AI and Digital Transformation: A Comprehensive Guide

For some time now, we have been discussing industrial transformations driven by digital technologies. The digital transformation journey has spanned digitization to digitalization, with various data technologies becoming the predominant driving force. To the extent that data-driven technologies are helping to manifest several advances in digital technologies as actionable, the next frontier in reshaping businesses is AI transformation. Developments in AI (mainly powered by advancements in machine learning combined with data-driven technologies) are pushing digital enterprises toward becoming intelligent enterprises. AI technologies are already revolutionizing not just how we perceive and do business, but also the business environment and its overall landscape. Take, for instance, how our online behavior has changed, how we shop today versus in the past, how we find a ride, or even how we socialize.

The rapid pace of developments in the technological space is making it increasingly difficult to keep track of, let alone interpret, them in the context of their impact on and consequences for businesses. One of the main effects of AI has been to enable the realization of value through advancements in other digital technologies in much faster, more efficient, and novel ways. While the literature is replete with use-cases that are being powered by AI, the focus of this article is on exploring how successful AI transformations can be undertaken in large enterprises. Current AI technologies are heavily reliant on the availability of right, relevant, and often abundant data. This dependency on data in turn makes the underlying digital technologies and data-related building blocks crucial for a successful AI transformational effort, since they power the acquisition, management, and organization of data as well as the mechanisms for the provisioning of AI solutions and user interactions. Consequently, any AI transformation is also a Digital transformation. In fact, digital transformation is a pre-requisite for AI transformation. It is important to realize that the rapid evolution in AI capabilities does not imply the replacement of existing digital initiatives but rather their strategic augmentation. Going forward, references to an AI transformation is to be understood in this broader sense.¹

The State of Affairs

From a business perspective, the companies that have succeeded in putting AI technologies into action have been native digital companies. That is, these companies either evolved alongside the technology itself or were themselves driving related data innovations and enabling technologies due to their internal growth needs combined with the unavailability of solutions in the market.

What does this mean for legacy organizations or for those that are not natively digital? Forced by external factors, they have started looking seriously into AI and digital technologies so as not to be left behind. In fact, a strong case has been made for legacy companies to digitally reinvent themselves. Realization of the need to transform and its urgency has resulted in a plethora of transformation initiatives in organizations in search of an elixir to all their digital challenges. Given related developments and enthusiasm, some have even perceived AI as a panacea.

However, we have already started to see some disillusionment with digital transformation efforts. While there have been isolated success stories (typically over-hyped), the necessary structural reforms remain a challenge. Most of the AI-driven innovation efforts tend to stay at the periphery and are almost never integrated into the core business.

Legacy Industry is finally realizing that AI transformations are not easy.

To understand why, let us first understand the fundamental nature of transformative forces.

What Necessitates Transformation?

Transformation typically is necessitated, or forced, as a result of various phenomena occurring simultaneously. These phenomena, taken together, change the businesses’ operational landscape in fundamental ways. Not just that, but the combined impact of these phenomena also affects the environment, mechanisms, and attitudes for addressing any consumer need. The advancement in data-driven technologies is one such phenomenon. However, this transformation is being compelled not just by data-driven technologies (including machine learning and AI) but also by additional phenomena that are (loosely speaking) co-occurring and complementary in making data technologies realizable. Among other advancements, these include our capability to crunch massive amounts of data in meaningful ways (predictive machine learning algorithms that would have taken years to compute in the late 1990s now require only a few minutes to a few hours) and our ability to transmit data at high speed with relatively high fidelity, thereby allowing for meaningful connectivity — add to that our ability to “sense” signals more widely, more precisely, and to act on them. Probably for the first time, we have a serious computing capability at the edge — devices that interact directly with either the environment or the end-user — that enables integrated decision-making capabilities. And finally, maturity in the software space is allowing for these pieces to come together in the form of powerful software products and services that can both be deployed efficiently and be integrated into existing systems.

While the building blocks for each individual factor have been in place for quite some time, it is the right combination of advancements coming together in each of these areas that makes the combined impact transformative. How do these developments translate into pressures on business? As it turns out, with increasing digitization, there is increasing pressure on both revenue and profit growth, since these technologies reduce economic friction.

Let us discuss why “disillusionment” occurs in non-native digital organizations and how organizations can ensure that AI transformation is meaningfully achieved. While the article focuses on AI transformation, the lessons it offers can be applied to any transformation in any industry as well as to any technological transformative phenomena.²

Why Do AI Transformations Fail?

There have been many attempts to understand why AI or digital transformation efforts fail. Most such analyses do identify some core gaps, including factors other than just digital technologies, such as: broader economic forces or market requirements; lack of investments in skills, infrastructure and processes’ upgradation; lack of calibration of digital investments to industry readiness, and; “rocking the boat” on the existing business model while chasing the moving target of a promising new (but unrealizable or too difficult) business model.

While these certainly are pertinent observations, I believe that they ignore some critical aspects. Transformations are seldom about technology. Consequently, failures seldom result from technological bottlenecks. There are other potential causes at play:

• These transformations are typically treated as be-all, end-all technology projects proposed to management by interested parties (both internal and external). As such, little effort or importance is placed on understanding the evolution of the industry landscape. Consequently, there is a lack of a coherent business vision. All the steps that follow from there are, proverbially speaking, rudderless efforts in one-off technological projects without an overarching strategic objective.
• The lack of vision and strategy is further exacerbated by resistance from legacy forces to new ideas and gaps in leadership that would otherwise provide the right skillsets. The result is that any effort, however well intentioned, can neither be assessed properly nor anchored in a broader transformation goal.
• Existing incentives and compensation structures encourage the maintenance of the status quo rather than risk taking for longer-term success. Transformations are inherently long term and risky efforts. Penalizing risk taking for long-term growth while expecting stellar outcomes of transformation efforts in the short term is misguided. For successful transformations, these incentivization and benefits structures must be either adapted or revised. The same goes for performance measures at various levels in organizations.
• Companies tend to morph all initiatives into a product focus aligned with traditional business practices. This myopic mapping not only results in setting up such initiatives for failure but also robs businesses of opportunities to either move up the value chain or pursue novel revenue opportunities.

Finally, many an attempt at AI transformation suffers from what I refer to as the playbook problem.

The Playbook Problem

We regularly see AI transformation playbooks being pushed out by experts (typically technology experts), consultants, and companies themselves. Most, if not all, of the most prominent playbooks focus on a few main points: perform small (but company-wide) AI projects, educate leadership on AI, have a data/analytics strategy, train the workforce on AI, and hire the right talent. The majority of playbooks have minor variations of these points, but overall the message is similar. These playbooks sell because their underlying message is simple: There exists a 4–5-step recipe that is easy to read and can form the basis for immediate “execution.”

I contend that the playbook approach is at best incomplete and, in any realistic case, will inevitably do a serious disservice to AI transformation efforts.

The primary motivators for such playbooks are the questions business leaders routinely ask — with the expectation of simple, deterministic answers. Here are two misguided questions that are typically asked by senior executives:

• What can AI do for me? — this is a general and open-ended question that demonstrates a clear lack of familiarity with the technology.
• What can AI do to transform my business? — this question originates from a combination of external forces (witnessing rapidly changing industry in which the business operates) and idealistic notions of a silver-bullet solution to help stay relevant.

Both questions establish an onus to prove (and thus guarantee) the business impact of AI transformation on the technologists. The natural inclination would be to focus efforts on automation, since automating existing functions is easier to understand and measure. This holds true especially in cases where a business provides no context to the technologists.

AI Transformation is not just automation. By only focusing on automation, businesses stand to leave a significant amount of money on the table, thereby losing out on crucial and strategic opportunities. Business leaders are better served by asking the right questions:

• Given the broader transformative phenomena at work, how should I envision my business in the changing landscape?
• How can I leverage novel technologies to evolve (as an organization) in step with changes in the business environment?
• Given how the world around me is changing, do I have a vision of the future, one which can allow me to adapt and grow the organization? Or am I clinging to a decades-old vision that I inherited?

These questions should force the industry to think about how the world is changing and how enterprises should evolve in response. Does this mean changing your business model? Adapting or building novel customer experiences? Driving efficiencies? Transforming internal business processes? All of the above?

Asking the right questions often also broadens appreciation for the opportunities afforded by a transformation, which would otherwise mistakenly be mapped onto a single dimension of customer experience. The answers to these questions are not trivial, as they presuppose not just an appropriate understanding of current technology but also a solid grasp on how the world in which the business operates is developing and what vision will best cultivate an apt response toward future growth.

Framing the discussion in this broader context also, by definition, necessitates a partnership between the decision makers, the technologists, and the technology strategists in the organization. This partnership is not just a good-to-have condition, but a necessary one. In most cases, however, this partnership never happens. The technologists take a backseat, being delegated the technological projects that the decision makers choose (with partial information at best), and are then expected to deliver results that transform the business. You can see that, by design, these initiatives are set up for failure, since neither party has the information it needs to make meaningful choices.

Organizations also often boast of initiatives taken to train the workforce in AI. This is at best a partial solution, one which can, in many cases, make things worse if not done carefully. Most training initiatives are pedantic in nature, almost never drawing the right connections between business objectives, problem formulation, and technological fit. Since no effort is made to educate the technologists on business operations and practices, AI execution teams are often kept in the dark when it comes to understanding the right metrics that the technology should optimize. It is critical that AI organizations promote visibility about the overall business processes, metrics, constraints, and strategies. Finally, the playbooks typically either entirely ignore or oversimplify the need for organizational readiness to undertake any transformation. Even worse, they presume readiness and hence take subsequent challenges completely off the radar, only for them to resurface much later in the process.

The overall takeaway is that there is no general playbook on how to undertake a successful transformation. A playbook can provide a rough guide or checklist, but each execution step is inherently subjective, and significant internal effort and commitment are required to address them. Further, It is important to realize that transformation initiatives for any given organization need to focus on the right objectives, since not all focal areas will yield uniform returns.

Undertaking a Successful AI Transformation

Below, I provide six core steps that every organization aspiring for a successful AI transformation should follow. This is not yet another playbook. Each step requires committed and intelligent execution. The execution of each step is not “templated,” i.e., the execution of each step can vary significantly depending upon factors ranging from the kind of business to the organizational structure to the transformation objectives to organizational capability and commitment to workforce readiness to skillset gaps, and so on.

Six Critical Steps for a Successful AI Transformation

1. Build an informed business vision looking outside-in: Typically, any transformation, by definition, applies to the core business or industry within which a business operates. As such, there are two key mistakes that companies often make when attempting to undertake a transformation initiative.

First — Looking bottom-up from a technology perspective: To leverage new advancements, companies often ask the questions I mentioned above, i.e., they look at what solutions a new technology can provide without understanding the broader context of what the business or customers need. This myopic inquiry results in an ever-increasing focus on the newest incremental developments and in an abundance of ever-changing buzzwords rather than in a comprehensive understanding of the fundamental changes taking place in the context of business needs. One of the most obvious instances of this phenomenon in recent memory is deep learning. I have witnessed innumerable occasions where the focus has mistakenly been shifted from discussing what the relevant business problems should be to the latest developments in the deep learning world. What is important is not necessarily the next incremental step in technological development. Rather, we need to consider how a combination of concurrent technological advancements combined with market expectations and changing realities are forcing a re-envisioning of business models and setting the foundation for future, albeit not immediate, “gratification.”

And second — Looking inside-out, or form-fitting an existing business to a new reality: This is the tendency to continue to do business as usual, repetitively asking the same questions about how a sizeable change can happen without committing any resources to adapting or evolving the current business model. Suppose you are an equipment manufacturer. Traditionally, manufacturers made the majority of their revenues from the margins on outright equipment sales. As such, manufacturing efficiencies and cost efficiencies once played a big role, as did the equipment’s unique selling propositions or USPs. Inside-out thinking would argue that the manufacturer should continue to address market pressures by aggressively cutting manufacturing costs and acquiring larger market shares. However, as we transition toward consumer behaviors that are increasingly focused on uninterrupted service and asset-operation in equipment-agnostic setups, the market is progressively driving businesses to provide robust service offerings. This pattern is being accompanied by the advent of new technologies, such as additive manufacturing, as well as by speedy design validations and operational feedback. These two phenomena, in conjunction, mean that on the one hand, equipment manufacturing is being commoditized, thereby exerting pressure on market shares. On the other, consumer expectations are driving service providers further up the value chain, such that if you wish to become a relevant player, USPs are going to come in service-related innovations.

It is in this context, then, that AI and related transformative phenomena can enable the organization to scale the value pyramid to capture not just a larger market share but also a larger revenue share (which, in the new world, are not necessarily the same things). This latter way of looking at the changing ecosystem is what I refer to as looking outside-in — that is, understanding how the market landscape and ecosystem are changing and establishing a clear vision about what role you would like to play in this new configuration. A business that always addresses relevant market needs is the one that will itself remain relevant in the long run.

A business vision that focuses on how to account for changing market expectations and how to build forward-looking business models will naturally lead to focal areas of benefit from rapidly evolving technologies. Putting an individual technology at the center makes sense only if you wish to automate processes rather than induce a fundamental transformation that will ensure a business succeeds and thrives in the long run. Looking outside-in will also illuminate areas in which new opportunities can be found as well as areas in need of further strengthening in existing lines of business. These can range anywhere from building operational efficiencies in existing business functions to refining or transforming business processes or customer experiences all the way to exploring and expanding the business model to develop novel markets and revenue opportunities.

Looking outside-in also allows for precise calibration to market expectations and readiness. The success of any product, offering, or service is extremely dependent on the timing and readiness of the market as well as of the industry. What is true for the automotive industry may not be true for banking, and vice versa.

Finally, it is critically important that AI leadership helps build this business vision, since these leaders perform a core role in informing the organization about the technological developments needed to identify and prepare for dependencies in that context. Moreover, they also represent one of the main execution vehicles responsible for seeing the AI transformation through, and hence they need to have a deep understanding of the business objectives.

Boardrooms must open up to AI leadership as AI and data-driven technologies become more central to a company’s vision and strategy.

2. Devising a solid business strategy: Knowing what you want makes it significantly easier to choose how you would like to achieve it. An informed business vision is priceless if an effective execution strategy is to be put in place. Such a vision would also enable stakeholders to ask the right questions as well as to decide on the metrics needed to measure and track progress. If Step 1 of establishing an informed business vision either is not accomplished or is vaguely put forward, then it will naturally be impossible to build any strategy. In fact, this happens much more often than commonly acknowledged.

In the absence of a coherent vision and strategy, technical teams (data science, AI, machine learning) are left pressured to come up with ground-breaking ideas, shifting accountability from decision makers to execution teams, which can be a critical mis-step. Can technical execution teams produce novel ideas or ultra-innovative solutions to drive exceptional market results, in abstract? Given that no anchoring of initiatives is established, combined with the fact that technical leaders are never part of vision development, it is impossible for these teams to produce results that can have a high business impact. In fact, even if the technology teams do come up with innovative solutions or ideas, they will almost never see the light of day, since organizational readiness is never factored in (see Step 4). In a nutshell, giving technical teams the onus of bringing about magical changes (or producing “wow” factors) without empowering them in any reasonable manner only sets them up for failure.

As a result, after a sustained exercise in trying to force aimless execution efforts, organizations often resort to “outsourcing a transformation strategy.” They do so by either tapping AI experts or selecting advising partners. Both of these actors can play a very important role in deciding what technology can or cannot solve (in the former case) or in providing education about how the landscape and ecosystem are evolving (in the latter case) — they can even identify potential areas that the organization should consider exploring. However, they can only augment Steps 1 and 2, not replace them. Researching ground-breaking algorithmic technologies and undertaking business transformations are not the same, even though they might have some overlap. The transformation strategy is much more than just a data- and analytics-strategy, as the rest of this article points out repeatedly.

To make matters even more challenging, the various internal structures in a large organization are not always aligned. For instance, the finance organization typically tends to contain budgets, while an innovation organization focuses on small proof-point ideas and projects. BUs typically are risk-averse and need certainty before they can adopt a solution, whereas leadership wants to see immediate results. External market and investor pressures can make matters worse. Hence, it is important that various organizations understand the expectations, buy into the overall vision, and support transformation efforts so that the journey from ideas to reality can be accomplished. Otherwise, the transformation effort becomes a tug-of-war between sub-organizations, almost guaranteeing its failure.

3. Focus on organizational culture and people: Transformations, even though touted as technology-driven, are first and foremost people- and culture-initiatives. The people aspect has multiple levels. A meritocratic choice of people to drive transformation efforts can be the key determinant of success. Moreover, transformation has implications for the entire organization. Clarity needs to be established, and the vision must be communicated not just to the core team but also to the other (sub-)organizations providing them with the relevant context, information, and details on the impacts of AI transformation. Understanding both context and impact is crucial to incentivizing various organizational structures and teams to support and participate in transformation initiatives.

As for the core transformation team, it is imperative to ensure that the most competent people lead the respective efforts. This determination is not a function of just domain understanding, loyalty points, or tenure with the legacy organization, but of the overall understanding of goals and the competence in technological depth needed to drive execution and broader ability in relevant problem formulations. This, then, typically returns to a point emphasized above on the necessity of partnerships between various stakeholders, including technologists, who should have a seat at the table not just during the execution of a vision but also on an ongoing basis, starting from the formulation of the business vision itself.

For a legacy organization, it can be extremely beneficial to incorporate external talent to inject a diversity of vision as well as an objective look at the problem. Having been embedded in a business and its processes for a long time can obstruct access to both novel ways of solving problems and to new and relevant problems. There is a reason why most of the disruption in various industries comes from the outside, often by people who challenge the notion that “the business has always been done this way.” It is important to inject new ideas, to be open to critical thinking and challenges to the legacy mindset. It is therefore not enough to just have new talent injected — this talent also needs to be empowered, lest the legacy process and mindset frustrate them or make them ineffective.

Legacy hierarchies must be broken up to enable freedom of expression, ideas, and critical thought. Flattening the organization is an absolute requirement.

Where is the critical skillset gap? — While a lot of emphasis has been placed on beefing up AI competency in the company, I argue that for most transformation efforts, the most critical skillset gap is not necessarily AI practitioners (researchers, data scientists, data engineers, and so on). Even though AI skillsets are extremely important, the biggest vacuum exists at the leadership level. The organizations are critically lacking in — and worse, underappreciating — these rare talents: leaders who can aptly bridge the gap between business objectives to AI/data-driven execution. These leaders need to be analytical thinkers and have a significant depth in AI and related technological understanding to be capable of quickly developing or adapting business priorities and objectives. These critical thinkers are typically also the main actors responsible for injecting disruptive ideas into business transformations.

4. Organizational readiness: Clear goals that provide the organization with stability must be established, thereby enabling them with resources and ensuring that the necessary components for taking technology to value are in place. For instance, a groundbreaking algorithm for performing a task may be amazing, but it will be of no use if the product roadmap never accounts for it — hence, Steps 1 and 2. The entire organization must be prepared to make sure that these efforts can be adopted, integrated, and scaled successfully. Below are the main aspects of this organizational readiness:

• Workforce readiness: Training the workforce on AI is not a one-dimensional task. Different parts of the workforce need different levels and different types of familiarity with AI. That is, training a large chunk of the workforce on building learning algorithms may not be the best route. Depending on how they will interact with and the role they will play in the new ecosystem, they will need to understand different aspects of AI. For instance, product deployment groups need a much better understanding of how to evaluate AI algorithms, not necessarily how to build them. Similarly, the parts of the workforce which would be utilizing the AI-powered capabilities must understand how their jobs, decision making, and dependence on the technology will be impacted as a result. Sales and external-facing organizations need to understand what the actual offering is in terms of impacting users’ or customers’ pain-points and what their clients should and should not expect from the new offering. Context-relevant skills development will also inject much-needed realism into the external perception of AI and tamper the hype which would otherwise result in misleading views on the business offering.
• Structural readiness: Innovation efforts typically stay, and often die, at the periphery of the organization, since little attention is paid to the internal structures and processes that can support their adoption, integration, and scaling. This is also one of the core challenges for internal venture, accelerator, or incubation efforts regularly pushed across the organization. One important aspect of this structural readiness is the capability to systematize AI and software development and deployment efforts in line with the transformation’s needs. Systematic building blocks, typically constituted as a platform, are extremely important for productionizing proof-of-concepts in a scalable, reproducible, and manageable way. In fact, the ability to scale the software function is one of the key differentiators between a non-native and a native digital business. It is important to pay attention to what solutions can best serve the scaling needs for the organization and reduce time to market. Platform efforts are necessary, but not all platform efforts are the same or can be replicated across organizations. Depending on organizational needs, these may include one or more horizontal development platforms (to enable common cross-divisional functionalities), customer-oriented platforms (to enhance seamless customer experience), product- or offering-oriented platforms (to support specific products or offerings), and so on. To support this, an efficient data strategy must be put in place that can cover aspects broader than just data organization and analytics readiness. It is important to understand the entire data landscape to appreciate and resolve the issues around data acquisition, ownership, governance, persistence, management, and so on. Building a data strategy is itself a critical and necessary exercise. Even though it is not within the scope of this article, it is important to realize that building data and analytics strategies cannot be done in isolation. They must be part of the larger transformation effort and have deep dependencies on what business objectives are being targeted.
• Brace for insights: One of the main reasons, often left out from the popular narrative, for intentional pushback on data-driven innovation efforts is the spotlight they shed on existing processes or methodologies. Often, existing processes are turfs controlled by select stakeholders in the organization. One of the interesting artifacts of AI and data-driven technologies is that they can introduce transparency and expose the weak spots or gaps in existing processes. These dark areas have traditionally survived either because they were never noticed (processes in large organizations evolve over time and are not necessarily intentionally sub-optimal) or were hidden intentionally. With AI transformation efforts, data will drive the outcomes. The efforts may also challenge conventional wisdom. In many cases, for instance, decisions are made based either on “expert” understanding of the markets or on traditionally held metrics. Often, these metrics have marginal, if any, correlation to reality. They are instead used as a mechanism to explain the events post facto. Ironically, they are still employed to make business decisions either because a better alternative has not been found or, more often, because they support the desired rationalization of decision making. The consequence of such processes is that they push the AI efforts to fit a set of pre-determined outcomes. Leadership must be aware of and work toward addressing these issues as they arise. These are not easy problems by any means. However, avoiding them will only further exacerbate existing vulnerabilities, which can culminate in disastrous endings.
• Change management: AI initiatives typically transcend individual teams, as they require various parts of the enterprise to undergo organizational change. It is important that this process is well thought out and prepared for before undertaking a transformation initiative. Anticipating and preparing for high-impact changes will not only help address natural anxieties and uncertainties but will also make the organization more receptive to the transformation’s outcomes. For AI transformations, a change management strategy must incorporate both the processes and the incentivization for the adoption of, and contribution to, data-driven technologies. This also includes preparing the workforce for data-centric thinking, training, and updating skillsets to enable the integration of AI capabilities in their processes and decision making seamlessly. Change management is one of the primary drivers of success when it comes to transformation initiatives, and organizations cannot afford to ignore it.

5. Avoid short-termism: Inherent risks and timelines must be realized and accepted. Transformation is a slow process and needs a long-term strategic vision backed by appropriate organizational and leadership support. While many aspects of transformation relate to efficient execution for enablement and are well understood, the innovation aspect is not as straightforward. When developing AI solutions, various hypotheses need to be developed, validated, and revised on a regular basis. Organizational maturity is key to allowing the entire innovation cycle to test ideas, validate them, enable them to accelerate, and scale them to adoption and integration. A short-term mindset can be perilous for these efforts.

It is imperative that organizations provide stability to both efforts and people. It is not uncommon to witness organizational leaders changing their minds on a frequent basis. This is an indicator of a lack of due diligence before initiating any transformation effort, or of the fact that the effort is serving a short-term purpose and is not aligned with the organization’s long-term strategic interest. There is something fundamentally wrong if the overall understanding of the landscape, and hence the vision, must be overhauled frequently or the strategy must be changed on a regular basis. It might take time to build the desired organizational structure, but it is important to avoid frequent re-organizations, changes in leadership, mandates, objectives, and goals. Constantly moving the goalpost is not only demoralizing but can also have systemic, undesirable implications.

While I will not delve into much detail on this aspect here, it is crucial that transformation efforts are shielded from inevitable internal politics and opportunistic takeovers at all stages. Predatory corporate politics not only sets efforts up for failure but can also systematically hurt the organization’s interests. While not often recognized, internal politics often derails any strategic transformation effort in the interest of maintaining the status quo. Any transformation effort will face massive challenges before it can attain escape velocity, allowing it to gain the momentum needed to enact irreversible change and ensure its longevity.

The strategic plan for long-term foundational change also needs to be reflected in the way in which various aspects of the transformation are measured. It is of course important to establish and track appropriate metrics. S.M.A.R.T metrics can certainly provide a good framework for tracking near-term milestones. However, it is important to realize that not all metrics may be quantifiable in the same way (e.g., translating to revenue) or even measurable directly (e.g., measuring culture change is difficult, even though certain surrogate indicators can be applied). In fact, the entire rationale for such transformations is not one-dimensional, short-term revenue or profit realization exercises. Transformations represent a long process, and if undertaken properly, they can yield long-term strategic advantages for an organization in difficult economic times or in outright downturns. Another aspect to keep in mind is that existing metrics may not be applicable and are not advisable for use as a gold standard, especially when they are not empirically grounded or validated. Objective, agreed-upon metrics are crucial to obtaining a consensus when evaluating the impact of a transformation. I discuss the issue of tracking here in the context of short-termism instead of earlier in the article for a reason. Most often the limited focus on demonstrating immediate outcomes and realizations of ROI’s results in employing the wrong measures to track progress or in entirely misdirecting the transformation effort, consequently jeopardizing its long-term success.

6. Assume risks and accept failures: However well informed, well prepared, and well strategized, transformation efforts run risks, since for legacy companies, uncharted waters are being tested. Hence, it is important to accept that there may be initial failures and to understand that there are inherent risks involved. When accounted for, and when lessons to be learned in subsequent iterations are anticipated, the transformation effort can lead not just to success but to surprisingly positive side effects, bringing the entire organization forward both technologically and culturally. While technology (AI capabilities, algorithms, data, communication capabilities, and so on) is typically listed as the biggest source of risks, most risk originates from non-technological sources. These include failing to undertake Steps 1 and 2 appropriately in favor of immediate rewards (i.e., short-termism), long-standing corporate bureaucracy, a culture deeply intent on maintaining the status quo, and a lack of appreciation for change management and organizational readiness.

There can certainly be failures in individual projects for a multitude of reasons, and a course correction might be necessary. It is also possible that some individual projects will need to be entirely scraped, but they will nonetheless result in important lessons learned about, for instance, the hardness of a solution or the availability of data, and so on. In fact, intermittent failures are inevitable and should be exploited to accumulate lessons across the organization. Building such organizational memory can be quite important, and sharing such lessons can provide long-term structural and intellectual depth to an organization. If the high-level steps to building a sound vision and strategy are robust, then the tactics can be adapted to reach its objectives. Some mistakes along the way will not put the entire effort at risk. However, a lack of high-level context can very quickly put the organization in a downward spiral, one which ironically is realized only when large-scale, often systemic, failures surface. By then, it is typically too late for course correction.

Further complicating the rapid pace of change in disrupting technologies, players as well as competitors can add to the uncertainties. There are both internal (and to a large extent controllable) and external (not always controllable or even foreseeable) factors that can introduce risks for the entire endeavor.

AI transformation initiatives should be seen not as binary initiatives but as foundational in preparing the organization for a changing world and to introduce within it a structural capability to adapt and evolve in response to these rapid changes.

This inherent agility is what allows the organization to drive innovation, rapidly adapt to the changing landscape, and evolve in sustainable manner.

Final Words

AI and Digital transformation are strategic exercises that address how the business can operate in a rapidly evolving landscape, identify key objectives, and develop the trajectories needed to attain those objectives. It is important that the entire transformation effort is anchored in a forward-looking business vision that sets a foundation for long-term growth and not immediate gratification. It is further critical to work toward developing a viable execution strategy as well as to enable the organization from both the personnel and readiness perspectives. Even though the article focused on internal readiness, mainly due to the emphasis on how to carry out a transformation successfully and efficiently, it is also very important to manage external perception. Customers and external stakeholders are increasingly aware of technology and the potential options they have to choose from when addressing their requirements. While it is important for the business to understand and address these requirements as part of its objectives for transformation and for an improved portfolio offering — one that is customer-centric and addresses customer requirements in the best possible manner — it is also important for the business to communicate these benefits to its customers and stakeholders in a realistic manner. It is certainly easy for advertising and marketing campaigns to become runaway organizations, taking on more fictionalized strategies to market a new product, offering, or even an entire organizational transformation strategy. However, keep in mind that with increasingly easier access to information, customers, partners, and other players in the ecosystem are becoming increasingly discerning. Such lofty claims and marketing attempts can yield a spike in benefits in the short term. However, it will not take too long for the realization to sink in that the reality of offerings cannot meet these promises. Misleading claims have long-term costs not just on revenues but also on the organization’s reputation and credibility. It is very important that the teams responsible for managing external perception are kept in the loop on the strategy, are made aware of the technology’s advantages and limitations, and have access to experts who can validate what is being promised on behalf of the organization and its offerings portfolio. An external campaign that draws on alignment with its internal strategy will be a significant asset in the long term. Just as short-termism imposes costs on the transformation initiative itself, it also imposes costs on establishing and reinforcing an organization’s credibility and reputation. Legacy organizations must also move away from the top-down driving of initiatives and products to a more customer-centric model, and they must accept and adapt to shorter, iterative launch cycles. Finally, it is important that the organization values a culture of meritocracy, discourages political opportunism, incorporates a diversity of opinions and perspectives, encourages the flattening of hierarchies, develops a long-term, growth-oriented outlook, and accepts the underlying risks of ambitious initiatives.

— — — — — — — — — — — — — — — — — — — — — — — — — — — — — —

Disclaimer: The views presented in this article are solely those of the author and do not represent those of any current or past employers.

[1] Further, I use the term AI loosely, since most such efforts are powered by a combination of technologies — and as far as AI goes, the main developments driving them are in fact from machine learning driven by data. As such, I am not using AI in its pure scientific sense.

[2] This article is not yet another discussion on all the different areas of our lives that are impacted by this ongoing transformation, be it social media, news, transportation, shopping, travel, and so on. Nor does it delve into the ethical or policy issues that surround them, even though I believe that it is crucial to discuss these issues at this juncture, too.