Hands-on
·
8 min read
The value of design is in everyone else’s budget.
Engineers read cost, product and design read value, and both are right on their own lines. One page per decision turns the argument into a trade. With cases from hardware and software.

In short
Design is booked as a cost, while its value lands in price, unit cost, tooling, logistics, time, service and the next product. Put one decision and every line it touches on a single page, and the argument between cost and value becomes a trade.
01 The bill
Cost thinking and value thinking read different lines of the same product.
02 The trade
A saving on one line pays for value on another. A pallet can pay for a light.
03 The evidence
Cause is shown for single decisions. The broad claims rest on correlation.
Design looks like a cost because that is how it is booked. Its value lands elsewhere: in price and volume, unit cost, tooling, logistics, time to market, warranty and service, and in the next product built on the same parts. Those lines belong to different people, which is why the same decision can look wasteful in one meeting and obvious in the next. The way out is modest. Take one design decision, put every line it touches on one page, and decide on the sum.
Two ways to read the same product
Most hardware teams know the argument. Engineers, project leads and purchasing think in cost: parts, tooling, assembly time. Product management, sales, strategy and design think in value: what a buyer notices, what the product can charge, what it says about the company. Both are doing their job. Each reads the lines it is responsible for (Exhibit 1).
Exhibit 1
On the left, how a design decision appears in the accounts, and what design decides. On the right, the seven lines where its effect shows up: the design levers that move each one, how it is measured and who owns the number.
Where a design decision is booked, and where it shows up
Booked as
One cost centre
Design hours, a dearer part, a tool insert. The ledger shows what was spent.
What design decides
- Form and proportion
- Material
- Colour and finish
- Details and gaps
- Touchpoints: buttons, light
Price and volume
What buyers pay, and how many buy
Design leversForm, colour and finish, touchpoints
Measured byWillingness to pay, win rate
Owned byProduct management, sales
Unit cost
Parts, materials and assembly per unit
Design leversMaterial, part count, details
Measured byBill of materials, assembly minutes
Owned byEngineering, purchasing
Tooling
Moulds, inserts, textures: a one-off cost
Design leversForm, undercuts, surface finish
Measured byTooling quote, tool life in shots
Owned byEngineering, manufacturing
Logistics
Freight, packaging, storage (and CO₂)
Design leversSize and footprint
Measured byUnits per pallet
Owned bySupply chain
Time to market
Launch date, late changes
Design leversWhen the form is frozen
Measured byCost of delay per month
Owned byProgramme manager
Warranty and service
Returns, repairs, service life (and EU repair rules)
Design leversMaterial, details, access for repair
Measured byReturn rate, warranty cost per unit
Owned byQuality, service
Platform and reuse
What the next product inherits
Design leversShared parts, rules, design language
Measured byShare of reused parts
Owned byProduct line management
What it means. The people who see the cost and the people who see the value read different lines. Put all of them on one page before the meeting, and the discussion is about the sum.
Sources and method
My own model. It extends the four classic trade-offs of product development economics (development expense, unit cost, product performance and time to market), as in Smith and Reinertsen, Developing Products in Half the Time (1991). Tooling cost drivers such as undercuts, side actions and surface finish follow common injection moulding guidance. Colour, material and finish as a design discipline: Del Curto and others, Politecnico di Milano. Platform and reuse after Baldwin and Clark, Design Rules (2000). EU repair rules: Ecodesign for Sustainable Products Regulation (EU) 2024/1781. Cost of delay after Reinertsen, The Principles of Product Development Flow (2009).
There is a second reason the argument is hard to settle. In 1987 Peter Gorb and Angela Dumas described what they called silent design: a large part of what a product becomes is decided by people who are not called designers. Where the screws sit, how thick a wall is, how a carton is sized. These are design decisions, taken in other budgets.
Consultancies have a name for weighing both sides. McKinsey calls it design to value and makes a point that matters here: unlike design to cost, it can end with a product that costs more to build, when buyers pay for the difference. Its roots go back to value analysis at General Electric in the 1940s, where Lawrence Miles defined value as function divided by cost. What follows is the small version, with no programme and a single decision.
Not every line is as easy to count. Unit cost and tooling come with a quote. Perceived quality, brand recognition and the fit with what customers need show up later, often in other products, and can only be surveyed or estimated. Marketing has the textbook tools for that half. David Aaker splits brand equity into awareness, perceived quality, associations and loyalty, and Jenni Romaniuk measures a brand’s distinctive assets by how many buyers link them to the brand, and to no other (Exhibit 2).
Exhibit 2
Design effects sorted by two questions. Does the effect show in this product’s profit and loss, or later and elsewhere? Can it be counted, or only surveyed and estimated?
Every design effect, by two questions
DirectShows in this product’s profit and loss
IndirectShows later, or in other products
Easy to measureCounted
Direct · easy to measure
Unit cost
Bill of materials
Tooling
Tooling quote
Logistics
Units per pallet
Warranty
Return rate
Indirect · easy to measure
Platform reuse
Share of reused parts
Ease of use
Task success, support tickets
Repeat purchase
Share of buyers who buy again
Hard to measureSurveyed or estimated
Direct · hard to measure
Price premium
Willingness-to-pay tests, conjoint
Time to market
Cost of delay, an estimate
Win rate
Deals won against a named competitor
Indirect · hard to measure
Perceived quality
Ranking tests with buyers
Brand recognition
Fame and uniqueness of an asset
Brand value
Awareness, associations, loyalty
Fit with customer needs
Kano survey, interviews, personas
Fit with the trend
Panels, sales over time
What it means. Put the top row on the one-page bill. Track the bottom row over years, with the same survey each time, and keep a share of the budget for it. A bill that leaves out the bottom row will always favour the cheaper design.
Sources and method
My own model. Brand equity as awareness, perceived quality, associations and loyalty: Aaker, Managing Brand Equity (1991) and Measuring Brand Equity Across Products and Markets, California Management Review (1996). Fame and uniqueness of distinctive assets: Romaniuk, Building Distinctive Brand Assets (2018). Perceived quality ranking: Stylidis and others, Chalmers University of Technology (2020). Needs as must-be and attractive quality: Kano and others (1984). Design in the balanced scorecard: Borja de Mozota, Design Management Review (2006).
A board, a pallet and a bill
On one hardware project the question was whether the circuit board should stay one large board or be split into two levels. One board was cheaper to make and to assemble. Two levels cost more, and they changed the housing: deeper, with a clearly smaller footprint. The cost view said one board. I cannot share the figures, so the exhibits here use invented values to show direction and proportion.
The footprint turned out to be the interesting number. I wrote a small sorting script in Rhino that takes the outer dimensions of a product and returns how many units fit on a Euro pallet. With the smaller footprint the count rose substantially. Sourcing and logistics then priced what that meant for freight, packaging and CO₂ per unit (Exhibit 3).
Exhibit 3
One decision: split the circuit board into two levels, so the housing gets a smaller footprint. Each row is one line of the bill. Bars to the left cost more per unit, bars to the right save. Lengths are relative and show no amounts.
One decision, every line it touches Illustrative
Electronics
Second board, connector, assembly
costs more
Housing
Deeper shell, a little more material
costs more
Freight
More units on every pallet
saves
Packaging
Smaller carton
saves
Storage
Less space per unit
saves
Sum per unit
All five lines together
net saving
CO₂ per unit
Follows freight, counted separately
lower
What it means. Seen from the electronics budget alone, the split loses. Across all lines it wins, and the surplus is a budget that can go into something the buyer notices.
Sources and method
Illustrative. The decision comes from a hardware project I worked on. Its amounts are confidential, so the bar lengths are invented to show direction and proportion, not measured data. CO₂ follows freight and is counted separately, for example with ISO 14083.
The saving in transport was larger than the extra cost of the second board. The decision that lost on the electronics line won on the bill, and it left a surplus. Exhibit 4 shows the mechanism with example cartons: a wide, flat box wastes part of the pallet, and a smaller, deeper one fills it.
Exhibit 4
A Euro pallet seen from above and from its right-hand side, the 800 mm edge. Both views are drawn to the same scale and loaded to the same height. Both cartons hold the same volume. The wide, flat one fits two by two and leaves strips of the deck unused. The narrower, taller one fits three by three.
One Euro pallet, two carton sizes Illustrative
Before
Wide and flat
Carton 420 × 360 mm, 120 mm high
From above2 by 2 per layer
From the right2 stacks, 15 layers
60 units per pallet
4 per layer, 15 layers of 120 mm
After
Narrow and tall
Carton 380 × 260 mm, 180 mm high
From above3 by 3 per layer
From the right3 stacks, 10 layers
90 units per pallet +50%
9 per layer, 10 layers of 180 mm
What it means. Units per pallet is a number design can move. It belongs in the brief as a target next to cost and size, and once it is there it works for every product in the range.
Sources and method
Illustrative carton sizes, not those of the project. Euro pallet of 1,200 by 800 mm and 144 mm high, loaded to 1,800 mm, which keeps the unit within the 1,950 mm upper loading height used in German retail logistics (CCG II, GS1 Germany). Simple grid loading without mixed orientations; weight limits are not considered.
Two things came out of it beyond the one product. Units per pallet became a target in the brief, a number that had not been there before. And because it is a rule about footprint and packaging, it applies to the next product in the range as well. This is where the single page meets the portfolio view the consultancies describe.
Spend the surplus where people notice
A surplus is only worth something if it goes to the right place. Research on perceived quality, much of it from Chalmers University of Technology with Volvo Cars, breaks the impression of quality into separate attributes and ranks them against effort and cost. The idea carries over to any product: list the elements that signal quality, estimate what each one adds and what it costs, and sort (Exhibit 5).
Exhibit 5
Six places where a product can signal quality, sorted by effect per unit of added cost. For each, the first bar is the effect on how the product is judged and the second is what it adds to the cost of one unit.
Six elements, weighed against their cost Illustrative
Alignment Start here
Even gaps, screws centred and mirrored
Light
A status light that reads as one piece
Control
The feel of the one button people press
Surface
Texture and material where the hand rests
Hinge
A cover that opens without play
Form
A shape that needs a more complex tool
What it means. Alignment comes first because it costs close to nothing when engineers know about it early. The budget found elsewhere then goes into one element people see or touch every day, and user research decides which.
Sources and method
Illustrative weights, not measured data. A method for ranking perceived quality attributes against effort, time and cost: Stylidis, Wickman and Söderberg, Chalmers University of Technology and Volvo Cars (2020).
Some of the strongest signals are nearly free. Screws that are centred, mirrored and evenly spaced make a housing look considered, and they cost nothing extra if the engineers know early that it matters. A short talk in each department about what the design is trying to do has paid back many times in my experience. Other signals cost real money: a light, the feel of a control, a surface. On one product the choice fell on the status light, backed by user research and by the company’s aim to be known for quality, and the number of LEDs was doubled. The pallet paid for the light.
For designers, and for external ones most of all, this changes the job. Ask how the costs are built, who decides and where there is room. When the bill is open, the cost side can say: this is the budget, put it where it does most. Product management then tests the choice with users.
Protect the detail
Value also has to last through the production run. Fine details in injection-moulded parts, such as a sharp chamfer or a surface texture, wear in the tool over time. How fast depends on the steel, the plastic and its fillers. The usual answers are to rework the cavity or to accept a softer edge. We made the fine element a replaceable insert instead. It is a small part of the tool, and each swap returns the moulded part to the designed state (Exhibit 6).
Exhibit 6
Sharpness of a fine moulded detail over the life of an injection mould. Without an insert the detail wears until it falls below the acceptance limit. With a replaceable insert it returns to the designed state at each swap.
One fine detail over the life of a mould Illustrative
Sharpness of the detail
First shotEnd of the run
What it means. The insert is a small, early expense on the tooling line. It buys constant quality over the run and avoids a stop for rework later, which is where the larger cost sits.
Sources and method
Schematic, not measured data. How fast a mould wears depends on the steel, its coating, the plastic and its fillers. Glass fibre in particular abrades cavities: comparative study of mould coatings in Coatings (2017).
On the bill this is an entry under risk and time: a small early expense against a later stop in production, and against a slow loss of quality that nobody signs off and everybody ships.
Software reads the same bill, later
In software the lines are the same and the timing differs. Hardware freezes a decision with the tool, so the page has to be written before. Software can measure afterwards. Ron Kohavi and Stefan Thomke describe a change to how Bing displayed ad headlines that raised revenue by 12 per cent, found through a controlled experiment. A study by Deloitte for Google across 37 brands linked a gain of 0.1 seconds in mobile load time to 8.4 per cent more retail conversions. Investment in architecture, in a fast front end or in careful motion can be argued the same way: name the line, then test.
What the evidence supports
The well-known numbers for the value of design are weaker than their reputation (Exhibit 7). McKinsey’s 2018 study of 300 companies found that those with the highest design scores grew revenue and shareholder returns far faster than their industries, and it says clearly that this is correlation. The Design Value Index chose its companies with hindsight. The often-quoted rule that 70 or 80 per cent of cost is fixed in design goes back to an informal survey, as Karl Ulrich and Scott Pearson found when they traced it in 1993.
Exhibit 7
Seven kinds of evidence that design pays, from strongest to weakest. The dots are my reading of how well each one separates cause from coincidence.
Evidence that design pays, by strength
Controlled experiment
Kohavi and Thomke (2017): one change to ad headlines at Bing, revenue up 12%
ShowsCause, for one change
Watch outSoftware only, short-term measures
Peer-reviewed firm studies
Hertenstein, Platt and Veryzer (2005); Gemser and Leenders (2001)
ShowsEffective design goes with higher returns, depending on industry and strategy
Watch outSmall samples, few industries
Large correlation study
McKinsey (2018): 300 listed companies over five years
ShowsHigh design scores go with faster growth
Watch outCorrelation; the method is not fully published
Stock index
Design Value Index (2015): 16 companies, 211% above the S&P 500 over ten years
ShowsA direction
Watch outCompanies chosen with hindsight; no longer updated
Commissioned study
Forrester for IBM (2018): a return of 301% over three years
ShowsA way to calculate
Watch outPaid for by the vendor; four clients interviewed
Survey
Oxford Economics for DesignSingapore (2025): 270 organisations, over 90% report positive outcomes
ShowsHow users of design see it
Watch outSelf-report
Folklore
“70% of cost is fixed in design”
ShowsNothing
Watch outTraced to an informal survey by Ulrich and Pearson (1993)
What it means. The broad claims rest on correlation and self-report. Cause has only been shown for single changes. That is a good reason to price one decision at a time.
Sources and method
Figures as published by the sources named in each row; links are in the list at the end of the article. The rating in dots is my own judgement, not a measurement.
I read this as support for the small version. Cause can be shown for single decisions, so that is where the case for design is strongest.
Limits
Not everything fits on the page. Brand effects build slowly and across products, and a company that only funds what can be priced in advance will stop making the things that later define it. Keep a share of the budget for work without a bill.
Design also costs. Marina Candi found that firms which use design openly attracted more new customers and were less profitable than firms where design stayed silent. Ulrich and Pearson saw no clear link between the quality of industrial design and manufacturing cost, in either direction.
The page itself can be gamed, because whoever writes it chooses the lines. It should be written together, with purchasing and logistics in the room.
What would prove me wrong
If teams that write such a page make the same decisions as teams that do not, it is paperwork. The same holds if the savings found on one line are routinely taken as margin and never reach the product.
A note on the numbers: figures from projects are withheld for confidentiality. Exhibits 3 to 6 use invented values and say so.
Sources (25)
- Weighing value and cost
- McKinsey, Capturing the full potential of design to valueMcKinsey, Design to value at scaleLawrence D. Miles and value analysisStylidis, Wickman and Söderberg, Perceived quality of products (2020)
- Silent design
- Shams and Lam on Gorb and Dumas, Silent design (1987)Candi, The sound of silence, Design Studies
- Cost folklore
- Ulrich and Pearson, Does product design really determine 80% of manufacturing cost? (1993)Barton, Love and Taylor, Design determines 70% of cost? (2001)
- Brand and hard-to-measure value
- Aaker, Measuring brand equity across products and markets, California Management Review (1996)Romaniuk, Building Distinctive Brand Assets (2018)Borja de Mozota, The four powers of design, Design Management Review (2006)Kano model
- Design and performance
- McKinsey, The business value of design (2018)Fortune on the Design Value Index (2017)Hertenstein, Platt and Veryzer (2005), summaryGemser and Leenders (2001)Forrester for IBM, Total Economic Impact (2018)Oxford Economics for DesignSingapore (2025)Yale Paprika, The Value of Design (2019)
- Logistics and tooling
- Design for logisticsDowlatshahi, The role of logistics in concurrent engineering (1996)Mould wear with glass fibre, Coatings (2017)Rhino
- Software
- Kohavi and Thomke, The Surprising Power of Online Experiments, HBR (2017)Milliseconds make millions, Deloitte for Google (2020)

