Rethinking Bench Strength in the Age of AI

Most organizations have succession plans, and many have invested seriously in the processes and structures behind them. Yet according to DDI’s research, 80% of HR leaders still lack confidence in their organizations’ leadership pipelines.

That gap suggests succession challenges may run deeper than effective process execution alone. It is worth looking more closely at what is driving the lack of confidence, and what it would actually take to rebuild it.

AI advancement, continuous change, and shifting leadership demands are raising more fundamental questions: how relevant current succession criteria remain in the new environment, whether current successor pools will meet evolving business needs, whether development is intentional enough, and what is happening to leadership pipelines at their source.

The challenge is ensuring that organizations are identifying, developing, and sustaining the right future leaders in an environment where both leadership demands and the actual opportunities to develop them are changing simultaneously.

Before AI entered the picture, most talent leaders were already aware that succession management wasn’t working as well as we would like it to. Pipelines existed on paper in ways they didn’t always exist in practice. Talent reviews happened annually, and then a year passed with little meaningful action before the same conversation repeated. Plans included successors who have never been asked if they would actually take the job when the time came.

AI is now adding new complexities to a system already under strain. Roles have never been entirely static, changing in line with business strategy, priorities, and other organizational factors. Today, we must also factor in the impact of fast-evolving technology, causing redistribution of work between people and AI. This makes it very hard to predict what many senior roles will look like in five years, let alone plan for their succession.

On a more hopeful note, while uncertainty around specific future roles is real, there is growing clarity around broader leadership expectations. For example, expectations are shifting from relying on expertise to exercising judgment, from managing people to orchestrating human and digital work, and from functional excellence to systems thinking. I explored these shifts in more detail in an earlier article in this series.

These shifts have direct implications for succession. While uncertainty around specific future roles remains high, greater clarity around leadership expectations gives organizations an opportunity to act and verify bench strength by reviewing succession candidates against these criteria. As a result, someone currently in the pipeline may no longer belong there. Instead, someone currently overlooked may emerge as a strong candidate.

A few years ago, at Orbia, we invested significant effort in defining future-fit leadership expectations through deliberate partnership with senior executives and high-potential leaders rather than relying on historical models or generic frameworks. The effort resulted in a clear model aligned to the business context, culture, and direction, reducing the risk of building succession pipelines toward the wrong destination.

Leadership requirements will keep changing, and no model will stay relevant indefinitely. However, we can extend its shelf life by building in elements that encourage ongoing adaptation. For example, “developing yourself and others” was one of four core expectations we built into the future-fit leadership profile at Orbia. A model built that way allows for continuous evolution and doesn’t become obsolete at the first sign of disruption. Consequently, if the model stays relevant longer, readiness assessments remain valid longer, reducing the need for constant re-evaluation of the pool against a moving target.

With criteria updated and pools reassessed, the next challenge to address is successor readiness. It is a function of two things, accumulating experience and targeted development working together over time: stretch assignments, exposure to complexity, sponsorship, and focused work against specific gaps. Assessed at any moment, readiness is a point-in-time read. The expectation is that it moves in the right direction and at a pace that reflects real investment.

Yet most of us have participated in talent reviews where successor readiness hasn’t changed in years. Lists are calibrated annually, names are reviewed, readiness is assigned, but no development plans are put in place or fully executed. Development falls short. Progression stalls.

Part of the explanation lies in how development works in practice. Most meaningful leadership development happens on the job, through stretch assignments, delegation of increasing complexity, coaching, and feedback in real situations. That work is not in HR’s hands. It depends on managers and leaders who are genuinely stretched, balancing development against operational demands that rarely pause to make room for it. We can design development programs, curate experiences, and build frameworks. What we cannot do is manufacture the daily managerial intent that drives the 70 percent in the 70-20-10 model, and that is often where execution quietly breaks down.

This is not a new challenge, but it becomes more consequential in an environment where leadership expectations are evolving faster and where opportunity windows may narrow more quickly than before.

As leadership expectations shift toward stronger judgment and systems thinking, the challenge of developing those capabilities becomes even more acute at the very source of the pipeline. Historically, they were often built gradually through early-career experiences, apprenticeship, and exposure to increasingly complex work. Routine tasks, while operationally simple, frequently served as developmental building blocks. As AI automates more entry-level and transactional work, those foundational learning opportunities may erode.

MIT’s Andrew McAfee and others have raised important concerns around this apprenticeship disruption. Analysis of current entry-level hiring trends reinforces the concern: organizations reducing early-career headcount for efficiency reasons are simultaneously making succession decisions that will have consequences five to ten years down the line, whether they recognize it or not.

This raises a broader bench strength question: succession risk may no longer begin at senior leadership reviews alone. It may begin much earlier, through decisions about job design, automation, and early-career talent pathways.

There is broad agreement that succession should function as a live discipline rather than an annual event, continuously refreshed and responsive to changes in strategy, people, and the shape of the roles themselves. In practice, however, this has always been difficult to operationalize at scale. Business leaders don’t have the time for constant talent reviews, and aggregating meaningful talent intelligence across large organizations more than once or twice a year has been, practically speaking, extremely difficult.

Where AI brings a genuine improvement opportunity is in the quality and currency of the signals feeding into talent review. Modern talent intelligence platforms can now aggregate, across the course of a year, signals that previously could go untracked between formal cycles: goal progress, learning activity, continuous feedback patterns, engagement trends, and predicted attrition risk for candidates in the pipeline. It allows the cadence of succession discussions to remain manageable, while the information supporting them becomes meaningfully more current.

What that picture still requires, however, is judgment. Knowing that a successor’s engagement has dropped, or that their goal progress has stalled mid-year, tells you something happened. It doesn’t tell you why — whether the slow down reflects a capability gap, a change in the organizational environment, or something personal. Understanding that requires someone who knows the person, the role, and the context well enough to interpret the signal.

We can and should leverage new technology to improve the process, but we should also acknowledge that there are challenges it will not solve. That limitation in the data layer sits alongside a deeper one in the criteria themselves. New systems can optimize data collection, identification, and tracking against whatever criteria organizations give them. However, if those criteria have not been updated, even more advanced tools will simply surface the wrong people more efficiently. Better talent intelligence infrastructure will not replace the harder upstream work of defining what future-fit leadership actually means.

While succession is often considered an HR topic, the real reason leadership succession gaps matter is business continuity. At critical moments, insufficient bench strength delays strategy execution, weakens transformation, and increases organizational vulnerability when it can least be afforded.

None of this is straightforward. Succession planning sits at the intersection of strategy, organizational design, and talent development, and it has always required more sustained attention than most organizations realistically give it.

What has changed is the complexity of getting it right in the time window we have. The criteria are shifting faster. The pipeline is under pressure at both ends. And the window for building genuine bench strength, the kind that holds up when it matters, is becoming much narrower than before. We may have to give it more focus this time.