AI and the hidden climate cost of ‘dark data’
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....
by Bettina Büchel Published 27 May 2021 in Technology • 4 min read
Take time to do your research as mistakes can prove costly
Shared services operations first evolved about three decades ago so the idea of centralizing processes and having strong IT infrastructure that supports them , some would say, is a thing that exists but nobody talks about any longer. Yet the emergence of digital technologies has instilled a new sense that we can move to the next level.
Interesting digital technologies for shares services include:
1 Process robotics, a commercial off-the-shelf software that automates human activities for repetitive, rule-based tasks.
2 Cognitive automation systems designed to uncover insights that can enable decision-making by using judgment, reasoning, and remembering.
3 Artificial intelligence that helps make predictions.
While these digital automation technologies sound exciting, they require not only a business case but also strong planning efforts to ensure implementation. Let us, for a minute, assume you have a business case for your digital toolbox.
Are you at level 1? If you are at the sub-optimized stage, you are just starting to think about a service delivery model to optimize your non-core business processes, such as finance and accounting, or human resources. You probably have recognized that fragmented processes are holding you back and are costly, as well as unreliable and the quality could be improved. You are therefore starting with a limited number of functions to develop processes that will service the business with limited governance to start with.
Are you at level 2? If you are at the rationalized stage, you already have a consolidated service delivery model and are able to leverage economies of scale from transactional services across a number of functions with a delivery center set-ups where you might benefit from wage arbitrage. There is likely to be variation in terms of inclusion of processes and supporting technology and governance still needs to be further developed. Outsourcing models are tactically used.
Are you at level 3? If you are strategic stage, then you have a shared service model which seeks to be at the stage of continuous improvement in terms of process optimization and technology leverage and governance with global process owners.
Are you at level 4? If you are at the stage of having integrated business services, then you have a multi-functional service delivery model with coordinated processes, technologically-supported and a governance system that allows smooth interaction with the business. Typically, this also means having found a balance of capabilities internally and externally that deliver services with clearly defined global process owners.
Implementation is not only about getting the technology integrated to achieve higher performance, but also about managing the speed, resources, sponsor commitment and change activities necessary to ensure success. Here are some reflections on lessons learned:
Wanting to move too quickly – you need to understand your starting point to be able to reap the benefits of the business case. For instance, if you are still using paper-based invoices and want to automate immediately, then you are trying to skip a step which makes reaping the full benefits of the new digital automation toolbox difficult. From paper-based to data-based to automated requires having a service delivery model where core business processes are data based and already standardized, only then you can automate.
The relative immaturity of the digital toolbox applications – ranging from automated PO approval to robotized payroll processing to procurement risk analytics – along with the lack of standards of how to best integrate these tools into the existing landscape leads to unrealistic expectations. As a result, companies have to rely on external providers who do not know much about the companies’ unique environment, its cultural context and the specificity of the requirements given the industry context. Therefore, the implementation does not respond to the specific needs of automation and, as a result, the initial financial calculation of the business case is unrealistic.
Underestimating complexity – Global operations, multiple different businesses, process multiplicity and complex IT environments with different servers and applications can be difficult to integrate so considerable expertise and resources are likely to be tied up in integrating digital technologies. There is a clear danger of not allocating sufficient time, bandwidth and testing resources. In particular, testing can easily be cut short as the project becomes under time pressure to deliver the business case. This is made worse by not having enough data available or in a consumable format which can easily be used to “learn” from performance to identify patterns. This again can easily lead to delays and challenge the business case.
Investment in change management pays off – any change in process or technology usage typically leads to a change in workflow, formats or inputs. If this is combined with lift and shift processes going on, then any changes will be expensive. The longer you drag the implementation of process shifts, the more expensive they become.
“Our initial transition from function to shared services was planned for 12 months and then lasted 28 months – more than twice as much as initially expected. People did not want to follow what the company wanted to do – we had already paid a third-party service provider and we continued to pay personnel plus the project management support and travel – all of this made it super expensive.”
But even if you only have new digital tool adoption, you will likely run into process changes. Each time these process changes occur, there is a need to recode the robotic tools as the rules for automating processes and extracting information from standardized documents needs to be updated.
Before you proceed with any digital shared service implementation, do a check for the following: (1) degree of process change, (2) degree of technological change, (3) degree of standardization of process, and (4) degree of internal capabilities – the greater the process and technological change and the lower the standardization of the process and available capabilities, the more you need to potentially re-evaluate your ability to deliver on your business case.
Professor of Strategy and Organization at IMD
Bettina Büchel has been Professor of Strategy and Organization at IMD since 2000. Her research topics include strategy implementation, new business development, strategic alliances, and change management. She is Program Director of the Strategy Execution and Change Management open programs, as well as teaching on the flagship Orchestrating Winning Performance (OWP) program.
20 hours ago • by José Parra-Moyano, Christine Legner, Konrad Schulte in Technology
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....
18 April 2024 • by Quentin Gallea in Technology
GenAI will allow human talent to focus on more creative, complex tasks, leading to greater job satisfaction and well-being....
16 April 2024 • by Amit M. Joshi in Technology
Chris Tung is President of Strategic Development at Chinese e-commerce multinational Alibaba Group. He tells Amit Joshi how the company is integrating AI to optimize its customer experience....
15 April 2024 • by Annabelle Gawer in Technology
AI-as-a-Service will impact executives and their strategic decisions. It’s crucial to assess the benefits and repercussions it will bring....
Explore first person business intelligence from top minds curated for a global executive audience