[ ENTERPRISE PILLAR 03 ]
Convert scientific and user needs into a product and process whose critical knowledge can survive scale-up, transfer, and change.
This is the scientific and technical development capability; clinical execution and capital-facility delivery have separate pillars.
The capability framing below, its failure modes and the boundary with neighbouring pillars are SPEQ’s practitioner reading — not a regulatory requirement, and not an assessment of any organization.
THE CAPABILITY
The output of development is not the product. It is the justified account of why the product and the process are the way they are: which decisions were taken, what evidence stood behind each, and the boundaries within which each still holds. The material thing is the visible result; the durable one is knowledge with its uncertainty still attached to it. Everything downstream is a consumer. Assurance tests claims this pillar made, manufacturing operates inside limits it drew, the submission is largely its argument rewritten for a different reader, and every later change is judged against boundaries it set. When the knowledge is thin nothing fails immediately — the organization simply loses the ability to tell a real problem from ordinary variation, and it discovers that several years afterwards, in an investigation that cannot reach a conclusion.
The work finishes long before anything can grade it. Development asserts behaviour at scales and over durations it has not observed: a characterization run at one volume must hold at another, a shelf life is an extrapolation from months of data about years of storage, an operating region is a space of which only a handful of points were ever tested. The central deliverable is an inference, and the feedback that would score the inference arrives at full scale years later, by which time the team has dispersed and the reasoning survives only to the extent that somebody wrote it down while it still felt obvious. Against that sits the strongest convergence pressure in the enterprise. Each additional experiment is a week of delay on a milestone that people are measured against, so the questions that get dropped are precisely the ones whose answers only matter later — which is what made them droppable. And the thing that eventually fails is never the study that was run badly. It is the study nobody ran, and a decision not to run something produces no protocol, no report and no signature, so it leaves the file looking complete. Fluency in every constituent activity is entirely compatible with a portfolio whose gaps are invisible by construction.
Each of these happens with the individual branches below being run competently. That is what makes them capability failures rather than performance problems.
Batches, conditions, results and charts are all present. The sentence explaining why this parameter matters and that one does not was never written, because at the time it was obvious to everybody in the room. Five years later the range can be reproduced but not defended, and the only route to moving it is to repeat work that was already done once, by people who understood it better than anyone available now.
The attribute list, the parameter ranges and the risk assessment are carried across from a predecessor or a platform. It is cheap, usually reasonable and frequently right, which is why it passes review without comment. What travels with it is a set of assumptions about a formulation or a process that differed in exactly the one respect that turns out to matter, and nothing in the copied document records that it was copied.
The receiving site gets the batch record, the specification, the equipment list and the qualification package. What it does not get is the failed variant, the abandoned approach, the reason behind an odd hold time, or the excursion that shaped a range. The first campaign is a competent imitation. The first unexpected result has nothing behind it to interpret with, so it becomes an investigation where it should have been a recognition.
A method built to characterize a handful of lots is asked to support disposition decisions on hundreds of them. Variability that read as scientific noise during development becomes a steady stream of results outside limits that are about the measurement rather than the material, and the investigations they generate look like product problems for as long as it takes somebody to think of questioning the method.
WHERE THIS STOPS
Development stops when the argument is complete, not when the product works, and the gap between those two dates is where most of the disputes live. The sharpest seam is with assurance. The studies look identical — same equipment, same analysis, often the same people — and the difference is purpose. One establishes where the limits ought to be; the other confirms the process holds inside limits already fixed. Run them together to save time and the organization permanently loses the ability to say which claim was tested and which was assumed. The second seam is the receiving manufacturing site, where nobody can name the moment a transfer is finished: batches are running, the site is producing, and the sending unit is still answering questions it no longer has budget to answer. The third is the analytical method, which belongs here while it is being defined and to the laboratory once it is routine, with no natural event to hang the handover on. The fourth is clinical, where this pillar characterizes the thing and clinical produces the evidence of what it does in people; the investigational supply is where the two touch.
Documentation records that studies were performed and what they produced. Knowledge is the part that lets a later reader act: why this was judged to matter, what was tried and rejected, where the boundary of the claim sits, and what would change the conclusion. A file can be complete and carry almost no knowledge, which is the normal state of a development history assembled under a deadline.
Mostly a posture, and treating it as a method is how it turns into paperwork. The commitment is to understand which attributes matter to the patient or user and why, then to build controls out of that understanding rather than out of convention. The tools usually associated with it are useful, but an organization can run every one of them and still be unable to say why a limit sits where it does.
When the process moves outside the conditions the region was inferred from: a different scale, a changed equipment geometry, a material with a different impurity profile, or simply enough elapsed time that the assumptions behind it have not been rechecked against how the process now behaves. It is a claim about a region, and like any claim it has a validity boundary that has to be maintained rather than assumed.
This pillar owns the scientific judgement of whether the changed material is the same in every respect that matters, because it holds the characterization and the reasoning behind each attribute. What it does not own is the decision to accept the change, or the obligation to tell anyone about it. Those sit with governance and regulatory, and the three assessments should reference one another rather than run in parallel.
CAPABILITY BRANCH MAP
Turning a target or an unmet need into a defined candidate: intended use, the population it serves, feasibility, and the early characterisation that decides whether to continue. It is where the product first becomes a specific thing.
The intended use written here propagates into classification, clinical design, labelling and the entire control strategy. Changing it later is not a document edit — it invalidates the reasoning built on top of it.
Designing the thing itself: dosage form or device architecture, materials, components, specifications and the human factors that determine whether it can be used correctly by the people who will use it.
Design fixes most of the risk and most of the cost before manufacturing ever begins. A material chosen for availability, or a device whose use error was never studied, becomes a control burden that operations carries for the product’s whole life.
Developing the measurements the product will be judged by: what attributes matter, which methods can measure them reliably, and how those methods will be validated, transferred and maintained across the lifecycle.
Every release decision, stability conclusion and process claim rests on a measurement. A method that was never robust does not fail loudly — it produces plausible numbers that quietly widen the apparent variability of everything downstream.
Understanding how the process behaves: which parameters matter, how they interact, what scale changes, where the operating range sits, and what control strategy holds the product inside its specification.
Process understanding is what makes validation an argument rather than a demonstration. Without it, an organisation can prove three batches worked and cannot explain why the fourth did not.
Working from the target product profile backwards: which quality attributes are critical, what risk each carries, which experiments resolve the uncertainty, and how that knowledge becomes controls rather than a report.
It is the difference between a control strategy that can be defended and one that can only be described. Where the linkage from patient need to critical attribute to control is explicit, a change can be assessed; where it is not, every change is a guess.
The formal engineering discipline for products developed under a device-style framework: user needs, requirements, design outputs, verification, validation, design reviews, transfer and the design history that records it all.
Design controls make the reasoning auditable — every requirement traceable to a need and to the evidence that it was met. It is also the framework a regulator will use to judge whether a design change was assessed properly years after the fact.
Moving product and process knowledge to the site that will make it: assessing site fit, equipment equivalence, method transfer, scale differences and the comparability of what comes out the other end.
The receiving site inherits the outcome but not the reasoning. Where only the procedure transfers and the rationale stays behind, the site can execute the process but cannot judge what to do when it deviates.
Establishing how the product changes over time and under stress, what packaging protects it, what shelf life and storage conditions follow, and the ongoing programme that confirms the conclusion holds.
Shelf life is a claim made to patients and regulators, and it is one of the few product attributes that cannot be re-tested after the fact. When stability is wrong, the correction is a market action rather than a document revision.
Preserving the reasoning behind the product: decisions taken, assumptions made, models used, experiments that failed, and the traceable record of why the process is set the way it is.
This is the knowledge that determines whether a change can be assessed in ten years. Losing it does not stop production; it makes every future change either unjustifiably conservative or quietly unsafe.
CONNECTED OPERATING MODEL
MATURITY ORIENTATION · SPEQ SYNTHESIS
HIGH-VALUE INTERSECTIONS
SOURCE BASIS
REGULATORY BASIS
The 14 standards SPEQ maps to this pillar, and the 3 regulatory bodies behind them. Which standards belong to a pillar is a SPEQ judgement; the bodies, disciplines and industries below are read from the standards themselves.
ICH · IEC · ISO
Ranked by how many of this pillar’s standards each one cites — showing 10 of 36.
These 15 standards govern the system classes this pillar depends on rather than the pillar itself. The distinction matters: a standard that governs a system is not thereby a standard of every capability that uses it.
21 CFR Part 21121 CFR Part 58OECD GLP Principles21 CFR Part 11EU GMP Annex 11ISPE GAMP 5 (2022)MHRA GxP DI (2018)PIC/S PI 041-1ISO/IEC 17025:2017USP <1058>ICH Q2(R2)ICH Q9(R1)ICH Q10FDA Process Validation Guidance (2011)ASTM E2500
PROFESSIONAL · READINESS ORIENTATION
Rate observable operation from 0 (not established) to 4 (adaptive). The protected output prioritizes operating dimensions and evidence—not a compliance score.