A sale is not completed when someone clicks “buy.” Operations must turn the commitment into a delivered outcome, verify that delivery, and remain ready to respond if something goes wrong.
Customers may see an order confirmation and a delivery; behind the scenes, checks, assignments, purchasing, preparation, controls, and decisions all take place. If those activities are disconnected, an order may appear “sold” even though no one has reserved the materials, prepared the service, or communicated a delay. This article shows how to map the full flow of a real operation, identify what each role hands off, and anticipate exceptions before they turn into improvisation. Process mapping guidance from ASQ and the National Institutes of Health (NIH) describes how to visualize inputs, activities, decisions, handoffs, and outputs [R01][R02].
1. Distinguish an operating process, value flow, project, and activity
An operating process is a repeatable sequence that transforms inputs into outcomes under defined conditions: handling each booking, assembling each order, or enabling each new account. An activity is one unit of work within a process, such as verifying an address. A project seeks a bounded result and is not normally performed in exactly the same way every day: setting up the production kitchen for the first time is a project; preparing daily orders is a process. A service is what the customer receives and experiences; several processes may support it.
By value flow, we mean the movement of materials, information, and work that contributes to fulfilling a promise. Not every step adds value the customer can see, but some invisible checks are necessary for safety, compliance, or reliability. Do not remove a required inspection simply because it does not make another meal appear faster.
Draw the current state first, not the ideal process from a manual. If someone reviews orders on WhatsApp and then copies them into a notebook, represent that duplicate entry: hiding it would prevent you from understanding errors and waiting time. ASQ recommends reviewing process maps with the people who perform the work, not only those who design it [R01].
2. From order to delivery: set boundaries and states
An operation does not necessarily begin on a website. One map might start when a request arrives, another when an order is accepted, and another when inventory replenishment starts. Choose a boundary for each map and document which earlier process creates its input. For an accepted order, the following is a useful starting sequence:
Request → scope and capacity check → acceptance and commitment → materials/slot reservation → preparation or service delivery → quality check → dispatch or activation → delivery confirmation → payment/receivable record → follow-up and closure.
Payment may occur before, during, or after delivery: a prepaid order, cash on delivery, and a B2B invoice due in 30 days require different states. Therefore, do not draw “payment” after “delivery” as the only possible path. Represent payment authorization, payment collected, and outstanding balance where relevant.
A minimal set of shared states might be requested, pending verification, accepted, in preparation, ready, in delivery, delivered, and closed, with on hold, canceled, incident, and return when applicable. Each status change needs an actor, timestamp, and exit condition. Avoid the ambiguous status “done”: does it mean produced, dispatched, or received?
3. Record inputs, outputs, owners, and handoffs
At each stage, answer five questions: what triggers it, what does it need, who decides, what does it produce, and who receives it? Add the system, evidence, and deadline if they matter to execution. A simple sheet is enough to begin:
| Stage | Input and condition | Owner | Verifiable output | Recipient / control |
|---|---|---|---|---|
| Accept order | Details, coverage area, time, and available slot | Sales/customer support | ID and confirmed promise | Operations; slot reserved |
| Prepare | Accepted ID, recipe, and available inventory | Kitchen/production | Prepared and identified order | Quality check; batch or composition verified |
| Verify | Order, requirements, and evidence | Quality lead | Approved or held due to an incident | Dispatch; documented decision criterion |
| Dispatch | Conforming order, address, and route | Delivery | Departure logged | Customer; status updated |
| Close | Receipt confirmation and payment status | Administration/support | Delivery and balance can be reconciled | Archive and support if needed |
One person may perform every function. An owner does not imply a department; it indicates who is authorized and expected to act at that point. In a one-person business, separating functions in a table still helps distinguish “I made the sale” from “I confirmed I could fulfill it.”
To prevent information loss at a handoff, define a minimum information package: order ID, product, quantity, version or specification, relevant conditions, promised time, essential contact details, and payment status. Do not collect or share more personal information than needed. When something changes, specify who updates the single source of truth and who must be notified.
4. Adapt the flow to products, services, and digital solutions
The general sequence changes depending on what you deliver:
| Type | Customer-visible work | Often-invisible work | What confirms closure |
|---|---|---|---|
| Physical product | Selection, purchase, receipt | Sourcing, inventory, picking, packaging, transportation | Correct item and verified delivery |
| In-person service | Booking, visit, outcome | Staff assignment, travel, equipment, authorizations | Service performed and agreed condition met |
| Digital solution | Payment, access, use | Provisioning, permissions, integrations, support, monitoring | Appropriate access and core function available |
| Hybrid service | Product plus recurring assistance | Inventory, staffing, and follow-up coordination | Physical delivery and contracted service recorded separately |
In software, “dispatch” may mean enabling an account and sending instructions; logistics mostly concerns information and technical capacity. In an in-person service, specialist availability and travel time are part of the actual offering. In manufacturing, batch control and traceability can be critical. Do not force everything into a diagram of physical boxes when the work happens in systems, people, or agreements.
An operating process map differs from a customer journey: the latter examines the customer's actions, expectations, and friction points; the operating flow shows how the business creates the response. They should align, but they are not interchangeable tools.
5. Distinguish inbound, internal, outbound, and reverse logistics
The Association for Supply Chain Management (ASCM) distinguishes inbound, outbound, and reverse flows and emphasizes the coordinated movement of goods, information, and resources [R03]. In a small operation, apply that distinction as follows:
Inbound logistics must connect with receiving, inventory, and replenishment controls so commitments do not rely on unavailable stock.
- Inbound: purchasing, transporting, and receiving inputs before production; check quantities, condition, and applicable requirements.
- Internal: storage, movement into production, preparation, where work in progress is held, and equipment availability.
- Outbound: order consolidation, packaging, route assignment, transportation, delivery, and proof of receipt.
- Reverse: returns, repair, reuse, reimbursement, safe disposal, or materials recovery, depending on the product and applicable rules.
Not every returned item may go back into saleable inventory: delivered foods, customized items, damaged products, and digital data each have different conditions. For a service, a “return” may instead mean rescheduling, correcting the work, or providing a justified remedy; there may be no physical item coming back.
Define which part you control and which you delegate. If a third party transports orders, retain a checkpoint for confirming they received the shipment, departure time, delivery status, and failure handling. Delegating execution does not eliminate the need to fulfill the promise made to the customer, although specific legal and contractual obligations depend on the case.
6. Locate inventory, capacity, scheduling, and proof of delivery
A well-described flow answers before accepting an order: do we have the materials, people, time, and coverage to fulfill it? Total inventory is not the same as usable availability: some items may already be reserved, damaged, expired, or stored elsewhere. The same applies to staff who appear “available” but already have travel commitments.
Distinguish four objects:
- Available inventory: physically suitable inputs that are not already committed.
- Reservable capacity: effective slots per time window after breaks, earlier tasks, and quality limits.
- Scheduling: assignment of orders to shifts, resources, and external partners.
- Delivery evidence: evidence proportionate to the risk, such as a confirmation, a technical activation log, or acceptance of completed work.
Do not promise 15 deliveries in one hour if the kitchen can prepare 15 but the delivery team can complete only 9. Useful capacity depends on the limiting stage and the schedule. Nor is it always necessary to keep every resource fully occupied: spare capacity can protect punctuality when things go wrong. Detailed capacity and demand calculations belong in the dedicated P6 articles; here, simply record where they affect the flow.
7. Design exception paths with decisions and communication
A flow that accounts only for success describes an aspiration, not a ready operation. Map the normal route and what happens when a condition fails. Use this structure: detection → containment → decision → owner → communication → record → resumption or closure.
| Exception | First responsible action | Decision and limit | Minimum evidence |
|---|---|---|---|
| Input unavailable | Hold preparation; verify actual stock | Substitute only if permitted and the customer agrees; otherwise cancel or reschedule | Shortage, alternative offered, decision |
| Payment rejected | Do not mark payment as collected | Authorized retry, alternative method, or cancellation according to policy | Status and receipt, without storing sensitive card data |
| Supplier delayed | Update the expected availability | Reschedule and communicate before making another promise | New ETA clearly identified as an estimate |
| Incorrect address | Pause dispatch or contact customer | Correct when verifiable and feasible; handle changes in cost and timing | Validated address and agreement |
| Change/cancellation | Check preparation status and terms | Accept, reschedule, or apply disclosed policy | Request time, decision, communication |
| Technical failure | Stop unsafe operation; activate appropriate backup | Restore service, log pending order, or communicate unavailability | Incident, owner, actual state |
A safety exception is not treated like a commercial preference: it may require stopping service, isolating a product, and following applicable protocols. In food service, never substitute an ingredient when restrictions or allergies are involved without verified information and competent controls. WHO food-safety principles include cleanliness, separation, and suitable preparation practices [R04].
Communicate what happened without promising a time you cannot yet support. A useful message states the situation, confirmed options, and the next update. Do not make customers investigate across departments to learn where their orders are.
8. Add checkpoints that prevent losses and harm
You do not need a signature at every step; you need checks where the cost or harm from an error warrants them. Place them at three moments: before promising, before a hard-to-reverse action, and before declaring delivery complete.
If you need to define what makes a delivery conforming, see product and service quality criteria; this workflow identifies where to check them in operation.
| Checkpoint | What to verify | If it fails |
|---|---|---|
| Acceptance | Coverage, price, availability, and feasible time | Do not confirm an impossible promise |
| Start of preparation | Correct order, suitable inputs, applicable restrictions | Block and resolve before production |
| Dispatch | Identification, quantity, condition, and address | Hold, correct, and recheck |
| Delivery/closure | Receipt, incident, and payment or outstanding balance | Do not automatically close the obligation |
On a website, the equivalent control might prevent enabling an account with incorrect permissions. For industrial equipment, it may involve a regulatory inspection: detailed criteria require appropriate sources and qualified personnel. Avoid confusing an operational control with a formal audit or the initial solution-design tests already covered in Pillar 5.
9. Measure cycle time, waits, work in progress, and errors
The initial goal is to make the operation visible, not yet to prove a cause or optimize it. Useful definitions:
- Cycle time: how long an activity takes or how often a unit is completed, depending on the boundary chosen. State start and end precisely.
- Total response time (lead time): from the selected initial event—for example, acceptance—to delivery, including waits.
- Waiting time: elapsed time with no active work on that order.
- Work in progress (WIP): accepted orders that have not reached closure, without counting canceled orders as completed.
- Error rate: orders with incidents / orders assessed × 100. Define what counts as an incident.
- Promise fulfillment: deliveries within the agreed window / eligible deliveries × 100.
Illustrative timing example: order accepted at 12:00 p.m., preparation started at 12:10 p.m., preparation completed at 12:36 p.m., dispatched at 12:40 p.m., and received at 12:55 p.m. Preparation time is 26 minutes; acceptance-to-receipt lead time is 55 minutes. The 10 minutes before preparation and the 4 before dispatch are observed waits, but we cannot claim that the entire remaining time is hands-on work: transportation and quality checks have not been separately timed. This prevents adding incomparable figures together.
At minimum, log the ID, timestamp, state, owner, and any incident. Keep measurement proportionate; do not turn operations into a collection of stopwatches that gets in the way of delivery.
10. Validate the flow with one complete simulation and one failure
Before declaring it ready, run a full end-to-end trial order with someone acting as customer and someone observing the flow, even if one person normally performs everything. This is not the same as validating market demand or conducting a formal product test: it tests process coordination.
After validating the workflow, turn its critical controls into operating procedures that can be performed and supervised.
- Define a realistic request, a time, and closure criteria.
- Follow every state change rather than skipping “obvious” steps.
- Verify every handoff: who knows what to do, with what information, and by when?
- Simulate a shortage, cancellation, or schedule change before finishing.
- Check whether the promise was blocked when necessary, who notified the customer, and what was recorded.
- Repeat the normal scenario and record decisions that require a detailed procedure.
Worked example: a recurring lunch order with two incidents
Almuerzo Cerca — entirely hypothetical example. At 10:05 a.m., order AC-042 arrives with a promised delivery window of 12:45–1:00 p.m. Support verifies service area, quantity, and slot, then accepts the order at 10:08 a.m. While reserving ingredients, the kitchen notices at 10:12 a.m. that a side dish is unavailable. It does not substitute silently: staff verify alternatives that are safe and compatible with the information available, ask for the customer's agreement, and record the change at 10:20 a.m. If safety or compatibility cannot be confirmed, the order is placed on hold and rescheduling or cancellation is offered.
At 12:15 p.m., the customer asks for delivery after 1:15 p.m. Dispatch checks whether this can be accommodated without compromising food safety or other customers' delivery commitments; it communicates a new window only after confirming feasibility. Assume the customer agrees to 1:15–1:30 p.m., preparation ends at 1:05 p.m., the check occurs at 1:07 p.m., dispatch happens at 1:12 p.m., and the order is received at 1:25 p.m. The order is marked delivered, with payment reconciled or an outstanding balance identified. Recording “delivered” alone is insufficient: the substitution approval, schedule change, and delivery evidence must be retained.
Simulation outcome: two essential controls emerge: do not accept substitutions without the relevant confirmation, and do not change delivery windows without rechecking capacity. The flow remains the same; the exact steps for performing each control will be defined in an SOP.
Deliverable A: swimlane map
This table is a textual representation of swimlanes; it can be turned into a visual diagram with rows for each owner and arrows for handoffs.
| Sequence | Customer | Sales / support | Operations / kitchen | Quality / delivery | Shared record |
|---|---|---|---|---|---|
| 1 | Submits request | Verifies and accepts | Confirms capacity | — | ID, time, terms |
| 2 | Responds to proposed change | Requests approval | Detects shortage, reserves inputs | — | Documented decision |
| 3 | Requests a new window | Notifies customer | Adjusts schedule | Confirms route and feasibility | Agreed window |
| 4 | — | — | Prepares and labels | Checks and dispatches | Status and conformity |
| 5 | Receives order | Handles incidents | — | Confirms receipt | Delivery and payment/balance |
Deliverable B: reusable process record
| Field | What to enter |
|---|---|
| ID / name / version | Flow identifier and actual review date |
| Trigger / end | Event that starts it and evidence that closes it |
| Input / output | Data, materials, and outcomes for each stage |
| Owner / recipient | Who acts, decides, and receives each handoff |
| States | Allowed states and conditions for changing them |
| Constraint | Capacity, hours, safety, coverage, privacy |
| Checkpoints | What is checked and what prevents moving forward |
| Exceptions | What triggers a hold, who decides, and how to communicate |
| Evidence | Where records are kept and their retention under applicable rules |
| Metrics | Definitions of time, incidents, and fulfillment |
Mistakes and limits. A map that is too general is useless (“process order” as its only step); one that documents every click becomes impossible to maintain. Do not measure capacity using only the fastest stage, or mistake a sent notification for confirmed receipt. Physical-product logistics do not necessarily map literally onto digital services. Food safety, consumer protection, privacy, transport, and other obligations must be checked for the relevant country and activity; this map does not replace those checks.
What you should have by the end
A bounded flow reviewed with the people who carry it out; a map of owners, inputs, outputs, and states; risk-based checkpoints; exception paths; a small set of clearly defined measures; and evidence from a normal simulation plus a failure scenario. The next step is not to draw more arrows: it is to help people execute and supervise critical steps consistently.
References
- [R01] ASQ. Flowchart: Process Flow Diagrams & Maps. https://asq.org/quality-resources/flowchart
- [R02] NIH, Office of Research Services. Process Mapping. https://ors.od.nih.gov/OD/OQM/Pages/Process-Mapping.aspx
- [R03] ASCM. What Is Supply Chain Logistics? https://www.ascm.org/topics/logistics/
- [R04] World Health Organization. Five Keys to Safer Food. https://www.who.int/activities/promoting-safe-food-handling/five-key-to-safer-food