |
| Quality control and quality assurance | Checks | Supervisor checks, counts, tolerances and what happens when they fail |
| References | Related documents | Linked SOPs, reference cards and forms, with their versions |
Two details from the EPA guidance matter on a warehouse floor. Health and safety warnings belong in their own section and also “at the critical steps in the procedure”. And the guidance says the level of detail may differ with how critical the process is, how often it is done and how many people use the SOP.
One SOP per process, not per screen
Organize SOPs around the work, not the menu. “Receive a pallet against an ASN” is a process with a trigger, an outcome and an owner. “The receiving screen” is not. A process SOP can span several screens and devices, and a single screen can appear in several SOPs. Readers look for the job in front of them, so title each SOP with the job.
SOP, reference card and checklist
Warehouses need several document types, and mixing them up produces procedures nobody reads.
| Document | Purpose | Typical length | Where it lives |
|---|---|---|---|
| SOP | The full, approved procedure, including roles, exceptions and checks | Several pages | Document system, linked from the station |
| Quick reference card | The scans in order and what each error means | One page | Laminated at the dock door, pack bench or count cart |
| Checklist | Proof that steps were done, filled in each time | One page | With the records of that activity |
EPA’s guidance is clear on the last one: “the checklist is not the SOP, but a part of the SOP.” The same logic applies to reference cards. Each card is derived from one SOP, carries that SOP’s ID and revision, and changes when the SOP changes.
Writing system steps from the real screens and scan prompts
Write each step as one action, using the exact words on the screen in quotation marks, and say what the user sees when the step worked. A step that says “receive the pallet” is a goal. A step that says “Scan the SSCC barcode on the pallet label when the screen shows ‘Scan pallet ID’” is an instruction.
EPA’s writing guidance is the right starting point: “concise, step-by-step, easy-to-read”, in the active voice and present tense, with “you” implied rather than written. On a handheld procedure, that becomes a handful of rules.
- One action per step. If a step contains “and then”, split it.
- Quote the prompt. Copy the prompt text exactly as the device shows it, including capitalization. People match words, not meanings, at speed.
- Name the barcode. Pallets and cases carry several barcodes. Say which one: the SSCC on the pallet label, the GTIN on the case, the location label on the rack beam.
- State the confirmation. Every step ends with what the screen shows next, so the user knows the step worked.
- Use the floor’s words, then the system’s. If the floor says “pallet tag” and the system says “LPN”, write “pallet tag (LPN)” the first time.
- Show the screen. A screenshot from your own configuration beats a paragraph. Vendor demo screenshots show fields you do not use.
- Put warnings at the step. A forklift or dock-plate warning in a safety section on page one is not enough on its own.
| Step | Action | Screen prompt | You should see | If not |
|---|---|---|---|---|
| 1 | Select Receive, then By ASN | "Scan ASN or PO" | Supplier name and line count | "ASN not found": use RCV-03, receiving without an ASN |
| 2 | Scan the SSCC barcode on the pallet label | "Scan pallet ID" | Item EX-20118, quantity and lot from the ASN | "Pallet not on ASN": set the pallet aside, tag it, call the supervisor |
| 3 | Count the cases and enter the count | "Confirm qty" | "Qty matches ASN" | Count differs: enter the counted quantity and select SHORT or OVER |
| 4 | Scan the location label on the staging lane beam | "Scan staging location" | Pallet status "Received, staged" | "Location invalid": scan the beam label, not the door sign |
| 5 | Close the receipt after the last pallet | "Close receipt? Y/N" | A receipt number | Open pallets listed: scan them, or mark them missing with the supervisor |
Example (fictional). The old receiving SOP at a fictional distributor says: “Receive the pallet into the system and put it in staging.” The rewrite for the new WMS becomes the five steps in the table above. Testing shows that receivers scan the door number sign instead of the staging lane label in step 4, because both are yellow. The SOP gains a photo of the beam label, and the site reprints the door signs without a barcode.
Screenshots and photos that stay accurate
Take screenshots from the test environment with fictional data, never from production records with customer names or prices. Crop each one to the part of the screen the step refers to, and mark the field or button. Note the system release in the file name, so that when a release changes the screen, the owner can find every image that needs replacing.
Photos of the physical world age more slowly but matter as much: where the SSCC sits on a typical inbound pallet, what a valid location label looks like, which printer feeds which door. One good photo often removes a paragraph of text.
Write for the configuration, not the product
Two sites on the same WMS can have different prompts, different required fields and different tolerances. Write from your own test environment, with your own settings switched on. If a step depends on a setting, such as whether an item requires lot capture, say so in the step. For example: “If the screen shows ‘Scan lot’, scan the lot barcode on any case label.”
Writing the exception paths
Most of a warehouse SOP’s value is in its exceptions. For each step, list the messages the system can show, what each one means, what the user does, and who decides when the user cannot. A procedure that covers only the clean receipt teaches people to improvise the moment a label is missing.
Start from the system’s own message list for each transaction. Then add the physical exceptions the system cannot see: a damaged pallet, a wet carton, a trailer seal that does not match. Then write the exception table once, and point to it from the steps.
| Message or condition | What it means | User action | Decided by |
|---|---|---|---|
| “Pallet not on ASN” | The SSCC is not on this ship notice | Set aside, tag, call supervisor | Supervisor, with purchasing |
| “Qty exceeds tolerance” | Count is outside the configured tolerance | Recount once, then enter the count | Supervisor approves over-receipt |
| “Lot required” | Item is lot-controlled and no lot was captured | Scan the lot on a case label, or key it from the label | Receiver |
| “Location invalid” | Scanned label is not a valid location | Scan the beam label for the lane | Receiver |
| Damaged pallet | Physical damage the system cannot see | Photograph, receive with a damage reason code, move to hold | Inventory control |
| Device offline | No connection at the door | Follow the downtime procedure; do not close the receipt on paper | Supervisor |
Reason codes and authority
When an exception ends in an inventory adjustment or a quantity override, the SOP names the reason code to use and the role that approves it. Do not reproduce the full reason-code list in every SOP. Link to the controlled list, and keep the approval thresholds there. Our note on inventory adjustment reason codes and approval rules covers how to build that list.
Exceptions that cross departments
Many exceptions start on the floor and finish in an office. A short shipment notifies purchasing; a damaged pallet may become a freight claim; a lot on hold goes to quality. The receiving SOP should end at the handoff: which queue or report the exception lands in, what information the receiver must capture for the next person, and who owns it from there.
Do not write purchasing’s procedure inside the receiving SOP. Link to it by ID. When each department owns its own steps, a change in one office does not leave a stale copy in three other procedures.
The stop rule
Every SOP should say when to stop. “If the screen quantity and the shelf disagree, do not adjust. Log the item and location and call inventory control.” A clear stop rule prevents the well-meant fix that turns a small variance into a negative balance.
The downtime procedure
Write the procedure for working without the system before go-live, not during the first outage. It covers what is recorded on paper, which tasks stop entirely, and who enters the paper records when the system returns. Print it and keep it at each supervisor’s desk.
Drafting SOPs during UAT from the test scripts
Draft SOPs during user acceptance testing (UAT), using the test scripts as the skeleton. By UAT, the screens are stable enough to describe, the scripts already list the steps and expected results, and the testers are the people who will use the procedures.
Waiting until after go-live is the usual alternative, and it builds in a gap. People work the first weeks from memory and training notes, habits form, and the SOP ends up describing whatever the habits became. Writing earlier, from specifications, has the opposite problem: the screens change during the build.
Mapping a test script to an SOP
A well-written UAT script already holds most of an SOP’s content. Our UAT test plan template uses the same columns as most test plans, and each one maps to part of the procedure.
| UAT script column | SOP section it feeds |
|---|---|
| Scenario | The trigger at the start of the procedure, such as “A trailer arrives with an ASN” |
| Steps | The numbered steps, with prompts added from the screen |
| Expected result | The confirmation at the end of each step, and the check at the end of the procedure |
| Test data | A worked example in the SOP or on the reference card |
| Tester | The role that performs the procedure |
| A failed or blocked result | An exception path the SOP must describe |
The last row is easy to overlook. Every defect logged in UAT describes something that can happen on the floor. Even after the defect is fixed, the behavior that caused it, such as a label without an SSCC, still needs an exception path.
Screens that are still changing
Some screens will still change after UAT starts, usually because a defect fix moves a field or adds a prompt. Draft those steps anyway, flag them in the draft, and list them on the release plan. The owner rechecks every flagged step in the final test build before approval. A known-provisional step is far easier to finish than a blank page written in the week before go-live.
Who drafts, and when
Assign one SOP owner per process before UAT starts, usually the area lead or a documentation specialist working with them. The owner sits in on that area’s UAT sessions, drafts from passing scripts, and marks any step whose screen is still expected to change.
EPA’s guidance says SOPs should be written by “individuals knowledgeable with the activity”, essentially “subject-matter experts who actually perform the work”. A technical writer can make a procedure clear. Only the people on the floor can make it true. In our engagements, documentation is drafted alongside the build and finished against the system as it went live, which is the same principle.
Keep traceability in both directions
Record which requirement and which UAT script each SOP step came from. When a requirement changes, you can find the affected procedures in minutes. That traceability starts with requirements written as testable statements, covered in writing logistics software requirements your team can test.
Testing an SOP with the people who will use it
An SOP is ready when someone new to the task completes it with the document alone, in the training environment, without asking for help. Watch them, note every hesitation, and fix the document rather than coach the person.
EPA’s guidance sets the bar: SOPs “should be written with sufficient detail so that someone with limited experience with or knowledge of the procedure, but with a basic understanding, can successfully reproduce the procedure when unsupervised”. It adds that “it is especially helpful if draft SOPs are actually tested by individuals other than the original writer before the SOPs are finalized.”
A test protocol for one SOP
- Choose a tester from the role who has not done this task on the new system.
- Give them the SOP, the reference card, a handheld and the scenario data. Nothing else.
- Watch without speaking. Note each step where they pause, reread, ask, or do something the SOP did not say.
- Run at least one exception from the exception table.
- Ask afterward which words confused them, and write down their words for things.
- Revise the SOP, then test with a different person.
Two clean runs by two different people is a sensible bar for approval. Record the tester, the date and the revision tested, because that record is evidence of the review.
Example (fictional). A cycle count SOP (INV-04, draft revision 0.3) is tested by a picker moving into inventory control. She stops at step 6, “Confirm the count”, because the screen shows two buttons: “Submit” and “Recount”. The SOP did not say which. The owner checks the configuration: “Recount” sends the task back to the same counter, and “Submit” routes a variance over tolerance to a second counter. Step 6 becomes two lines, one for each button, and the next tester completes the count without stopping.
Approval
EPA’s guidance says the immediate supervisor and the quality assurance officer generally review and approve each SOP, and that electronic signatures are acceptable. In a warehouse, that usually means the operations manager for the area and whoever owns quality or inventory control. Approval comes after the user test, not before it.
Documenting control points in the SOP
Mark the steps that are controls: approvals, counts, reconciliations, holds and segregation of duties. Say who performs each one, when, and what record proves it happened. If a control exists only in the system configuration and not in the procedure, nobody will notice when it stops working.
The US Government Accountability Office’s Green Book, Standards for Internal Control in the Federal Government, is written for federal agencies, but it states that nonfederal entities, including nonprofit organizations, may adopt it as a framework. The 2025 revision (GAO-25-107721) was published May 15, 2025, and is effective beginning with fiscal year 2026. Its language on procedures is directly useful in a warehouse.
Under principle 12, “Management should implement control activities through policies and procedures.” Management documents “in policies what is expected and in procedures specified actions that implement policies” (paragraph 12.02). Procedures “may include the timing of when a control activity occurs and any follow-up corrective actions” (paragraph 12.04). And when there is a significant change in a process, management reviews it “in a timely manner after the change”, noting that changes “may occur in personnel, business processes, or information technology” (paragraph 12.05). A new WMS is that change.
Control points a warehouse SOP should name
The Green Book’s list of common control activities includes physical control over vulnerable assets such as inventories, where management “periodically counts and compares such assets to control records”. It also covers authorization of transactions, including assigning capabilities to users’ credentials in an information technology system, and segregation of duties.
| Control point | Where it appears | Evidence |
|---|---|---|
| Over-receipt approval above tolerance | Receiving SOP, quantity step | Approval record with user ID and time |
| Inventory adjustment approval | Cycle count SOP, variance step | Adjustment with reason code and approver |
| Count before adjustment | Cycle count SOP | Recount task completed by a second counter |
| Quarantine or hold release | Receiving and quality SOPs | Status change by an authorized role |
| Shipment verification | Shipping SOP, load step | Load scan of every pallet against the order |
| Separation of receiving and adjusting | Roles section of each SOP | Role permissions in the system |
Write the role, not the person. “Inventory control lead approves” survives turnover. “Maria approves” does not.
When the rules require written procedures
Some warehouses work under rules that require written procedures outright. FDA’s current good manufacturing practice rules for finished pharmaceuticals require written procedures for warehousing drug products, including quarantine before release and storage conditions (21 CFR 211.142). They also require written distribution procedures, including oldest approved stock first and a way to determine where each lot went (21 CFR 211.150).
For foods on FDA’s Food Traceability List, the traceability plan must describe the procedures used to maintain the required records, “including the format and location of these records”, and how traceability lot codes are assigned (21 CFR 1.1315). The plan must be updated as needed to reflect current practices, and the previous plan must be kept for 2 years after an update. A new WMS changes the format and location of those records. FDA has said it will not enforce the Food Traceability Rule before July 20, 2028, following a congressional directive (FDA), but the plan still has to match the system when enforcement begins.
Version control tied to software releases
Give every SOP an ID, a revision number, an effective date, an owner and the system release it was written against. Then make SOP updates part of the release: a change that alters a screen, a prompt or a rule does not go live until the affected SOPs and reference cards are revised, approved and trained.
EPA’s guidance describes the document control basics. Each organization should use a numbering system, and each page should carry a short title or ID, a revision number, a date and “Page 1 of” the total. The organization should keep a master list of all SOPs, with fields including SOP number, version, date of issue, title, author and status. Outdated versions should be archived “in a manner to prevent their continued use”.
A register that links SOPs to releases
| SOP ID | Title | Revision | Effective | Written for release | Owner | Next review |
|---|---|---|---|---|---|---|
| RCV-02 | Receive a pallet against an ASN | 3 | 2026-10-05 | 2.4 | Receiving lead | 2027-04-05 |
| RCV-03 | Receive without an ASN | 2 | 2026-10-05 | 2.4 | Receiving lead | 2027-04-05 |
| INV-04 | Cycle count and recount | 1 | 2026-09-14 | 2.3 | Inventory control lead | 2027-03-14 |
| SHP-01 | Load and ship a customer order | 2 | 2026-09-14 | 2.3 | Shipping lead | 2027-03-14 |
The register above is fictional. The column that matters most here is “Written for release”. With it, a release note that touches the receiving screens produces a list of SOPs to check in seconds.
Numbering revisions and keeping a change log
Use whole numbers for approved revisions and decimals for drafts, so revision 0.3 is a draft and revision 3 is the third approved version. Never reuse a number. At the end of each SOP, keep a change log: revision, date, what changed, why, and the change request or release that caused it.
The change log answers the question an auditor or a new manager asks first: when did this step change, and why? It also shows whether the SOP is changing with the system or lagging behind it.
The release gate
- Every change request names the processes it affects.
- The SOP owner for each process reviews the change in the test environment and marks the SOP as “no change” or “revise”.
- Revised SOPs go through the same user test and approval as new ones.
- Reference cards are reprinted with the new revision and date, and old cards are collected from the stations.
- Affected roles are retrained before the release reaches production.
- The release is approved for production only when steps 2 to 5 are recorded.
EPA’s guidance says SOPs should be updated and re-approved “whenever procedures are changed”, and also “systematically reviewed on a periodic basis, e.g. every 1-2 years”. Use both triggers. The release gate catches the changes you planned, and the periodic review catches the drift you did not.
Printed copies
Printed reference cards are the weak point in version control, because nobody can update paper remotely. Put the SOP ID, revision and date in the corner of every card, so an old card at a station is easy to spot. When a revision issues, the owner collects the old cards by hand. EPA’s guidance also stresses that current copies must be “readily accessible for reference in the work areas”, so a procedure stored only in an office system does not meet the need on its own.
Linking SOPs to training and audits
Each SOP should name the roles that must be trained on it and the practical check that proves they can do it. Each audit should use the SOP’s own steps as its checklist. When training, audits and SOPs point to each other, a finding on the floor leads straight to the document or the person that needs fixing.
EPA’s guidance notes that SOPs “can be used as a part of a personnel training program” and that they are “frequently used as checklists by inspectors when auditing procedures”. Both uses depend on the SOP matching the system.
The training link
Build a matrix of SOPs by role. Each cell says whether the role must be trained on that SOP and which practical check signs them off. When an SOP revision issues, the matrix gives the list of people to retrain. Our onboarding and training page describes the same matrix from the training side.
The audit link
A floor audit is a supervisor or quality lead watching a task and comparing it to the SOP, step by step. Use the SOP’s numbered steps as the audit form, and record for each step whether it was done as written.
| Audit result | Likely cause | Action |
|---|---|---|
| One person skips a step | Training or habit | Coach and recheck that person |
| Several people skip the same step | The step is wrong, unclear or slower than a workaround | Revise the SOP, or fix the system, then retrain |
| Everyone follows the SOP and errors still occur | The procedure itself is flawed | Process owner redesigns the step |
| The SOP describes a screen that no longer exists | Release gate was bypassed | Revise the SOP; review how the release was approved |
The second row is the important one. When many people deviate in the same way, the floor has usually found something the SOP missed. Treat it as a draft revision, not a discipline problem.
How often to audit
Audit by risk, not by calendar alone. Processes that move inventory value or customer orders, such as receiving, adjustments and shipping, deserve frequent audits in the first weeks after go-live, when habits are forming. Once audits come back clean for several cycles, lengthen the interval. Any release that changes a process resets its interval to the shorter one.
Deviations
Sometimes the SOP cannot be followed: a device fails, an item arrives in a form the procedure did not foresee. Give people a way to record a deviation, with what they did instead and who approved it. In regulated drug manufacturing, 21 CFR 211.100 requires that any deviation from written production procedures be “recorded and justified”. Even where no rule applies, a deviation log is a cheap and steady source of SOP improvements.
A rollout plan and checklist for rewriting your SOPs
Rewrite SOPs in the same sequence as the project: inventory what exists, map it to the new processes, draft in UAT, test, approve, train, and then review after go-live. Each step has an owner and a date in the project plan.
Sequence
- Inventory existing SOPs. List every current procedure, reference card and checklist, with its owner. Mark which processes the new system changes.
- Map old to new. For each new process, note which old SOPs it replaces, merges or retires.
- Assign owners. One owner per process, named before UAT starts.
- Draft in UAT. Owners draft from passing scripts and record every exception that testing exposes.
- Test with users. Two clean runs by people new to the task.
- Approve and register. Approval by the area manager and the quality or inventory owner; entry in the register with the system release.
- Train from the SOPs. Training uses the approved SOPs and cards, and trainers feed corrections back to the owners.
- Retire the old documents. Remove old binders and cards from the floor on go-live day, and archive them.
- Review after go-live. A short review a few weeks after go-live, using audit results and the deviation log, then the periodic cycle.
Checklist
- Every process in scope has one SOP, one owner and one ID.
- Every SOP uses the control block: ID, revision, effective date, system release, owner, approver.
- System steps quote the exact screen prompts and state the confirmation for each step.
- Every SOP has an exception table, a stop rule and a pointer to the downtime procedure.
- Safety warnings appear at the steps where they apply.
- Control points name the role, the timing and the record that proves them.
- Each SOP has passed a user test with someone new to the task.
- Reference cards carry the SOP ID, revision and date.
- The register lists the release each SOP was written for.
- Change requests cannot reach production until the affected SOPs are revised and roles retrained.
- Where rules apply (drug warehousing, food traceability plans), the written procedures they require are updated for the new system.
If your current SOPs predate the system you run today, the inventory and the user test will show where to start. The go-live readiness checklist includes reference cards at each station as a go-live condition. For help writing and testing procedures against your own screens, see our documentation service or describe your system and processes.
