Constraints do not appear after design; they are part of the design problem. A solution that works only with unlimited money, materials, or time is not yet ready for delivery.
During development, a team may discover that a planned material is unavailable, a software integration creates vendor dependence, or a service experience requires more staff than the business can supply. The answer is not indiscriminate cutting. It is identifying which limits are mandatory, which decisions can change, and which alternatives preserve the core outcome. Systems engineering uses trade studies to compare effectiveness, cost, risk, and feasibility under explicit constraints [R01]. Here we adapt that reasoning to smaller products, services, and processes.
1. Distinguish constraints, objectives, assumptions, and preferences
A constraint limits what may be done; an objective describes what we seek to achieve or optimize; an assumption is a condition not yet confirmed; and a preference favors an option without necessarily making others infeasible.
| Statement | Initial classification | What to verify |
|---|---|---|
| The team has an approved budget | Financial limit | Actual amount, scope, and reserves |
| We want to deliver as soon as possible | Objective or preference | Which date and delay cost matter |
| The vendor will deliver parts in five days | Assumption | Quote, capacity, and reliability |
| The solution must meet an applicable obligation | Mandatory condition | Competent source and jurisdiction |
| The client prefers a blue interface | Preference | Whether this serves a real need |
Classification does not solve the problem; it tells you what to negotiate, test, or never ignore. A contract deadline might be real; an internally suggested date may be adjustable. Do not present internal preferences as legal requirements.
2. Include financial and total-cost constraints
Separate development budget, cost per unit or transaction, operations, maintenance, repair, support, and retirement or replacement. An inexpensive product to build can be expensive to run. For physical goods, include waste, inspection, packaging, and logistics. For software, consider hosting, security, support, integrations, changes, and vendor dependence. For services, account for actual staff hours and exception handling.
Use ranges where evidence is weak. Do not turn an early quotation into a guarantee. Look at recurring costs and the consequences of failure, not only purchase prices. Pillar 2 business viability and Pillar 4 budget methods provide estimates; this article addresses how those estimates change the design.
Cost assessment depends on demonstrated capabilities: the resource and capability-gap inventory helps identify what exists, what is missing, and when it is available before settling on a design.
3. Account for materials, manufacturing, and supply
In physical products, changing material can affect strength, compatibility, finish, safety, maintenance, and future supply. A substitution is not justified solely because it is cheaper or easier to obtain. Record the required properties and test whether the substitute preserves them.
Manufacturing introduces tolerances, batch variation, equipment limits, inspections, instructions, and worker safety. A component that works in one prototype may fail across one hundred units produced with ordinary variation.
For services, the equivalent could be an ingredient, a scarce skill, or an outside provider: replacing it may change experience, capacity, or quality. Distinguish interchangeable components from those requiring renewed validation before delivery.
4. Consider technical and human constraints
For digital products, examine device compatibility, connectivity, storage, interfaces, performance, technical capacity, security, and vendor continuity. If the solution only works with reliable internet, determine how failure will be communicated and whether recovery is needed.
For services, examine shifts, training, fatigue, languages, accessibility, dependence on specific people, and coordination across functions. A screen may save seconds for customers while adding hours of staff work behind the scenes. Human capacity is part of the design, not a problem to be addressed only after launch.
Do not mistake a current limit for a law of nature. A vendor can sometimes be replaced and a capability developed through training, but these choices carry cost, delay, and risk.
5. Separate mandatory conditions from negotiable ones
Safety, accessibility, data protection, and regulatory requirements vary by product and jurisdiction. Research what applies. Do not infer universal obligations from another country's guide, and do not presume an exception because a project is small.
A mandatory condition should not receive a low score that can be offset by speed or visual appeal. If a design cannot operate lawfully or safely, reject or redesign it. Negotiable constraints allow trade-offs: less customization in exchange for simpler delivery, or a narrower initial service area to maintain quality.
A useful classification is “blocking,” “needs evidence,” “negotiable,” and “desired enhancement,” with reasons. The Pillar 4 guide to legal and operational requirements covers research of authoritative obligations.
When a condition directly affects safety, reliability, or the quality of the outcome, translate it into verifiable quality criteria. An accepted constraint does not automatically make the final performance acceptable.
6. Compare design trade-offs without manufacturing certainty
A trade-off occurs when improving one property worsens another. Higher strength may add weight or cost; more features can increase maintenance; longer opening hours may demand more staff. Compare options against core value and consequences.
| Decision | Potential benefit | Potential cost or loss | Needed test |
|---|---|---|---|
| Alternative material | Better availability | Unknown strength | Relevant-use testing |
| Initial manual process | Lower software investment | Limited capacity | Load and error pilot |
| Single-area service | More controlled timing | Narrow initial reach | Real delivery trial |
| External API dependency | Faster initial integration | Vendor changes and outages | Contract, recovery, and testing |
Do not hide unacceptable risk inside a favorable average score. Record missing information, plausible scenarios, consequence ranges, reversibility, and approval authority. NASA's trade-study concept is a methodological reference, not a requirement to adopt full aerospace documentation [R01].
If a trade-off makes the original proposal infeasible, return to comparing solution alternatives rather than forcing one option. The review should preserve the reasons and costs of the change.
7. Manage a constraint discovered during development
When an unexpected limit appears:
- Confirm what changed and what evidence supports it; distinguish verified fact from warning.
- Determine which requirement, actor, and outcome are affected.
- Classify whether safety or compliance blocks delivery or alternatives remain possible.
- Generate options: substitute, simplify, change the workflow, alter the date, limit capacity, or stop.
- Assess costs, new tests, residual risk, and human impact.
- Record the decision, owner, and evidence needed before choosing the new design.
- Update requirements, prototypes, budget, design, and version coherently.
Do not promise the same date, cost, and quality when a fundamental condition has changed. Replanning is not failure; it acknowledges that commitments were based on information no longer valid.
8. Give extra attention to irreversible or costly-to-reverse choices
A screen can often be reworked. Recalling a physical part, reversing a data leak, or renegotiating a contract may be costly or impossible. Identify irreversible choices and test their assumptions earlier.
A simple register can capture decision, reversibility, evidence, outstanding test, cost of correction, and owner. If reversibility is unknown, do not assume that it is easy. For high-stakes choices, obtain appropriate specialist review in materials, safety, or applicable law.
Worked example: replacing a material in a physical product
A fictional company designs a protective enclosure for an indoor device. It plans to use material A, but the supplier reports that A will not be available on time. Material B is offered at lower cost. Accepting B solely for availability would be a mistake.
The team records required properties: intended impact resistance, dimensional stability, compatibility with components, thermal behavior, and real-world exposure. It separates properties supported by manufacturer documentation from those needing tests. It also checks manufacturing lead times, rework, and assembly failure modes.
Illustrative decision: use B for fit prototypes and initial tests only, not in customer deliveries without evidence that relevant use conditions are satisfied. If an essential condition fails, find another material or revise scope. The decision preserves learning without asserting that two materials are equivalent because they look similar.
The same logic applies to software: replacing a library means examining functionality, security, licensing, compatibility, and maintenance, not merely whether sample code compiles.
Common mistakes and limitations
Treating the budget as the only criterion; assuming materials are available; ignoring staff and training; accepting untested substitutes; inventing legal requirements; hiding lifecycle costs; confusing “can be demonstrated” with “can be delivered”; scoring options arbitrarily; persisting because of sunk costs; recording the decision but not the consequences.
No matrix removes the need for professional judgment and appropriate testing. Physical, sanitary, technical, and legal risks may require specialist expertise.
What you should have when finished
A design constraints matrix with evidence sources, classification, affected requirements, alternatives, grounded costs and risks, outstanding tests, decisions, and owners. Someone should be able to explain why the final solution differs from the original idea without claiming that the constraints magically disappeared.
References
- [R01] NASA. Systems Engineering Handbook — Trade Studies and System Design Processes (public methodological guidance, not adoption of aerospace standards wholesale). https://www.nasa.gov/reference/4-0-system-design-processes/