Digital transformation has made businesses smarter and more capable than ever before. It’s also made nearly every industry much more intensely competitive, as the tools for digitalization are more equally available to all players. Beyond their own unique proprietary resources, all competitors have access to essentially the same third party data. Where companies can still distinguish themselves, though, is in their capacity to use the right data to make sharp, timely decisions.
For every customer interaction, decision science and predictive analytics can enable businesses to make the most of all available data to anticipate customer needs and prescribe the best action. Decisioning models based on analytics are transforming Origination, Account Management, Credit Risk Management, Collections and Recovery, Marketing and many other business processes as well.
Companies need to quantify customer credit risk, find ways to deliver better, more individualized marketing offers and comply with regulations governing their markets, without falling victim to fraud and other financial crimes – they key that allows innovators to win market share is their use of advanced technologies such as machine learning and predictive analytics. Competing with these companies can be challenging for those organizations that are earlier in the analytic maturity.
There is growing interest among prospective clients in building out a world class data science team, but this is often a difficult and expensive undertaking in today's extremely tight talent market. Actually, there an alternative: Work with an analytic partner to leverage their world-class experience to develop and deploy advanced predictive models and accelerate the maturity of your in-house teams.
One of the benefits of a partnership approach is that the client can have access to whatever types of models will be most effective in addressing their business problem, including the most advanced machine learning tools. But in many cases, the key to success is not the ability to come up with clever models – it’s the ability to operationalize those models. “Artificial intelligence” and machine learning are powered by advanced analytics, but many organizations struggle to develop and deploy analytics in a repeatable and cost-effective way.
The most effective advanced analytics are those developed by diverse teams that match technical people – modelers, data scientists and software developers – with business people who have important problems to solve. These people often speak very different dialects, so the best teams often include translators, either business people who grasp technical concepts or technical people who understand the earthier problems that business people deal with.
It is widely understood that competition for top data science talent is intense, and this includes a shortage of analytic leaders - people who have led analytics teams that pull together the right balance of data science and operational talent, under the right leadership. It’s a very human-scale problem, but ironically, one of the factors making it important now is the growing interest in AI and automation.
Analytic models have become much more complex and “black-box” in nature. Machine learning can achieve highly accurate results but understanding how a given result was reached is often impossible. Model explainability is important not only for the obvious reason – ensuring that regulators and customers have clear answers, but also, it’s an important step in model development to ensure that humans are “in the loop” for model development and can collaborate with this advanced technology to ensure it is designed to meet all stakeholder requirements. It’s also important from an ethical point of view, to ensure that there is no negative impact that exists in terms of social biases or other unintended outcomes that result from putting that model into production.
Regardless of the approach you take, a client partnership can make sense for many organizations to achieve a rapid step forward in their ability to successfully transform their company into an analytic leader. This may involve getting started by simply outsourcing a machine learning or “AI” modeling project but can also involve a longer term collaboration with internal analytic teams to augment their staff to reduce a backlog of projects, or provide more strategic executive guidance on model methodology, governance, as well as coaching and fostering collaboration of analytic and business leaders to achieve breakthrough results.
At the end of the day, it’s not often as simple as “build vs buy” or even “build vs buy vs partner”. A pragmatic, hybrid approach can ensure your organization is doing all it can to be an analytic innovator and customer experience leader. Regardless of the path you choose, FICO is here to help with world-class services, differentiated intellectual property, and our leading Digital Decisioning Platform.
Chisoo Lyons explores the issues facing “Model-Driven Organizations” in greater depth in a brief Hot Topic white paper, published by FICO. https://www.fico.com/en/latest-thinking/executive-brief/model-driven-organization-machine-learning-and-advanced-modeling