Will you please describe your role at Fractal Analytics?
I am the Chief Practice Officer for Insurance, Healthcare, and Hi-Tech verticals at Fractal. The Insurance practice is currently engaged with several top 10 P&C insurers in the US, across the Insurance value chain through AI, Engineering, Design & Behavioural Sciences programs.
Are you seeing currently any specific issues in the Insurance industry that should concern Chief Data & Analytics Officers? If so, what do you recommend to organizations to help them overcome these?
We see CDAOs grapple with challenges across multiple dimensions:
1. The urgent need to reskill the existing talent pool to ready them for a cloud and AI first world.
2. Lack of clear, unified, and scaled data engineering expertise to enable the power of AI at enterprise scale.
3. Progressing AI based solutions from proof of concept or minimum viable product (MVP) to production.
4. Regulations and compliance requirements, especially around pricing, risk selection, etc., present a significant barrier to adoption of the latest and greatest approaches.
5. Transforming the organization to collaborate/compete with insur-techs.
In addition, the traditional challenges remain. For instance, for a variety of reasons, in the short term, CDAOS are challenged with quantifying the benefits of analytics’ investments. Some of the work is very foundational, such as building an enterprise data lake and migrating it to the cloud, which enables other more direct value-added activities such as self-service. Moreover, rapid and full adoption of analytics insights can hit speed bumps due to change resistance in the ways processes are managed and decisions are made.
In the long run, we see a steep increase in the proliferation of all types of data due to IoT which will pose both challenges and opportunities. Companies need to be able to ingest, transform, and make the data available for analytics consumption in near real time. The data will enable companies to provide more personalized services and product choices. For instance, in Auto Insurance, connected and/or self-driving cars are expected to result in lower severity and frequency of accidents, leading to drastically lower premiums. Therefore, understanding customers for cross and up-sell is paramount.
Fractal’s recommendation is to take an incremental, test and learn approach to analytics to fully demonstrate the program value before making larger capital investments. There is usually a steep learning curve in terms of “doing AI right”, which is invaluable. Also, design thinking should play a large role in analytics in terms of how it will benefit the organization and exactly how people will react to and adopt the resulting insights. Finally, the test and learn approach and accountability should be strengthened so that algorithm-driven insights are given a fair shot to be incorporated into decision making.
In summary, Insurance carriers and brokers will need to ensure a sound data foundation and a smart use of the cloud to harness the value of the large amounts of disparate types of data. Analytics is a powerful capability enabler to help Insurers transform their operations and services. This progressive transformation leads to a better customer experience providing more value-added services that weren’t possible before. For example, leading carriers are using their apps to help customers improve their driving, offering special discounts on retail goods as a result, and subsequently cross and up sell their products and services.
What would you recommend to Chief Data & Analytics Officers who are looking to move into social media text analysis?
There are obvious concerns about privacy and ensuring that companies are conforming to privacy standards. These can be a moving target or “yet to be defined” standard. Beyond that, we recommend setting up the appropriate data management and engineering framework including infrastructure, harmonization, governance, toolset strategy, automation, and operating model. A properly set framework will ensure quality, timeliness, scalability, consistency, and industrialization in measuring and driving the return on investment. It is also important to have a strong test and learn culture to encourage rapid experimentation.
What is the most common mistake people make around data?
Underestimating the importance of data engineering to bring alive the power of AI. Consequently, building a world class team of data scientists, without investing equally on engineering. The rule of thumb I have seen work well is 2-3 data engineers for every data scientist in the organization.
How fast are the advances you’re seeing in AI now? What do you recommend to organizations to harness this but also show a solid ROI?
It is fast and slow. There is a dizzying pace in the development of autonomous robotics, generation of data from connected devices, improvement in algorithmic approaches, and the creation of stories around re-imagining businesses and value chains. There is a need to pivot these advances towards addressing inefficiencies across the value chain and simplifying customer experiences. The resulting cost savings can fuel the capital investment required to address growth objectives.
A playbook for this is to run multiple experiments in parallel and create ‘MVPs’ (fail/learn fast), as well as incorporate feedback mechanisms to enable an improvement loop, and scaling the ones that show the fastest path to ROI.
How can advanced analytics be used to improve the accuracy of forecasting?
The use of newer techniques, especially Machine Learning and Deep Learning, including RNNs and LSTMs, have high applicability in time series forecasting. Newer methods can work with large amounts of data and are able to unearth latent interactions. Such discoveries may enable significant reduction in forecasting errors as well as substantially improve revenue and profitability outcomes.
One of the most important benefits of leveraging advanced analytics in forecasting is the self-learning aspect of Machine Learning and Deep Learning techniques. Given enough trials and data, Machine Learning techniques are likely to add great value in the forecasting process.
What three pieces of advice would you give to CDAOs looking to get a better grasp of what customers really think of their products?
1. Apply a behavioural-sciences based approach. Emphasize the observation of human behaviour to understand what customers think rather than ask them. A significant proportion of human decisions are not just rationally driven. One approach is to use NLP techniques to analyze actual call center interactions with customers. Another is to recreate real-life situations (gamification) where people are brought into play games to observe the decisions and understand their emotional state.
2. Build multiple MVPs to test conceptually and learn from early user feedback. Incorporate these into subsequent releases.
3. Measure user adoption and engagement metrics to not just understand products take-up, but also to enhance the overall product propositions.
What advances do you see in Visual Analytics in the next five years?
There are three strong trends in Visual Analytics. First the interfaces will become more “natural”, meaning that the user will be able to type or converse in their language of choice on what they are looking for from the data. This helps the business analysts spend more time on insights and the needed actions rather than chasing data.
Second, the insights and anomalies in the data can be automatically detected and pushed out to the relevant users. This means that users will spend less time looking for patterns and insights and spend more time understanding and acting on the insights.
Third, the insights will find their ways to the user in multiple channels across multiple devices. Traditional Visual Analytics required that one sat down on a powerful desktop computer, but more and more the insights will be delivered via alerts, text links, collaboration tools that reside on mobile phone, tablet devices, and at some point, on VR devices.
What are you most looking forward to about CDAOI Insurance 2019?
To network and learn from peers, clients, and prospects.
What differentiates Fractal Analytics?
Fractal’s aspiration is to power human decisions in the enterprise by bringing together AI, Engineering, & Design.
We are rated as a Leader in the Forrester Customer Analytics Service Providers Wave, with industry leading scores across most contributing dimensions. We work with a significant number of the top 10 P&C insurers in the US. We continue to be rated as one of the best places to work by the Great Places to Work Institute.
Fractal has a strong partnership with major cloud providers including AWS, Microsoft Azure, and Google Cloud Platform. We are partnered with these cloud services providers in programs helping organizations navigate seamlessly into the cloud.
Lastly, Fractal’s clients provide us a Net Promoter Score on each one of our engagements, every quarter. Fractal’s 2018 Net Promoter Score is greater than 70.