The challenge of setting ambitious targets as a leader
26 April 2024 • by Angelica Adamski in Magazine
Executive coach Angelica Adamski shares the challenges facing Lyndsey in aligning and leading a new and globally dispersed sales team....
Technology
In the age of AI, the whiteboard remains an essential tool
Published 30 October 2023 in Technology • 8 min read • 
Audio available
Technology may lead to a more efficient process, but it’s not the best option if it worsens the customer experience – and without due regard to data cleansing the results can be catastrophic.
In the early years of this century, the UK national postal service centralized and automated its financial reporting systems for thousands of post offices around the country, typically run by small, independent business owners, known as sub-postmasters. The project, supplied by Fujitsu, was called Horizon. After the roll-out, many of the sub-postmasters began reporting bugs in the system, often registering a significant but incorrect shortfall in takings. The system frequently failed to compute sums accurately. Someone who trained sub-postmasters on the system reported years later that he had reported this systemic error to his bosses, but the flaws were not corrected. Some postmasters even made up the accounting discrepancy from their own funds.
Instead of acknowledging the possibility of incorrect data, between 2000 and 2013, senior Post Office executives began criminal proceedings against over 700 blameless individuals, some of whom were sent to prison for “theft”. The criminal convictions are in the process of being overturned, and a public inquiry has been launched, but many lives have been ruined. It has been described as the most widespread miscarriage of justice in British history.
Amid the understandable desire in the business world to maximize operational efficiency through automation, and to be data-led, there is not always the same priority applied to data cleansing as there is to project roll-out and delivery. Data is so readily available and so convenient to use that it requires discipline to ensure it is “clean” – that it states what it purports to state. Early in my tenure as Chief Operations Officer at LEGO, I and an American colleague devoted many hours to ensuring more visible and more accurate operational data so that we could better match supply to demand.
“Instead of acknowledging the possibility of incorrect data, senior Post Office executives began criminal proceedings against over 700 blameless individuals, some of whom were sent to prison.”
Throughout my career in supply chain management, I have fought against an irrational bias in which strategy and the implementation of new technology carry greater kudos, and often higher salaries, than operations, logistics, and maintenance. The Horizon scandal is perhaps the most extreme example of this prejudice.
Of course, more than 20 years on, digital technology has developed considerably since Horizon, but automation was nothing new in 2000. Robots had been building cars since the 1970s, and air traffic control was automated well before the turn of the century. While it could be argued that such an appalling error as had occurred in the Horizon scandal could not happen with more sophisticated systems, one could equally well argue the opposite: that the very sophistication of newer systems may exacerbate the human tendency to have too much respect for the technology – to neglect data cleansing, fail to ensure checks and balances within managerial systems and, crucially, to lose sight of the over-riding purpose of serving the customer. An understandable excitement about the potential of new technology can result in the failure to engage sufficiently in scrutiny or due diligence. If we adopt the latest innovations without a critical eye and imbue them with special powers, we will fail to gain the best use of them.
In large language models and other AI applications, for example, there are acknowledged problems of “model drift”, “data drift” and “concept drift” – models whose usefulness has decayed resulting either from such factors as a change in context or methods of obtaining data, inaccuracy, the incompleteness of relevant data, or the bias of designers. The need for continual updating and training is increasingly recognized as essential in the field of machine learning.
What are your vital data points?
In my years as head of the supply chain at the toy company LEGO, I would place a strong emphasis on data cleansing and on being led by what I called the “vital few” points of operational data – the key indicators showing performance as experienced by retail customers and consumers, and whether operations were performing and/or delivering. It was essential to ensure the maximum flexibility and responsiveness to enable supply to match demand. Sometimes I would push back against a proposal from an engineer to automate a certain process because, with the technology available at the time, it would have locked us into a less flexible process. One result of this was that, on one occasion, a colleague expressed surprise when I advocated in favor of the automation of a process for molding machines so that waste could be removed by a robot arm and recycled back into production in the correct dose. He had assumed I was “anti-automation”. Perhaps I could have explained myself to my colleagues better earlier, but the point was: automation must serve the business, not the other way around.
Leading systems change
The sustainability transformation issue
No organization can escape the need to transform to become more sustainable. The need to act is urgent. It calls for strong leadership, difficult decisions, and deep cultural change. In Issue XI, we explore how to build sustainable organizations to succeed in turbulent times.
A useful distinction, I discovered, was between efficiency and effectiveness: a more efficient process that worsens the customer experience is not the option to take. In the context of an innovative toy company with its own factories and a high turnover in its portfolio of products, dealing with fluctuating and seasonal demand, the matching of supply to demand requires constant discipline, anchored around the “vital few” data points and overseen by a multi-disciplinary team. The over-automation of the packing process could result in overly rigid processes in which we were packing the wrong things more quickly. Too much inventory is inefficient from a business point of view and has a negative environmental and brand impact. The efficiency of our molding processes, on the other hand, was continually enhanced through robotics.
There are essential disciplines to observe when weighing up where to deploy data and the latest technology. The central one is to maintain an overall understanding of the entire value chain of the whole business. This is allied to working back from the customer’s needs. A multi-disciplinary approach, in which all relevant divisions are represented, is essential. The decision on which parts of an operation to automate, and how, should be a joint managerial decision, not one that is left to technology gurus alone. Once this is decided, then the tech experts can oversee the implementation.
Visualize your operations as a team
An approach I introduced at LEGO – initially in our North American operations but eventually adopted globally – and used in other companies is that of the “visual factory”. This is anchored around a weekly operations briefing, which provides a concise yet full overview of the company’s operations, attended by all relevant personnel (and no one else!). The briefing starts on time, typically early morning, and only lasts for 30 minutes, or 40 minutes if there is a major issue to address.
The “vital few” points of data (and nothing else!) are written up on a whiteboard. They are coded green for “on course”, and red for “attention/action”. The kind of data used in this process may change over time – at LEGO, we would have a review every six months. Typically, data would cover areas such as customer service, complaints, progress against delivery, and business results as well as health and safety indicators.
The physical act of writing on the whiteboard serves to embed the importance of an action in the individual’s memory. It reinforces personal accountability too. The individual responsible for an action must report back to the group the following week.
The physical act of writing on the whiteboard serves to embed the importance of an action in the individual’s memory. It reinforces personal accountability too.
The handwriting and whiteboard may sound like an archaic approach in the age of AI, but it is not. The data will, of course, be up-to-date and supplied by the latest systems, fed in by retailers as well as internal sources on real-time systems. The idea is to select the points that most directly affect performance and to be anchored around those. A problem with data on a screen is that it is too easily absorbed and too quickly forgotten. And there can be way too much of it. Provided the data is accurate, up-to-date, and supplied by the most modern systems available, then the “visual factory” harnesses the best of data-led intelligence and cognitive human engagement and facilitates collaboration.
A business needs to find its own cadence, or rhythm, within which it ensures constant vigilance – therefore, such briefings may be more, or less, frequent than weekly in another sector. In the meetings themselves, this vigilance involves asking the “What if?” questions. You must ask yourselves continually: “What is the problem we are trying to solve here? How will automation help us?”
Another important feature within the “visual factory” approach is the presence of everyone in the room (or on the video call) who affects the outcome. They should be given freedom of expression. This includes relatively new or junior individuals: it is not a hierarchical approach, but rather one that enables collaboration, honesty, and listening. Had this approach been adopted at the UK Post Office, it is likely reports by sub-postmasters of worrying bugs in the system would have been taken seriously and corrected before it was too late.
That regrettable tale reminds us that automation and AI will not automatically improve operations, efficiency, or customer service. Yes, they can enhance, even transform, performance, but only when overseen by a well-trained, tuned-in, multi-disciplinary team with an informed and open-minded view of the whole value chain. The purpose of a business, after all, is to serve the customer, not to pursue technological change as an end in itself.
Authors
Related
AI: a dystopian future for academic freedom?
23 April 2024 • by Jerry Davis in Magazine
GenAI can deliver cost and productivity benefits for colleges and students, but there is a dark side....
Audio available
AI and the hidden climate cost of ‘dark data’
22 April 2024 • by José Parra Moyano, Christine Legner, Konrad Schulte in Magazine
Training and fine-tuning AI models requires troves of data, but storing and processing that data is financially and environmentally costly. Here are three ways to address the problem....
- Audio available
How ChatGPT can unchain you from mundane office life
18 April 2024 • by Quentin Gallea in Magazine
GenAI will allow human talent to focus on more creative, complex tasks, leading to greater job satisfaction and well-being....
- Audio available
Learn Brain Circuits
Join us for daily exercises focusing on issues from team building to developing an actionable sustainability plan to personal development. Go on - they only take five minutes.
Explore Leadership
What makes a great leader? Do you need charisma? How do you inspire your team? Our experts offer actionable insights through first-person narratives, behind-the-scenes interviews and The Help Desk.
Join Membership
Log in here to join in the conversation with the I by IMD community. Your subscription grants you access to the quarterly magazine plus daily articles, videos, podcasts and learning exercises.