6 steps towards successful AI implementation in banking & financial services
Written by Anna Suhr
10 January, 2022
The Nordic region has steadily built a reputation as an innovator and early adopter in areas such as telecoms, music streaming and payments. Now, approaching the next frontier of tech innovation, Nordic companies are involved in the early exploration for AI-based capabilities. Research indicates that two-thirds of key decision-makers throughout the region view AI as either considerably or very important to their business, with half stating that they expect their business to be at a mature or advanced level within the next three years.
For BFS companies this presents a clear market opportunity, with AI as the driving force for gaining a better understanding of consumer behaviors and making more contextually relevant interactions. It also brings operational benefits, shifting away from conventional, siloed structures towards more agile, data-driven and customer-oriented models.
Making this transition requires top-down involvement in order to reinforce it as a strategic priority. There are six fundamental steps when implementing AI into your own BFS organization’s approach:
1. Identify AI use cases
You need to have well-defined business objectives that you want your AI implementations to achieve. Are marketing messages falling flat? Are users frustrated with your customer service centers? Are sales teams missing obvious opportunities for upselling services? These are all problems to which AI can be applied for better outcomes.
By adding this layer of extra insights to successful business practices, AI can help to optimize your current ways of working and interacting with customers. You should be able to articulate the business reasons behind each application with simple explanations as to what AI is adding to the process and why it’s adding value.
We’ve seen progress in this area among our customers. A banking client wanted to gain an edge through improved customer insights so we helped them to assemble a data lake, connecting a series of previously fragmented and siloed internal and external data sources to enable better decision-making. Matched against clear business objectives, the project brought better customer understanding, leading to improved satisfaction and profitability.
2. Get your data in order
Once you have your objectives established, you need to establish whether your current IT infrastructure is able to support the ambitions you have. The process of becoming a data-driven organization involves coordination of several stages, from collecting, cleaning and analyzing raw information from multiple data sources, to acting on insights and automating repetitive or low-level tasks.
All of these stages, when activated at scale, have massive implications for the underlying IT infrastructure. Even where data initiatives are experimental in nature, it’s important to think ahead to how it would work as a full-scale operation. Considerations, therefore, need to be made regarding whether the environment offers the capacity, reliability, availability and agility necessary to cope with increasing demands.
3. Have data science capabilities in place
In BFS, there’s been a noticeable rise in interest in data science, growing in parallel with the imperative of gaining a deep understanding of customer behaviors and adapting banking services around those dynamic behaviors. With vastly higher volumes of data emanating from services relating to open banking, mobile banking and digital assistants, data science capabilities are in considerably high demand.
This trend involves a cultural change for many organizations; firstly, in attracting and nurturing data talent, but also in integrating capabilities into existing processes and services. It’s for this reason that defining roles that combine ‘hard’ data skills with a more rounded understanding of business outcomes and customer behaviors will be increasingly paramount to the success of AI implementations.
4. Integrate data back into IT systems
In seeking to become more data-driven, the challenge many BFS organizations encounter is less around developing AI capabilities and more to do with coordinating them. If AI implementations and applications run in siloes, the organization misses valuable opportunities for leveraging data and collaborating around it.
By integrating everything back into your IT systems, your organization will be able to ensure quality, accuracy and, perhaps most importantly for financial businesses, compliance and data protection. By establishing systems and processes in place for smooth and consistent integration, your organization will be able to learn from the implementation process and continue to make necessary improvements over time based on the insights being generated.
These don’t need to be wholesale changes to your established IT systems, rather targeted improvements to enhance the availability of cloud services, establish robust access protocols and lower the cost of data management.
5. Organize to feed the insights back into the business
The data you collect as part of your AI implementations needs to be used as the key trigger for your future decisions and actions. What worked? What didn’t work? How were errors identified and resolved? Here the learning process is just as important as the final implementation.
Successful AI practices are just as much about knowing what you shouldn’t do as what you should, and this comes from a continuous improvement-based approach. Using real-time data as the driver for optimization is only effective if the system is designed to allow key insights to be fed back into your business operations, with flexibility embedded in decision-making structures to adapt based on these insights.
6. Work on the adoption
Adoption of these practices needs to be for everyone, not just those from native data and IT parts of the organization. Throughout the process of AI implementation, your employees should be encouraged to make reskilling and upskilling in data a more focused part of their job responsibilities. This may involve learning new software platforms, completing cloud certifications, or improving data analysis skills.
AI can even be used to support adoption and training processes. For example, Natural Language Processing-based digital assistants and training programs that integrate Augmented Reality can be effective means for onboarding internal users onto new software and embedding continuous improvement into the user journey.
Successful AI Implementation
While these six steps are all necessary to ensure the successful implementation of AI practices, none of this can begin without management and C-suite executives being on board from the outset. From improving staff awareness and advocacy through to defining responsibilities and investing in infrastructure, having decision-makers that recognize the long-term value of becoming data-driven and see it as their mission to drive company change will be crucial in developing a supportive culture.
For more information regarding how your company can manage successful AI implementation, in particular across the BFS sector, our eBook outlines key challenges and recommended approaches.
