Finding the processes that data should inform and then making sure the data the process is informing is accurate, high-quality, and in the right format are the first steps in becoming a fully data-driven organization. Once these requirements are met, the company may set up clear, structured, and well watched data processes that will allow its divisions to use the data. These adjustments enable teams to efficiently convert insights into decisions that can be implemented, which eventually results in more informed outcomes.
For a number of reasons, a sound data-driven decision-making strategy is currently deemed necessary. First of all, it enables businesses to make more assured and proactive choices, ultimately resulting in cost savings. Years before the AI revolution would become widespread, a 2014 McKinsey Global research found that data-driven businesses were already 19 times more likely to be profitable and 23 times more likely to attract new clients.
According to a more recent McKinsey analysis on data-driven businesses from 2022, businesses that use AI to contribute 20% of their earnings before interest and taxes (EBIT) are far more likely to use growth-minded data strategies. According to the study, these practices include real-time information delivery, flexible storage that makes data ready for use, operating models that treat data as a product, efficient data sharing with rivals, and giving chief data officers the authority to create value for their companies.
Daniel Faggella, the CEO and head of research at Emerj, recently met with Mike Borrelli of Uniphore and Ciprian Porutiu of Marsh McLennan to talk about the difficulties and methods of successfully integrating AI technologies in businesses.
Data-driven risk consulting services and insurance solutions are offered to both business and consumer clients by Marsh McLennan, a professional services firm that specializes in risk, strategy, and people. Using generative, knowledge, and emotion AI in conjunction with workflow automation, Uniphore is an AI vendor that specializes in integrating AI technologies to improve consumer experiences across many industries.
Financial services executives can have a better understanding of the management approaches that will put them on the correct track to utilizing data-driven organizational capabilities by exchanging these contrasting viewpoints.
The paper that follows examines their viewpoints and offers executives the crucial takeaways from their discussion:
Prioritizing repetitive processes for early AI adoption: By concentrating on repeated operations for early AI adoption, a foundation for automation and efficiency gains is laid for data-driven decision-making.
Using a variety of data and user feedback to optimize: To improve client experiences and streamline procedures, user feedback is gathered, various data kinds are utilized, and results are measured.
Giving Repetitive Tasks Priority In The Early Adoption Of AI
Mike Borelli starts off the discussion by emphasizing the importance of giving priority to projects that have the potential to have a big commercial impact. He does, however, stress that the decision-making process should also take into account the long-term effects, the wider adoption landscape of AI, and any potential obstacles.
The difficulties in selecting various AI solutions for business application make it hard to confirm their legitimacy and effectiveness among a plethora of choices. He emphasizes how adopting numerous solutions can lead to data fragmentation, necessitating a more objective understanding of operations.
There are numerous firms who like to keep your information secret. Nowadays, data sovereignty is crucial because many of the bigger vendors charge for access to your data. That’s not a good strategy, and to Ciprian’s point, it means: Before attempting to determine where automation may have the greatest influence on your organization, how can we ensure that the data is easily available and accessible to help us make better decisions?
—Uniphore’s Sales Director, Mike Borrelli
There is no one-size-fits-all method for choosing AI solutions, Mike stresses to the Emerj executive podcast listeners, adding that the decision is based on the particular requirements and circumstances of each company. To create a basic structure, he advises tackling repetitious and readily automatable chores. By doing this, businesses are able to comprehend their procedures and create models and workflows, including client journeys, using a variety of chat and voice channels.
The foundation for additional automation and support, including the deployment of agent assist platforms, is laid by that preliminary work. Mike also raises the possibility that various innovations, including work summaries produced after-call, could continue to enhance human potential and promote efficiency gains.
From his point of view, Ciprian presents a methodical and philosophical strategy for tackling the difficulties of semantics and data management in the framework of regional agility. Echoing Mike’s earlier view of incremental growth, he advises starting small and evaluating what works before expanding.
Ciprian does, however, also recognize that multi-national corporations face difficulties standardizing data collecting and interpretation due to linguistic and semantic variations across nations. In his ideal world, big language models will be able to sort through unstructured material without the need for strict taxonomies. With this method, data would no longer need to be arranged into distinct lakes or warehouses because it would all be in one central area.
Using Diverse Data And User Feedback To Optimize
Ciprian then recommends beginning with more standardized jobs where the company and the client can both gain from direct access to information. Automating repetitive tasks, like filing claims, can help organizations increase productivity and streamline operations.
He also emphasizes how automation can result in immediate benefits, including automatically closing particular claims in accordance with preset criteria and thresholds.
This is where the expert’s interface transforms the visual intelligence of data reporting or that specific output of LLMs into something that speaks to the issue. These days, we frequently develop solutions rather than address issues. Therefore, the solution turns into the lovely tool we recently developed that can react to any kind of interaction without considering the possibility that the customer’s issue is something else.
— Ciprian Porutiu, Marsh McLennan’s VP of Operations & Delivery for Digital Transformation and Change Management
In order to make sure that bot designs satisfy user expectations and save expensive redesigns, Mike emphasizes the importance of obtaining user feedback throughout deployment. He also notes that businesses are increasingly using several forms of data (phone, video, chat, etc.) to create proprietary enterprise language models (ELMs).
Over the years, you can refine it to make it suitable for other nations, languages, and semantics that you need to achieve in order to make it strong enough that your clients are satisfied with the service as well. You can then take use of self-service and stuff. Because big businesses will feel more at ease with something they know has more in-house knowledge base, I believe that’s where it will start to go.
-Uniphore’s Sales Director, Mike Borrelli
Cyprian talks about the idea of a “flow framework,” which connects several software development layers to provide value in the form of improved revenue, efficiency, customer experience, and compliance. It focuses on the transition from project to product.
He changes the paradigm to achieve measurable business results by highlighting the significance of evaluating outcomes rather than just outputs:
The paradigm shift there is caused by the necessity to measure results rather than outputs, which allows you to receive those kinds of results directly and quickly change course. This is why a quick glance into the database, where the software’s modules are created, immediately yields value.
Now, you add it if the procedure adds value by improving the customer experience right away. Since there is a clear link between the commercial value and the technical terminology used by those who produce those modules, you can immediately stop doing it if you notice a negative reaction.
At Marsh McLennan, Ciprian Porutiu is the Vice President of Operations & Delivery for Change Management and Digital Transformation.
In his concluding remarks, Mike stresses the significance of data aggregation, analytics, and a feedback loop when implementing User Accepted Testing (UAT). To guarantee that insights are gained early and frequently, enabling optimization and fine-tuning, he emphasizes the necessity of iterating on these procedures during the implementation. Mike agrees that tone detection and sentiment analysis can offer important insights into consumer interactions.