Technology investments are usually justified by what a new capability makes possible. This report examines what the organization must do to support that capability, and why the full consequences of adding more belong in every business case.
Download PDFOrganizations evaluate technology by its cost, anticipated efficiencies, and expected returns, but some of the most consequential costs arrive after approval: maintaining integrations, validating information, resolving exceptions, training employees, and managing interdependent processes.
Individual initiatives are often evaluated more thoroughly than their cumulative effects. Automation frequently changes where work occurs rather than eliminating it, and undocumented processes can turn a straightforward project into an exercise in organizational discovery.
Drawing on research into the productivity paradox and the NIST AI Risk Management Framework, this report offers five considerations for evaluating the full consequences of a technology decision.
Over the course of my career in technology consulting and enterprise business development, I have participated in countless discussions about the benefits of new technology. These conversations typically begin with a business problem, followed by an examination of the capabilities available to address it. Eventually, the discussion turns to cost, anticipated efficiencies, implementation requirements, and expected returns.
There is nothing inherently wrong with that process. Organizations have finite resources, and investments require financial justification. Still, I have often questioned whether we give sufficient attention to what happens after an investment is approved and the technology becomes part of everyday operations. Some of the most consequential costs are not necessarily associated with acquiring or implementing a system. They are the responsibilities that remain long afterward: maintaining integrations, validating information, resolving exceptions, training employees, and managing increasingly interdependent processes.
These responsibilities accumulate as organizations introduce new capabilities without always reconsidering what is already in place. Over time, a business may become more technologically capable without becoming proportionately more effective. The value of an investment depends not only on what the technology makes possible, but also on what the organization must do to support it.
Contact centers offer a useful illustration, particularly given the rapid adoption of generative AI. I spent a significant portion of my career working with organizations evaluating and implementing customer experience technologies. Traditional telephony and workforce management systems evolved into sophisticated analytics, cloud contact centers, and conversational AI platforms, each addressing genuine operational challenges.
Today, automated interaction summaries can reduce administrative effort, knowledge systems can help employees retrieve information, and AI-enabled service platforms can handle routine requests that previously required human intervention. These advances offer meaningful opportunities to improve service quality and productivity. However, experience has taught me to distinguish between eliminating work and changing where that work occurs.
“Experience has taught me to distinguish between eliminating work and changing where that work occurs.”
An AI-generated summary may reduce documentation time, but its accuracy still matters. A knowledge assistant may accelerate access to information, but someone remains responsible for maintaining that information. Automated interactions may reduce routine service volumes while leaving employees to manage increasingly complex situations. Some work disappears, some changes, and some shifts to other people or functions.
These considerations do not necessarily diminish the value of automation. Additional oversight may be justified, particularly when accuracy, compliance, or customer outcomes are consequential. The issue is whether organizations adequately account for these responsibilities when evaluating anticipated benefits.
Consider a technology that eliminates ten minutes of administrative work compared with one that eliminates ten minutes but introduces several minutes of verification, correction, or coordination. Both may perform their intended function, yet their contributions to overall productivity differ. This distinction becomes increasingly important as businesses incorporate AI into operational decisions.
One recurring challenge in technology investment decisions is that individual initiatives are often evaluated more thoroughly than their cumulative organizational effects. A new analytics platform may improve reporting, another system may strengthen customer engagement, and a workflow application may automate a manual process. Each investment can present a credible business case independently, while their combined effect becomes considerably more complicated.
This is compounded by an issue I have encountered repeatedly in conversations with client organizations: the condition of their existing process documentation. In some businesses, documentation has become outdated; in others, procedures have evolved without being formally recorded. Employees may understand their individual responsibilities, but the organization lacks a reliable, end-to-end view of how work moves across business units, where decisions occur, and how different activities and systems depend on one another.
In these environments, much of the organization’s operational knowledge resides with experienced employees rather than in accessible, maintained documentation. People know which exceptions require special attention, which procedures no longer reflect actual practice, and which informal workarounds keep daily operations functioning. That knowledge can be remarkably valuable, but it also introduces organizational vulnerability when employees leave, responsibilities change, or systems are replaced.
The consequences become particularly apparent during technology modernization and automation initiatives. Before a process can be reliably automated, its underlying logic, dependencies, exceptions, and decision points must be sufficiently understood. When that information is fragmented or undocumented, implementation teams may spend considerable time reconstructing how the business actually operates, often relying on employees whose institutional knowledge has effectively become the organization’s process documentation.
What initially appeared to be a straightforward technology initiative can become an exercise in organizational discovery. Importantly, this work is not always created by the new technology. Some deficiencies existed long before the investment was proposed but remained largely invisible until the organization attempted to change something. The technology exposes them, and addressing them becomes part of implementation.
This distinction matters when assessing the true cost of technological change. Recovering undocumented process knowledge may ultimately improve the organization, but it requires time, resources, and management attention that may not have been anticipated in the original business case.
Similar problems arise when new systems are layered onto existing applications. Data must move between platforms, employees must learn different interfaces, and managers must reconcile competing information. Technical teams inherit maintenance responsibilities, while older systems remain operational because replacing them introduces additional risks. Organizations can gradually accumulate capabilities without adequately examining their total burden.
This goes beyond diminishing returns. It also concerns how costs and responsibilities are distributed, particularly when those consequences are not represented in the original financial justification. A department may realize measurable efficiencies while another absorbs additional work. A dashboard may improve visibility while increasing reporting and validation demands. The benefits may be genuine, but so may the burdens. Understanding both is necessary before concluding that the organization has become better off.
Research provides useful context for understanding why technological capability and organizational value do not always advance together.
Erik Brynjolfsson, Daniel Rock, and Chad Syverson (2017) examined the modern productivity paradox and the difficulty of observing productivity improvements from transformative technologies. Their work suggests that realizing economic benefits often requires complementary investments, organizational changes, and time. Acquiring a capability and realizing its full value are not necessarily simultaneous events.
This is particularly relevant to AI. Introducing powerful technology does not automatically produce comparable improvements in organizational performance. Existing processes, employee capabilities, governance structures, and operating models influence how effectively technology is used and what value it delivers.
The National Institute of Standards and Technology’s (NIST, 2023) AI Risk Management Framework emphasizes governance, monitoring, accountability, and ongoing risk management. These activities are essential to responsible AI adoption, but they also require organizational resources. Some additional work represents avoidable inefficiency, while other responsibilities are necessary to ensure reliability, safety, and appropriate oversight.
The objective should not be to eliminate every new obligation associated with technology, but to determine whether those obligations are justified by the value they create or protect. More automation does not necessarily demonstrate better overall performance, just as additional oversight does not necessarily indicate inefficiency. Both require evaluation within the broader operating environment.
My experience has led me to consider whether organizations need a more comprehensive approach to evaluating technology investments. Traditional business cases remain necessary, but they are more useful when they account for organizational consequences rather than focusing predominantly on anticipated improvements within a particular function.
None of this suggests reluctance to invest in innovation. A more sophisticated or expensive solution may produce substantially greater value by simplifying other processes, reducing risk, or enabling new capabilities. The relevant consideration is whether the resulting benefits justify the full commitment.
I remain optimistic about AI and other emerging technologies. Their capabilities continue to develop, and organizations are still learning how to incorporate them into meaningful applications. At the same time, I believe we need to distinguish more carefully between technological advancement and organizational improvement.
A company can acquire sophisticated systems without improving its decisions, collect additional information without developing greater understanding, and automate activities while increasing the demands placed on employees. Conversely, a well-considered investment can reduce unnecessary effort, improve customer outcomes, and allow people to concentrate on work requiring their judgment and expertise.
Looking back across my career, I suspect some of the most consequential technology decisions were not simply about what an organization implemented. They were also about what it determined was no longer necessary, what responsibilities it was willing to accept, and whether the resulting organization worked better.
As AI becomes more deeply embedded in business operations, these judgments will become increasingly important. Technological capabilities will expand, but organizational resources, managerial attention, and human capacity remain finite. Recognizing those constraints does not mean resisting progress. It means being more deliberate about how progress is defined and evaluated.
The question is no longer simply what technology allows us to do. It is whether doing more leaves the organization, its employees, and its customers better off after the full consequences are considered. To me, that is a more meaningful measure of progress.
“Technological capabilities will expand, but organizational resources, managerial attention, and human capacity remain finite.”
Brynjolfsson, E., Rock, D., & Syverson, C. (2017). Artificial intelligence and the modern productivity paradox: A clash of expectations and statistics. National Bureau of Economic Research, Working Paper No. 24001. https://doi.org/10.3386/w24001
National Institute of Standards and Technology. (2023). Artificial intelligence risk management framework (AI RMF 1.0). https://doi.org/10.6028/NIST.AI.100-1