Process & Operations · Operational end-to-end processes
Every station in the material flow carries a condition: documented status, monitored storage conditions, an approved manufacturing instruction, a tested batch, a documented release. If one of them is missing, the material may not be released — regardless of its actual quality. Significon designs, implements and validates the processes in which that evidence comes about: from incoming goods to release.
The material flow
Quality control paces the flow: it decides on release and onward movement and is therefore part of the flow it steers. Above all of it sits traceability — batch, status, location and time have to match at every station, not only in the report.
The problem
The four patterns below sit in almost every system landscape that has grown over years. None of them is an individual's mistake. Each follows from the same assumption: that the evidence can be assembled later.
01
Material status, deviations and approvals are kept in spreadsheets, mail threads and paper travellers, because the leading system does not hold them in the form they are needed in. The process then runs in two lanes: the one the system knows, and the one the audit asks about.
02
A label is a regulated information carrier, not a printout. Where templates are maintained locally and versions distributed by email, version control, audit trail and an unambiguous approval are missing. One wrong label stops the batch — and the question in an audit is not who printed it, but which version was approved.
03
The mapping study exists, the route is validated, the report is signed. Then the packaging changes, or the forwarder, the season, the warehouse layout. Qualification is a state with validity, not a document with a date — without a link to change control it lapses and nobody notices.
04
The requirements from Delegated Regulation (EU) 2016/161 and from the DSCSA were in many cases placed as a separate system next to the existing printing and warehouse processes. The result is two data streams that do not run in sync, and traceability that holds on paper and shows gaps in a recall.
Three situations
A data logger shows twelve hours above the permitted range. The batch is standing at the customer’s goods receipt. What decides now is not the temperature but the state of the data: which packaging specification was qualified, which route was validated, which stability data cover the excursion, who is entitled to decide? If those facts do not sit together, an assessment turns into a hold — and the hold into a delivery delay that nobody catches up on.
The inspector’s first question is the batch. The second is the chain: which manufacturing instruction in which version, which label in which version, which storage conditions over which period, which test result, who released it and when. A process that holds this chain answers in minutes. One that does not answers in weeks — and every week produces the next question.
A wholesaler reports a pack whose identifier cannot be verified. What counts now sits in GDP Chapter 6 — Complaints, Returns, Suspected Falsified Medicinal Products and Medicinal Product Recalls — and in GMP Chapter 8 — Complaints, Quality Defects and Product Recalls: who shipped which packs where, in which aggregation, with which status? The ability to recall is not a statement you make. It is a query that returns a result within hours — or fails to.
The approach
So we do not start at the document but at the point where the information arises. First the process picture across system boundaries: which station produces which evidence, who decides, which system leads. From that the requirement in URS and FS — in the same language in which testing will later happen. Then the architecture, and only after that the implementation.
What stands at the end is not a validation project but validation documentation that accrues, because process and system were built from the start against the condition they have to meet.
AI appears in these processes only as a framed step: defined input, human approval, audit trail. The binding frame for it today is Annex 11 — not a draft.
Which condition has to be met is written in GMP and GDP. Which system meets it is the follow-up question. We answer them in that order, because the reverse order bends the condition to fit the system function that happens to exist.
For every piece of information there is exactly one source that holds it and one path along which it is passed on. Dual maintenance is not redundancy but two truths about the same batch — and in an audit the weaker one counts.
A qualified state only holds if every change to route, packaging, equipment or warehouse layout feeds back into it. That coupling belongs in the design, not in operations — added later it becomes a collection task without an owner.
The solution
The flow is one; the work on it divides into six sections. Each has its own condition, its own systems and its own evidence.
Procure-to-Pay
Only qualified suppliers deliver material, and at goods receipt, testing decides whether the batch may enter the flow. We design supplier qualification and goods receipt testing as a coupled process — including the qualification of contract manufacturers and contract laboratories.
Store-to-Ship
From receipt through storage under monitored conditions to dispatch, material status, cold chain, labelling and traceability all hang on the same data. We design the processes for it, carry out the qualification of transport and storage, and put labelling and serialisation on one path instead of two.
Plan-to-Produce
Recipes and resources are not master data, they are the manufacturing instruction. That puts them under the same control as the result: version, approval, change, evidence. We design the regulated parts of the manufacturing process and implement them in validated form — with URS and FS that the later testing reads from.
Sample-to-Release
Release paces the flow. It needs test results that have demonstrably travelled from the laboratory into the releasing system — not retyped, not emailed. We design the testing processes in the digital laboratory and the integration between quality management and LIMS.
Order-to-Cash
What gets sold is not a quantity, it is a released batch — at the right time, at the agreed price, to the right customer. We design batch allocation and dispatch release so that traceability reaches the customer and carries a recall when it matters.
Engineering & Maintenance
Manufacturing is only permitted on qualified infrastructure. We are building this field out toward predictive maintenance: equipment condition data, AI-assisted anomaly detection and demand-based spare parts ordering, tied to qualification status instead of a fixed calendar schedule.
Why Significon
Whoever writes the URS knows the test that will later check it. So requirement and specification are written here in the form in which testing happens — the validation documentation accrues during the work, not after it.
We do not sell software. Whether ERP, warehouse management, LIMS or a vendor label management system: we design process, architecture and integration, and in system selection we sit on your side of the table.
What we do not deliver, we say in the first conversation and not during the project. A request that falls outside is cheaper for both sides than a proposal that has been made to fit.
Which station in your material flow does not carry the evidence?
A conversation in which we walk the chain once — not a proposal.