Here, I delve deeper into the topic and suggest three questions that all CTOs or CIOs must answer to reap the rewards of ML in the long-term.
1. Where will the value come from?
When identifying the right use cases for machine learning, it is critical to pinpoint exactly where the value of the program will come from; how big is the impact on board-level business priorities? How quickly will it deliver value? If you are optimizing your internal operations, a common KPI is reduced resource burden, or supporting a process that is already being carried out manually by employees. By automating repetitive, slow, and data-driven processes, you can increase throughput and reduce time and cost; one area we have seen ML make a significant impact on is in customer service. The rise of transformers has led to groundbreaking language-based applications such as Google’s BERT or Open.Ai’s ChatGPT. This exponential growth in the world of conversational AI allows for chatbots that can better understand and generate human-like language for customers 24/7.
Another internal process ripe for automation is risk management, particularly improving risk analytics and implementing ML-based forecasting models. As well as reducing cost and resources, many ML use cases can help drive additional revenue. For instance, with recommendation systems in digital stores, companies can deliver relevant content, drive more traffic to their business and significantly improve customer engagement.
2. What data is available?
Every ML use case relies on data at its core. When choosing a business challenge to solve, you must identify if the necessary data is available within the business – and if it is the right data. I will talk more about the importance of data in detail in my next article, including how to capture it, cleanse it, and prepare it for data science (often the most time-consuming part of a project).
However, to scale any ML project successfully you must also look to create a data-driven culture within your organization first. If you are a large enterprise business, it is likely that most of your employees have been collecting data throughout their employment (intentionally or not!). For example, your marketing team will be gathering information from Google Analytics and their automation platform. Your sales team will be using CRM systems daily. These data-savvy people have the capacity to turn data into insightful information – and upskilling them in data governance will have a significant benefit.
3. Who are the key stakeholders?
It is a given today that there needs to be a strong collaboration between IT and business stakeholders to understand how ML can make tangible improvements to the bottom line. At the top level, the role of the C-Suite in these programs cannot be understated. They may not have a direct influence on the day-to-day workings of the project; however, not only will these leaders be able to identify the right opportunity areas for the project (including those that suit the boardroom’s long-term goals), but they are also responsible for driving it to successful fruition.
At the project level, the key stakeholders will range from project managers and domain experts to scrum masters and third-party project partners. These stakeholders will be deeply involved in defining the goals, identifying the requirements needed to fulfill these goals, and managing the implementation itself. Additionally, there will be data specialists, data engineers, and data scientists who look after the data prep, model selection, tooling, and programming. I will discuss these tasks in more detail in my next article.
Getting the groundwork in place as described above is phase zero in the journey to successful machine learning at scale. At Cuesta, we take a holistic look at your business and technology, finding connections and identifying obstacles – and create ML solutions that can achieve real ROI at scale. Read more about our offerings here.