Why Companies Don’t Adopt New Technologies
This image was created with the assistance of DALL·E 2
After recently attending the Canadian Science Policy Conference one comment made by Jean-Charles Fahmy (CEO of CENGN) stood out as to why companies don’t adopt new technologies:
Companies don’t have a strategy; they don’t have a sandbox to test and innovate new technologies; they don’t have the required skills; and they have insufficient understanding of the return on investment.
Whilst these are all important we also believe that there are other problems facing businesses.
Problem 1: Strategy and Culture
Many organizations are lacking a strategy of how, or even which, new technologies will be integrated into their business. Without this attempts are likely to be fragmented, have insufficient buy-in and may miss the most important opportunities the technology provides. Even with a strategy, without a culture to support bottom up experimentation with new technology the most effective ways to adopt the technology are likely to be missed.
Problem 2: Skill Gap
By definition there will be a limited pool of people expert in or even familiar with a new technology. The likelihood that a company already employs these experts (unless its primary business is to be on the cutting edge of tech) is low and hence there can be steep technical challenges to roll out a new technology. Providing training to acquire these skills is non-trivial and requires an organizational culture that values learning.
Problem 3: Uncertain Return on Investment
Understandably a business will be wary to invest in a technology without confidence in what the returns will be. However, especially due to a skill gap mentioned previously, this uncertainty may be more due to a lack of knowledge internal to the company than intrinsic to the newness of the technology. This also heightens the risk of significantly underestimating (or overestimating) the return to be expected, and potentially missing out on implementing the most productivity enhancing advances. Yet acquiring the expertise to properly evaluate a new technology is itself an investment that may have no returns.
Problem 4: Integration with Legacy Systems and Processes
Every organisation has a set of tools, systems and processes to currently perform its tasks. Except for the absolute most transformational technology many of these legacy techniques will still be in use and will need to interface with the new technology in some way. This can be immensely complex and solving these challenges are often far more difficult than adopting the new technology itself.
Problem 5: Inertia and Insufficient Trust
The set of tools, systems and processes currently used will also to a large degree be highly functional. If the new technology only appears to be an incremental improvement on what is already in use then there is likely to be significant resistance to changing to the unknown (with all the bureaucratic measures such as rewriting documentation that goes along with this). For more disruptive technology employees may be more hesitant to embrace this, at least until they have established trust that it works at least as well as the old systems. And for the most transformational employees may be wary of how it could impact their own job security if there is any level of distrust in management.
Problem 6: Regulation and Cybersecurity
Finally many companies work in highly regulated sectors, such as banking, where compliance already uses vast resources of the business. In these cases it may simply not be legal to adopt a new technology under current regulations. Even if it is, the issues of inertia and integration with legacy processes can be magnified many times as making a mistake simply isn’t an option.
Even for organizations without sector specific regulations, the need for information security can still pose a significant challenge as new technology may increase the vulnerabilities and surface area susceptible to cyber attack. Data sovereignty regulations can also pose a challenge — especially when working with small companies located in different countries. However it is often legacy systems that are the most vulnerable to attack and security is often enhanced by newer technology.
Recent Examples
To illustrate these problems we consider the following case studies: that of a new HR system at a university and software development at a financial services organisation.
HR Infrastructure at a University
The University’s rollout of a new integrated IT system for managing finances and HR serves as a compelling case study illustrating the challenges organizations face when adopting new technologies. Despite meticulous planning, a four-year lead time, and extensive testing, the $40 million initiative encountered significant post-implementation issues. Payroll errors surged, professors struggled with accessing their research grants, and promised benefits like an intuitive user interface and a useful chatbot failed to materialize even six months after the launch.
This case is a clear demonstration of problem #5: inertia and insufficient trust. The project’s emphasis on replacing existing systems without effectively communicating the need for the change to end-users resulted in skepticism and resistance. The large number of errors that occurred was a case of problem #2: skills gap, where an inadequate training programme and the absence of ongoing support exacerbated adoption challenges. Despite detailed planning and materials developed prior to the rollout this quickly was found to have been insufficient, and was exacerbated by a lack of clear response. Finally it shows problem #1: strategy and culture. Whilst significant effort was made to ensure there was a strategy this was often far too centralized, with an overconfidence of decision makers resulting in actively turning away the participation of end users who wished to be involved in testing.
In all this case is a warning of some of the issues that arise when initial testing and planning is confined to too small a group of administrators — especially for an organization as diverse and fragmented as a university. With a greater focus on the importance of aligning strategy with user buy-in, maintaining organizational agility during implementation, and recognizing that seemingly ‘safe’ technology replacements can carry significant risks, many of these issues could have been avoided.
A financial services organisation
A financial services organisation was developing software to automate the provision of services that were manually performed by a central service request team and various engineering teams. As the due date approached for the completion of the software, the question of adoption was raised: what is the plan for adopting these new capabilities? Who will provide training or communication to the users? When will adoption be complete?
With hindsight, it was clear that the software development teams were focused on code delivery and not user adoption. Project tracking and reporting had focused on code released into production and, only towards the end, considered adoption. Furthermore, the service owner had not been proactively considering adoption plans until the code was ready to go live. Clearly, new software with limited user adoption provides a poor return on investment. This example illustrates challenges #1: Strategy and Culture and #5: Inertia and Insufficient Trust.
Artificial Intelligence
Right now AI is the technology that is rightly high on the agenda of CEOs due to the vast opportunities (and risks) it provides. Whilst all the problems outlined above still apply, problem #2: the skill gap is even more apparent for all but the most consumer facing (such as ChatGPT) of AI tools. In addition there is one important additional problem largely unique to AI.
Problem 7: Data Quality
The recent success of AI has come on the back of great advances in machine learning. Whilst there are a number of techniques, they all require large data sets that ideally are complete, high quality and labelled. For most AI applications “garbage in” results in “garbage out” — no matter how good the algorithms are.
Whilst many organizations may already have this data, it is likely to be fragmented across siloed systems and have issues with quality. Cleansing and consolidating this data requires substantial data engineering efforts.