Making the financial case for design without overclaiming

Design is valuable, and the value can be measured. The difficulty is that the two claims most often made for it are not the claims the evidence supports. The measured lineage from satisfaction to profit is real, conditional and local. Other figures circulate instead. A design-led portfolio beat a stock index, and a quartile of companies grew faster than their peers.

This sixth of ten units, after the fifth made the assumptions binding, argues that presenting the second as the first hands an informed client the question that ends the conversation. A case that survives prices the problem in the client's own numbers and leaves the arithmetic checkable.

The measured lineage runs from satisfaction to returns, and it is conditional

Anderson, Fornell and Lehmann (1994) tested whether improving customer satisfaction brings economic returns. They used a national customer satisfaction index and accounting measures such as return on investment. They found support for a positive effect of quality on satisfaction and of satisfaction on profitability. That is the strongest single result behind the commercial case. They also found that increasing market share can lower satisfaction. Anderson and Mittal (2000) answered the objection that satisfaction measurement is a trap. The chain is conceptually sound, but its links are asymmetric and nonlinear. Rust and Zahorik (1993) modelled satisfaction, retention and market share together. They found the returns to improvement depend on the firm's position and on which customers improve. Heskett and Sasser (2010) set out the service-profit chain that many business cases borrow. It runs from internal service quality and employee satisfaction, through productivity and customer value, to loyalty and profit. They present it as internally consistent logic rather than a tested causal model. The lineage is measured. Its measured form is a relation whose size depends on the situation.

The two figures everyone quotes choose their own winners

McKinsey and Company (2018) tracked the design practices of 300 publicly listed companies over five years. It built the McKinsey Design Index from the actions most correlated with financial performance. Top-quartile scorers increased revenue 32 percentage points and total returns to shareholders 56 percentage points faster than their industry counterparts. The report states that it found a strong correlation, and that an index of this kind cannot establish causation. It also records that over 40 per cent of the companies surveyed were still not talking to their end users during development. Design Management Institute and Rae (2015) publish the second: a market-capitalisation-weighted portfolio of sixteen design-led companies whose ten-year returns were 2.11 times the S&P 500, a 211 per cent outperformance to 2015.

Why an index cannot carry a causal claim

Both figures are built by selecting the companies first and reading their returns afterwards. A company that had already grown large could afford the design function that earns it a place in the portfolio, so the causality may run that way. Well-run firms may also be good at both for the same underlying reason. That is the confounding the reports acknowledge when they call the finding a correlation. The failure is not in the studies, which are candid about method, but in the quotation, where the word correlation is dropped between the report and the slide. Anderson and Mittal (2000) and Rust and Zahorik (1993) show why that matters, because both find the return to experience work depends on where a firm stands. Kimbell (2012) presses further. If design is a situated achievement distributed across people and objects rather than a method held by a team, attributing a firm's growth to a design function is itself the contested step.

The Design Management Institute's selection rules make the problem concrete. To enter the 2015 index a company had to show that design operated at scale across the business. It also had to show that a design executive sat at the top table, that investment was growing and that the chief executive backed it. The company also had to have been listed in the United States for ten years. Sixteen firms cleared that bar: Apple, Coca-Cola, Ford, Herman Miller, IBM, Intuit, Nike, Procter and Gamble, SAP, Starbucks, Starwood, Stanley Black and Decker, Steelcase, Target, Walt Disney and Whirlpool. A firm that failed, was taken over, or cut its design function during the window is absent by construction. The portfolio therefore cannot hold the counter-example a causal claim would need.

A famous metric that failed to replicate

Reichheld (2003) argued that one survey question, the likelihood to recommend, is the most reliable indicator of firm growth. He proposed the Net Promoter Score as the number management should track. Keiningham, Cooil, Andreassen and Aksoy (2007) attempted to replicate the claim. They used longitudinal data from 21 firms and more than 15,500 interviews from the Norwegian Customer Satisfaction Barometer. Using the industries Reichheld had named as exemplars, they failed to reproduce the asserted clear superiority of Net Promoter. A case built on someone else's aggregate can be overturned by the next study. A case built on the client's own data can only be overturned by the client's own data.

What a defensible case looks like

Bias (2005) established cost-justifying usability, setting out methods for estimating the cost of usability activity against returns from reduced support, rework and training. The argument is made in the organisation's own financial terms. Karat (1990) built the flagship figure in that tradition. Using data from three usability tests of each of two applications, she compared the projected dollar value of reduced end-user task time against the cost of the usability work. She reported a savings-to-cost ratio of about 2 to 1 for a small development project and 100 to 1 for a large one. The ratio differs fiftyfold across two projects in one study, which is the argument for estimating locally. Karat (2005) gives the form the local estimate should take. Rajanen (2007) adds that the cost-benefit models rest on assumptions rarely tested, so their ratios are illustrations rather than forecasts. Measure the cost of the problem in the client's own records, the support contacts, abandoned applications, rework and time on task, and count the users the interface excludes: an older operator, or a reader in a second language. Price the change in those units, and name the assumption each number rests on. Mijac, Jadric and Cukusic (2024) show that the terms have widened. They extend the DeLone and McLean success model with user experience and co-creation, and validate it with more than 550 end users.

Contested question: is the value causal or associational, and does the distinction matter?

The literature records two argued positions and a third that reframes the question. The first is that the quoted studies demonstrate association and no more, and the studies themselves hold it. McKinsey and Company (2018) calls its finding a correlation its index cannot turn into causation. Design Management Institute and Rae (2015) publish a hypothetical portfolio built by selecting companies on the practices it then credits. Anderson and Mittal (2000) and Rust and Zahorik (1993) support the same reading from inside the satisfaction literature.

The second is that the figures demonstrate cause, held in practice rather than argued, since the reports are cited as proof and the qualifier dropped. No paper here defends that reading.

Kimbell (2012) cuts against both, arguing that if design is distributed and situated, the unit of causation is not the thing an index scores.

What would settle the disagreement is a comparison in which design practice varies while strategy, market and firm quality are held constant. The ideal is a pre-registered two-arm test inside one organisation, or several pooled, since a portfolio index cannot supply a counterfactual. Keiningham and colleagues (2007) show the shape of the test that can defeat a widely adopted metric.

The association claim rests on the two index studies, which their own authors call correlations, and on the conditional findings of Anderson and Mittal and of Rust and Zahorik. The causal claim has no supporting paper here. While the dispute is open, a designer should price the problem in the client's own numbers and leave the arithmetic checkable. The strongest objection is that local measurement is easy to game, because a single instrumented decision can be chosen to show a saving and the client's own data then certifies a case that would not generalise.

Where the argument is heading

The direction of travel is from an aggregate ratio to a local measurement. Mijac, Jadric and Cukusic (2024) show experience quality being absorbed into an established success model rather than argued against. Karat's measurement method (1990) and business-case form (2005) describe how to build it, and Rajanen (2007) sets the limit on what it may claim. Kimbell (2012) points to the harder road, that if value is produced across an organisation the measurement that matters will be the organisation's. The skill the next few years will reward is not quoting a portfolio's return. It is instrumenting one decision in one organisation and showing the client how to check it.

The sources this unit rests on

• Anderson, E. W., Fornell, C. and Lehmann, D. R. (1994). Customer Satisfaction, Market Share, and Profitability: Findings from Sweden. Journal of Marketing. doi:10.1177/002224299405800304 Tested whether improving customer satisfaction brings economic returns, using a national customer satisfaction index and accounting measures such as return on investment. Found a positive effect of quality on satisfaction and of satisfaction on profitability, and that increasing market share can lower satisfaction, so the unit's strongest evidence for the lineage is itself conditional.
• Anderson, E. W. and Mittal, V. (2000). Strengthening the Satisfaction-Profit Chain. Journal of Service Research. doi:10.1177/109467050032001 Answers the charge that satisfaction measurement is a trap by defending the chain's soundness while refusing its arithmetic simplicity. Argues the links are asymmetric and nonlinear, so uniform improvement does not yield uniform returns, which is why the case must be local.
• Bias, R. G. (2005). Cost-Justifying Usability. Cost-Justifying Usability. doi:10.1016/b978-012095811-5/50022-5 A collection that establishes cost-justifying usability as a practice, setting out methods to estimate the cost of usability activity against returns from reduced support, rework and training. It is the origin of the discipline's business-case toolkit.
• Design Management Institute and Rae, J. (2015). 2015 dmi:Design Value Index Results and Commentary. Design Management Institute. https://www.dmi.org/page/2015DVIandOTW/2015-dmiDesign-Value-Index-Results-and-Commentary.htm Tracks a market-capitalisation-weighted portfolio of sixteen publicly traded design-led companies against the S&P 500. Reports ten-year returns 2.11 times the index, a 211 per cent outperformance over 2005 to 2015, from a hypothetical portfolio rather than an experiment.
• Heskett, J. L. and Sasser, W. E. (2010). The Service Profit Chain. Service Science: Research and Innovations in the Service Economy. doi:10.1007/978-1-4419-1628-0_3 Sets out the service-profit chain from internal service quality and employee satisfaction through productivity and customer value to satisfaction, loyalty and profit. Presents it as internally consistent logic rather than a tested causal model.
• Karat, C. M. (1990). Cost-Benefit Analysis of Usability Engineering Techniques. Proceedings of the Human Factors Society Annual Meeting. doi:10.1177/154193129003401203 Compared the projected value of reduced end-user task time with the cost of the usability work, using data from three usability tests of each of two applications. Reported a savings-to-cost ratio of about 2 to 1 for a small development project and 100 to 1 for a large one, a fiftyfold spread across two projects.
• Karat, C. M. (2005). A Business Case Approach to Usability Cost Justification for the Web. Cost-Justifying Usability. doi:10.1016/b978-012095811-5/50004-3 Sets out a business-case approach that translates reductions in user time and task error into financial terms and frames the result as a decision for managers rather than a technical claim. It is the canonical statement of the argument form the unit teaches.
• Keiningham, T. L., Cooil, B., Andreassen, T. W. and Aksoy, L. (2007). A Longitudinal Examination of Net Promoter and Firm Revenue Growth. Journal of Marketing. doi:10.1509/jmkg.71.3.039 Used longitudinal data from 21 firms and more than 15,500 interviews from the Norwegian Customer Satisfaction Barometer to replicate Net Promoter research. Using industries Reichheld named as exemplars, it failed to reproduce the asserted clear superiority of Net Promoter, the unit's strongest example of a famous metric failing replication.
• Kimbell, L. (2012). Rethinking Design Thinking: Part II. Design and Culture. doi:10.2752/175470812x13281948975413 Argues that design thinking is better understood through design-as-practice and designs-in-practice. Treats design as a situated achievement distributed across people and objects rather than a method held by one designer, which challenges attributing a firm's performance to a design function.
• McKinsey and Company (2018). The Business Value of Design. McKinsey and Company. https://www.mckinsey.com/capabilities/tech-and-ai/our-insights/the-business-value-of-design Tracked the design practices of 300 publicly listed companies over five years and built the McKinsey Design Index. Reported that top-quartile scorers grew revenue 32 percentage points and shareholder returns 56 percentage points faster than peers, that this is a correlation the index cannot turn into causation, and that over 40 per cent of the companies surveyed were not talking to end users during development.
• Mijac, T., Jadric, M. and Cukusic, M. (2024). The role of user experience and co-creation in measuring the success of digital services in higher education. Behaviour and Information Technology. doi:10.1080/0144929x.2023.2206924 Extended the DeLone and McLean information system success model with user experience and co-creation, and validated it with more than 550 end users of higher-education digital services. Found both to be significant success factors, evidence that experience quality now sits inside an accepted model.
• Rajanen, M. (2007). Usability Cost-Benefit Models: Different Approaches to Usability Cost Analysis. Proceedings of the Ninth International Conference on Enterprise Information Systems. doi:10.5220/0002407803320336 Compares approaches to usability cost analysis and argues that the models rest on assumptions that are rarely tested. Concludes that headline ratios should be treated as illustrations rather than forecasts.
• Reichheld, F. F. (2003). The One Number You Need to Grow. Harvard Business Review. https://hbr.org/2003/12/the-one-number-you-need-to-grow Argues that the likelihood to recommend is the most reliable indicator of firm growth, and proposes the Net Promoter Score as the metric management should track. It supplies the claim the replication tests.
• Rust, R. T. and Zahorik, A. J. (1993). Customer satisfaction, customer retention, and market share. Journal of Retailing. doi:10.1016/0022-4359(93)90003-2 Modelled satisfaction, retention and market share together. Found the returns to improving satisfaction are real but depend on the firm's position and on which customers improve, so no single figure transfers between firms.