Two questions to help managers rethink bureaucracy in the AI age

In a world of rapidly changing business environment, one thing appears to remain stable: a company’s bureaucracy. “In this company, bureaucracy is the real barrier to change” told me Heinrich, a discouraged employee at a chemical company.

Heinrich’s remark is a common complaint from non-executive employees. The reality is that most established organizations are still shaped around bureaucratic norms, codified in detailed procedures, rules, and authorization steps, creating an unbearable burden.

Based on extensive research and applications at companies like Netflix, Handelsbanken, Enel, Affinity Plus Bank, and other hands-on projects, I have developed a framework of control levers that companies can apply to reduce the bureaucratic burden of a process (see also a previous IMD article by my colleague Gabriele Rosani and my late leader and friend Alessandro Di Fiore with whom I first brainstormed this approach).

The framework (called also Trust-by-Design), fine-tuned with the work on field, is now based on the following levers:

·??????? Boundary rules

·??????? Guiding principles

·??????? Context setting

·??????? Data-driven decisions

·??????? Sounding board

·??????? Peer review

·??????? Social control

·??????? Detection-based control

Yet, companies typically have dozens of different processes; therefore, selection and prioritization are key.

Clear criteria and guiding questions can help leaders select which processes in their organization they should focus for reducing/eliminating bureaucracy (and which not).

It's important to note that problem is not bureaucracy per se. Instead, it is the organizational burden generated by bureaucracy. When dealing with bureaucracy, managers need to answer two fundamental questions:

1. Within a specific area, is bureaucracy a burden?
2. If so, can we reduce the burden?

While the first question evaluates how harmful the existing bureaucracy is in a given area, the second question is about the feasibility of a low-burden solution.

The answer to these questions should not be driven by perceptions, opinions and feelings (as often happens) but rather should follow a more structured analysis.

To answer the first question—Is bureaucracy a burden?—it is key to calculate a “Bureaucracy Burden Index” for each process in order to set a baseline. The burden index is a percentage score from 0 to 100 with four main brackets: processes scoring low are not considered burdensome, while the ones scoring very high bring significant pain to the organization.

To calculate the score of the index, I recommend to combine a mix of quantitative data and qualitative elements. Quantitative metrics like cycle times, wait times, process variability, and approval steps provide evidence of the bureaucratic burden.

Modern techniques like process mining can contribute to audit processes and extract knowledge from event logs. In addition, structured interviews with the process owners and a group of users help investigate current frustrations (e.g. hassle, frictions, wasted time, related disengagement), potential benefits (e.g. gains in speed, flexibility, freedom, innovation), as well as the costs of reducing bureaucracy (e.g. risk of compliance, variability of standards, irreversibility of wrong decisions, severity of consequences).?

Generative AI tools like ChatGPT can assist in developing more objective and consistent scoring, enabling better quantitative comparisons across processes.

?

While managers may be tempted to first address the processes with the highest burden index, in reality, it is wiser to carefully assess whether the necessary technical and organizational enablers are present to a sufficient degree.

To answer the second question—Can we reduce the burden?—there are three main enablers to consider. Let’s call these the “3 Ts”:

·??????? Technology: availability of AI-powered tools and presence of digital systems enabling quicker and more informed decisions, better transparency, and more efficient tracking and control.

·??????? Talent: on the people side, having higher talent density favors empowerment and freedom. As discussed in my Harvard Business Review article on 'Human Judgment in the AI Age', when employees have the skills and the knowledge to exercise judgement, they can make good (and safe) usage of data and of AI-generated advice, knowing how to properly prompt and how to interpret the outcome.

·??????? Trust: a precondition for freedom is to have leaders that trust both technologies and people, by giving clear boundaries for autonomy to their teams, including contextual guidelines on how to use AI tools.

Examples of sub-drivers for analyzing each of these elements are illustrated in the Table below.

?

THE “BUREAUCRACY BURDEN” MATRIX

The two questions combined generate a matrix, where the horizontal axis measures the level of the burden and the vertical axis the presence (or absence) of enablers to reduce/eliminate burden.

While the matrix can be used for evaluating a single process, it is particularly powerful when applied to a set of processes or policies to understand the relative positioning and decide where to act first and how to phase the resulting actions.? The matrix has three zones: Keep, Reduce, Eliminate.

?The ELIMINATE zone.

In established organizations it’s not difficult to find areas where the legacy bureaucracy burden can be eliminated. Over the decades rules and requirements accumulated, but many are simply a relic of the past. For example, maintaining outdated onboarding procedures which are heavy in paperwork, where digital solutions are easily available.

Automation can play an important role in reducing hassle and frustration for employees—see, for example, the case of American Express and Microsoft’s new AI-powered solution facilitating business travel payments for employees.

Sometimes bureaucracy is imposed by external regulators and cannot be avoided, but its burden can be minimized thanks to modern technologies (for example, GDPR compliance with robotic process automation).

In rare cases, companies can achieve a zero bureaucratic burden by simply eliminating the root cause of the bureaucracy, without relying on technology. For instance, Netflix and Virgin stripped back vacation and travel policies. Such an unorthodox approach requires very high talent density and trust as a precondition.

The KEEP zone.

It needs to be remembered in any assessment of an organization’s red tape that bureaucracy is not always necessarily a bad thing. There are situations where it should be maintained.

Think of when a company does not have proper technology enablers, when it’s difficult to digitalize and automate specific processes, or when there is a lack of trust in technologies and/or people. For example, policies and procedures related to harassment or anti-bribery especially when the context is unfavorable (as in countries which feature prominently in the high corruption index).

The REDUCE zone.

Based on my experience, most processes end up being positioned in the middle zone of the matrix, where employees are given more freedom and empowerment within defined guidelines, context, and trusted judgement. For instance, at Ritz-Carlton people empowerment is a key enabler: to ensure an exceptional and memorable guest experience employees are trained to follow guiding principles rather than strict codes, and are encouraged to find creative solutions to fulfill customers’ requests.

In other sectors, the workforce is empowered to make real-time decisions enabled by new technologies, think for example of AR/VR used by field maintenance operators.

New technologies like generative AI have the potential to support people in making better and faster decisions, however higher judgement and new forms of prompting, interpretation and control will be needed to counterbalance the associated risks (e.g. hallucinations, and biases).

The three types of enablers (Technology, Talent, Trust) are often intertwined when balancing freedom and control. Working from home is an area where investing in the 3Ts is particularly important.

At Enel, an energy company I have worked with, instead of simply rewriting a detailed and prescriptive new standard operating procedure, they defined principles and guidelines on where, when and how to work. These were displayed in a simple way, with practical examples and a series of criteria for judgment (including type of work and task, geography, technology available, productivity, and personal preferences).

Enel applied a similar logic to other policies (like travel, recruiting and recurring investments) using a mix of the levers described above, from simplification (guiding principles, boundary rules) to decision support (context setting, data-driven inputs, sounding board, peer review) to smart monitoring (social control, post-detection).

?

TIPS TO APPLY THE MATRIX

Beyond gut feeling.

Beyond mere intuition, evaluations based on anecdotes or perceptions should never replace more structured assessments, which should be grounded in fact-based quantitative scoring.

I recommend using a mix of focused interviews and quantitative metrics to ensure a precise positioning on the axes of the matrix and a baseline to monitor future progresses.

When conducting the analysis, it’s advisable to involve tech experts and other departments since the beginning, including control functions like legal, audit, compliance. Based on my experience, once they understand the logic and the approach, leaders of control functions become influential promoters of burden elimination or reduction.

Context matters.

It’s important to note that different companies (or even different units or geographies of the same company) might decide in opposite ways about the same process.

For example, a company like Netflix would position the travel and expenses policy in the Eliminate zone of the matrix (thanks to a high talent-density and a trust-based culture) while a large global consulting firm may resist complete freedom and prefer checks and balances on a process that impacts heavily on the cost structure of the organization.

Phase the shift.

Abruptly shifting from “Keep” to “Eliminate” can easily backfire when proper enablers are not in place.

For example, a utility company fascinated by the promises of the “Eliminate” zone, drastically removed some authorization steps for digital expenditures. As they lacked two critical enablers: digital talent to make good decisions and smart real-time visibility on cost side, costs skyrocketed mainly due to duplications and integration issues. As a consequence, the worried top management went back to an even more rigid authorization procedure, thus worsening the initial burden.

To avoid backlash, phasing the shift both temporally (by gradually removing controls while investing in enablers) and spatially (starting in a sub-area and then extending to other areas) is critical.

CONCLUSION

Bureaucracy is not universally good nor bad.

By contrast, the burden always negatively impacts people’s motivation and wellbeing, and thus should be prioritized. If, where and how to reduce the bureaucratic burden depends on the presence of three key enablers in the organization: AI-powered technology, talent and trust.

These vary from company to company based on the context and on the legacy system of the organization. A simple matrix can help visualize what areas of a firm should and could be simplified and what areas should or could not.

Having clarity on these fundamental questions is a good starting point for addressing bureaucracy in a more pragmatic way.