Product Development: free practice, theory and problems

1. Process and customer needs

What is it about?

Product development doesn't start with a solution, but with understanding what the user actually needs. Experience shows that changes become steadily more expensive the later in the process you make them – a change in the concept phase costs almost nothing, while the same change after production ramp-up can cost a hundred times as much. That's why good development teams spend a lot of time early on gathering and understanding customer needs before they even start thinking about solutions.

Concepts and formulas

• Customer need: a statement in the customer's own words about what they want to achieve (e.g. "I want the backpack to keep out rain"), independent of any solution.
• Specification (requirement): a measurable quantity derived from the need, with a unit and a target or limit value (e.g. "waterproofness ≥ 5000 mm water column").
• Importance (weight) $w_i$: how much the customer values need $i$, often found through a survey or pairwise comparison. The number of pairwise comparisons of $n$ needs is $\binom{n}{2} = \frac{n(n-1)}{2}$.
• House of Quality (QFD): a matrix linking customer needs (with weight $w_i$) to technical characteristics through a relationship strength $r_{ij}$ (typically 1, 3 or 9 for weak, medium, strong). Technical importance of characteristic $j$: $I_j = \sum_i w_i r_{ij}$.
• Function analysis: breaks the product down into what it must do (functions), not how – this opens up the solution space.
• Stage-gate: development is split into phases separated by decision points (gates) where the project is reviewed before being allowed to continue.

How to solve the problems

1. Clearly separate need (customer's language) from specification (measurable, with a unit).
2. When ranking or weighting needs: use pairwise comparison or given weights directly.
3. For QFD problems: multiply each need's weight by the relationship strength to the technical characteristic, and sum.
4. Remember that "what the product must do" (function) comes before "how it will do it" (solution).

Example

Three customer needs have weights $w_1=0.5$, $w_2=0.3$ and $w_3=0.2$. In the QFD matrix they have relationship strength 9, 3 and 1 to the technical characteristic "weight". What is the technical importance of weight? $I = 0.5\cdot 9 + 0.3\cdot 3 + 0.2\cdot 1 = 4.5 + 0.9 + 0.2 = 5.6$.

Answer: technical importance 5.6 – this is the highest among the technical characteristics tested against these needs, so "weight" should be prioritized in the further design work.

Common mistakes

• Writing a specification that is really a solution ("use aluminum") instead of a target ("mass < 1.2 kg").
• Jumping straight to solutions before the needs are understood and prioritized.
• Assuming the customer's first suggested solution IS the need – ask "why" to find the underlying need.
• Forgetting that the relationship strength in QFD should reflect how much a technical characteristic actually affects the need, not just whether it does.

Concepts in this part

Practise process and customer needs in the app →

2. Concept development and selection

What is it about?

Once the customer needs are understood, the team must find and choose among several possible solutions. The goal is not to find "the perfect idea" right away, but to systematically generate many concepts and then narrow them down in several rounds. Experience consistently shows that teams who explore several concepts before deciding end up with better solutions than teams who lock onto the first idea.

Concepts and formulas

• Brainstorming: generate many ideas without criticizing along the way; evaluation comes later.
• Morphological matrix: splits the product into sub-functions, lists solution alternatives for each, and combines them into concepts. The number of possible combinations (regardless of whether they are actually feasible) is the product of the number of alternatives for each sub-function: $N = n_1\cdot n_2\cdots n_k$.
• Concept screening (Pugh matrix): each concept is rated criterion by criterion with +, 0 or − against a reference concept. Net score = number of pluses − number of minuses (zeros don't count). Used to quickly weed out the weakest concepts.
• Concept scoring: a finer-grained evaluation of the remaining concepts, where each criterion gets a weight $w_i$ (often summing to 1, or given in percent) and each concept a score $s_i$ on each criterion. Weighted total score: $S = \sum_i w_i s_i$.
• Prototype: an early, simplified version of the product. A "works-like" prototype tests function, a "looks-like" prototype tests appearance and ergonomics – rarely both in the same prototype early in the process.

How to solve the problems

1. For the number of concepts from a morphological matrix: multiply the number of alternatives for each sub-function.
2. For the Pugh net score: count the pluses and minuses, subtract. Zeros don't affect the result.
3. For weighted concept scoring: multiply each score by its weight, and sum. Check that the weights actually represent what the team considers most important.
4. Remember that a high score on one criterion cannot "buy off" a concept that fails an absolute requirement (e.g. safety) – such requirements should be checked as a separate, absolute filter before weighting.

Example

A concept is rated on three criteria with weights 0.5, 0.3 and 0.2, and receives the scores 4, 2 and 5. $S = 0.5\cdot 4 + 0.3\cdot 2 + 0.2\cdot 5 = 2.0 + 0.6 + 1.0 = 3.6$.

Answer: weighted total score 3.6.

Common mistakes

• Choosing the first concept that comes up, without comparing it to alternatives.
• Criticizing ideas during the brainstorming itself – it stifles creativity and produces fewer ideas.
• Blindly trusting a high weighted score even though the concept fails a critical requirement (e.g. safety) that isn't included in the weighting.
• Confusing concept screening (rough, +/0/−) with concept scoring (finer, numeric points) – they serve different purposes.

Concepts in this part

Practise concept development and selection in the app →

3. Design for X and production

What is it about?

A concept also has to be manufacturable, cheaply and reliably, in the quantity you actually need. "Design for X" (DFX) is an umbrella term for designing the product with a specific goal in mind from the start – for example easy assembly (DFA), cheap production (DFM), or easy disassembly and recycling. The choice of production method, tolerances and materials has major cost consequences, especially as volume increases.

Concepts and formulas

• DFA (Design for Assembly): fewer parts, simpler and faster assembly. Ask for every part: does it need to be separate?
• DFM (Design for Manufacturing): designing parts so they are easy and cheap to produce – uniform wall thickness, standardized hole sizes, correct draft angle for molding.
• Fixed cost vs. unit price: prototyping (3D printing) has low fixed cost but high unit price; injection molding has high fixed cost (the mold) but low unit price. The break-even point: $Q^* = \dfrac{FC}{c_p - c_m}$, where $FC$ is the mold cost, $c_p$ the unit price of the expensive method and $c_m$ the unit price of the cheap one.
• Material cost: mass $= V\cdot\rho$, cost $= \dfrac{\text{mass}}{1000}\cdot\text{price per kg}$. Consider adding a waste factor for material lost to chips, sprues or supports.
• Tolerance stack-up (worst case): when several dimensions with individual tolerances $\pm t_i$ lie in a straight chain, the total worst-case tolerance is the sum $T = \sum_i t_i$. A more realistic statistical method (RSS – root sum square) gives $T_{RSS} = \sqrt{\sum_i t_i^2}$, which is normally smaller (more optimistic) because it is unlikely that all deviations are at their maximum at the same time.
• Standardization: using identical parts (e.g. the same screw type) across products reduces inventory, purchasing and assembly complexity.

How to solve the problems

1. Identify whether the question concerns assembly (DFA), manufacturing (DFM), material cost or tolerances.
2. For method choice: set up total cost as a function of quantity for each method, and find where the curves cross.
3. For material cost: compute the mass from volume and density first, convert to kg, and multiply by the price. Add any waste factor.
4. For a tolerance chain: use the sum of tolerances for the worst case, or the root of the sum of squares (RSS) if the problem asks for a statistical assessment.

Example

A linear chain of three dimensions has tolerances ±0.05 mm, ±0.08 mm and ±0.12 mm. What is the total tolerance in the worst case, and with the RSS method?

1. Worst case: $T = 0.05 + 0.08 + 0.12 = 0.25$ mm.
2. RSS: $T_{RSS} = \sqrt{0.05^2 + 0.08^2 + 0.12^2} = \sqrt{0.0233} \approx 0.153$ mm.

Answer: 0.25 mm in the worst case, but only about 0.15 mm with the statistical RSS method.

Common mistakes

• Setting unnecessarily tight tolerances "just to be safe" – this increases cost without any functional benefit.
• Using the worst-case sum when the problem actually asks for a statistical (RSS) tolerance stack-up, or vice versa.
• Forgetting the waste factor in material cost when the production method produces a lot of chips or scrap.
• Choosing a production method based on unit price alone, without considering the fixed cost (mold cost) and expected volume.

Concepts in this part

Practise design for X and production in the app →