NOTE: This is Part 2 of a two-part op-ed series by Dr. Andrew Maxwell, the Bergeron Chair in Technology Entrepreneurship in the Lassonde School of Engineering at York University, examining why diffusion, decision quality and governance determine the economic returns from innovation. Part 1, which looked at why innovation funding alone rarely produces economic impact, was published on February 18.

Innovation policy discussions often assume that once a technology has been successfully adopted by early users, diffusion will naturally follow and economic impact will accumulate.

In practice, diffusion is neither automatic nor economically uniform. Some innovations spread widely yet leave little trace on productivity or regional development, while others trigger cascading effects that reshape entire sectors.

The aim of this article is to extend prevailing models of innovation-driven growth by explicitly incorporating adoption, diffusion, and use – and by showing why new evaluation and decision processes are required at each stage if innovation is to translate reliably from insight into economic value.

Understanding this difference requires moving beyond the question of whether adoption occurs, to a more consequential one: what kind of diffusion occurs, and what it enables.

Adoption describes a discrete decision by an individual organization to use a new technology. Diffusion refers to the spread of that use across organizations, sectors and regions over time. The distinction matters because the barriers to diffusion are not simply scaled-up versions of adoption barriers.

Even when early use is successful, diffusion can stall due to organizational inertia, lack of complementary capabilities, or the difficulty of coordinating change across multiple actors. As a result, innovations may be technically proven and commercially viable, yet remain confined to a small set of users.

Diffusion is often treated as a binary outcome: an innovation either spreads or it does not. This framing obscures an important reality. The economic consequences of diffusion depend critically on how innovations are used and what they enable.

Some forms of adoption primarily support consumption or incremental convenience. These innovations may diffuse rapidly, but they generate limited spillovers beyond the immediate user. Their impact on productivity, organizational capability, or economic structure is often modest.

Other innovations function as enabling technologies. Their adoption reorganizes production processes, coordination mechanisms, or information flows in ways that create complementarities across users.

In these cases, each additional adopter increases the value of adoption for others, accelerating learning, secondary innovation, and further diffusion. Network effects, shared standards and capability-building dynamics amplify impact over time.

From an economic perspective, the difference is not the speed of diffusion alone, but the presence or absence of multiplier effects that link adoption to broader system transformation.

 Why diffusion outcomes are uneven

Empirical evidence consistently shows large productivity differences among firms operating in the same industries, even when they have access to similar technologies. A small group of frontier firms captures the majority of gains, while others struggle to follow.

These gaps are rarely explained by access alone. They reflect differences in managerial capability, workforce skills, organizational practices and the ability to integrate new technologies into existing operations. For many organizations, adoption requires simultaneous changes to routines, roles and incentives – changes that are costly, uncertain and difficult to coordinate.

When these complementary conditions are unevenly distributed, diffusion remains partial and impact remains concentrated.

Diffusion outcomes are shaped not only by the characteristics of innovations, but also by the tools and criteria used to select, evaluate and prioritize them. When decision frameworks focus narrowly on immediate performance, localized efficiency or short-term feasibility, they may favor innovations that diffuse easily but generate limited system-level impact.

By contrast, selection approaches that consider complementarities, learning spillovers and network dynamics are more likely to advance innovations whose diffusion reshapes broader economic or social systems. These tools influence not only which innovations advance, but how diffusion is expected to create value.

Yet many innovation systems lack practical instruments for making this distinction. Diffusion is encouraged, but rarely differentiated by its likely economic consequences.

If innovations differ in their multiplier effects, then diffusion support cannot be treated as a generic objective. Innovation systems need ways to identify and prioritize high-impact adoption pathways, not simply higher adoption rates.

At present, the absence of such instruments means that systems often optimize for what is most visible – uptake, deployment or pilot proliferation – rather than for forms of use that enable learning, coordination and capability-building at scale. Adoption becomes a metric of activity rather than a driver of transformation.

This is not merely a technical gap; it is a design problem in how innovation systems make decisions.

Why decision quality determines long-run impact

Innovation systems are often described as pipelines, ecosystems, or portfolios of activity. These metaphors capture important features, but they overlook a central reality: innovation systems are decision systems.

At every stage – from research selection to project continuation, from early adoption to scaling – decisions determine which ideas advance, which stall, and how innovations are ultimately used. Over time, the cumulative effect of these decisions shapes economic outcomes far more than the volume of innovation activity alone.

Policy often emphasizes expanding inputs: funding levels, numbers of projects supported, or volumes of outputs produced. These measures are easy to track and politically attractive, but they reveal little about whether innovative potential is being converted into productive use.

Without disciplined decision-making, innovation systems tend to default to continuation rather than selection. Projects persist because they are promising, because effort has already been invested, or because stopping them is institutionally uncomfortable. The result is a system that grows larger, but not necessarily more effective.

A critical but underappreciated aspect of decision quality lies in the tools and criteria used to choose between options at each stage of the innovation process. These tools implicitly define what kinds of outcomes are valued and which pathways are prioritized.

When evaluation frameworks emphasize short-term performance, localized efficiency or easily observable outputs, they may systematically favor innovations with limited spillovers. Adoption becomes a success metric in itself, regardless of whether it enables broader learning, coordination, or capability development.

Conversely, tools that recognize multiplier effects, complementarities and secondary adoption dynamics are better suited to identifying innovations whose use reshapes systems rather than merely improving isolated outcomes.

The same challenge applies to decisions about which innovations should continue, be redesigned, or be stopped. Many innovation systems lack disciplined criteria for making these choices based on evidence of real-world use and impact.

This is partly a measurement problem, but it is also a governance problem. Resources are limited, decisions are distributed across institutions, and incentives often reward continuation over selection.

Without clear accountability and feedback loops, learning remains fragmented and repeated inefficiencies persist.

A quality improvement perspective on innovation systems

Viewing innovation systems through a quality improvement lens shifts attention from expanding activity to improving yield. In quality-oriented systems, performance improves by identifying where value is lost, understanding why, and refining processes accordingly.

Applied to innovation, this perspective reframes failure as information rather than embarrassment. The objective is not to eliminate uncertainty, but to ensure that decisions evolve as understanding improves – particularly at key transition points such as adoption and diffusion.

Improving decision quality raises unavoidable questions of responsibility and authority. Who is empowered to make continuation or termination decisions? How are trade-offs between short-term performance and long-term system impact resolved? And how are decision-makers supported with instruments that reflect the true sources of economic value?

These are governance questions as much as analytical ones. Addressing them requires aligning incentives, accountability and decision tools with the realities of how innovation creates impact.

Innovation remains essential to long-run prosperity. But innovation alone is not enough.

The economic returns from innovation depend on how ideas are selected, how they are used, and how learning is incorporated over time. Strengthening adoption, diffusion and decision quality does not require abandoning research investment or narrowing ambition. It requires aligning processes, tools, and governance with the realities of how innovation creates value.

Innovation systems succeed not because they generate ideas, but because they learn how to turn potential into practice.

R$