Having completed the exploration of our global trends in our previous post on services, investments and job roles, we now turn to our key themes. In our next round of posts, we will explore 11 key functional and technological areas within insurance, starting with today's installment on analytics and AI.
The following statistics on big data, analytics and AI are drawn from the extensive survey we conducted as part of our Global Trend Map; a full breakdown of our respondents, and details of our methodology, are available as part of the full Trend Map, which you can download for free at any time.
Insurance, relying as it does on predictions about complex future events, has always been a data-hungry, data-driven industry. The big-data explosion of the past decade is therefore something that insurers have followed with keen eyes:
Analytics has applications across all the major lines. Health and auto are two obvious examples given the ready availability of connected health devices and in-car sensors, which make data easier to capture and, as an extension of this, models easier to feed. This facilitates usage-based insurance (UBI), which we explore in more detail in our forthcoming post on Internet of Things, whereby actual living/driving habits inform policy prices (read ahead straight away by downloading the full Trend Map for free).
Analytics also has obvious applications for predictive maintenance and security in commercial, auto and P&C/general lines, particularly where valuable assets (like property) are in play. Analytics is also growing in home insurance thanks to the increasing prevalence of connected-home devices, with Berg Insight estimating that there were approximately 18 million smart homes in Europe and North America by the end of 2015.
We asked respondents whether legacy systems and silos represented "somewhat" or "very much" of a problem for their businesses, and then created a "burden score" based on a weighted combination of these two figures. Insurers clocked up a burden score of 138, brokers and agents 103 and technology partners 105. (Reinsurers score 123.)
There are two key takeaways from this. Firstly, that silos and legacy systems are a problem for the entire insurance ecosystem. And secondly, that carriers are generally harder-hit (comparing insurers and reinsurers to the rest of the industry), which may well reflect their position as the central node of the industry into which all the other players feed.
From descriptive analytics to AI: What's your flavor?
With all new technologies or methods, there is generally a gap or lag between what is theoretically possible and what finds its way into commercial practice. We asked insurers and reinsurers what forms of analytics they were deploying out of a possible six options:
As we can see, every form of analytics has attained at least a modest level of penetration, and we can tentatively construe from this an adoption curve of different analytics formats running roughly from predictive, descriptive and diagnostic (high degree of current adoption) through to behavioral, prescriptive and machine learning/artificial intelligence. So, while most respondents have developed capabilities to describe and predict, only a minority have advanced beyond this toward prescriptive and AI capabilities.
The donuts above indicate the share of analytics work (as a proportion of the whole) being undertaken by respondents working in the areas of underwriting and claims – this is an intuitive way to compare the prominence of different flavors both within, and between, these two areas.
A larger proportion of the analytics work undertaken by underwriting respondents appears to fall at the early stage of the adoption curve (descriptive and predictive) and a smaller proportion at the later stage (moving toward machine learning and AI), when compared to respondents working in claims. This implies that claims either encourages more advanced analytics than underwriting – which may be oversimplifying things – or that, for whatever reason, it leads underwriting on analytics maturity.
See also: Why to Refocus on Data and Analytics
Join us for our next post, on digital innovation, where we talk about the rise of mobile and the many flavors of digital strategy. Or, if you'd like to access all 11 key themes straight away, simply download the full Trend Map free of charge.
For any inquiries relating to the Insurance Nexus Global Trend Map, this content series or next year's edition, please contact: Alexander Cherry, head of research and content at Insurance Nexus (alexander.cherry@insurancenexus.com)
According to IBM, the world generates 2.5 quintillion bytes of data every day, with 90% of the world’s data having been created in the last two years.While insurers, and most companies for that matter, have for a fair while had more data than they’ve known what to do with, analytical and machine-learning models are now sufficiently mature and sophisticated for them to start reaping the much-heralded rewards of the big-data revolution. This is not without its challenges, though, with silos and legacy systems in particular acting as a drag on innovation. Analytics is being deployed pretty much everywhere, and by everyone, in the insurance ecosystem, so this post covers:
- Analytics and data usage across various insurance ecosystem players
- Overall data strategy
- Issues with silos and legacy systems
- Contrasting flavors of analytics in use: descriptive, diagnostic, prescriptive, predictive, behavioral and, of course, AI!
"Causing the greatest stir out of all today’s analytics tools is AI, which stands to revolutionize the whole insurance industry over the next 2-5 years, from robo-advisers and chatbots through to claims automation and mitigating fraud. While analytics teams retain the greatest degree of oversight, AI capabilities are currently being embedded across the whole insurance organization." – Helen Raff, head of content at Insurance NexusAs we shall see, the leadership on many of these measures is provided by reinsurers. This is evidence of their driving the whole ecosystem forward, and in this they often take the lead over insurers. We also see this more generally; for example, the giants Swiss Re and Munich Re have been particularly active in accelerator-based innovation over the past two years. See also: 10 Trends on Big Data, Advanced Analytics Is your investment in/focus on analytics increasing? 84% of all respondents are increasing their investment in analytics. This conforms to the stats we presented in our earlier post on services, investments and job roles, where analytics was second only to digital innovation for increased carrier investment. Drilling down further into the responses of different company types, we see that similar proportions of insurers (82%), brokers and agents (76%) and technology partners (85%) are increasing their investment. Of interest is the clean sweep by reinsurers, exemplifying the leadership trend we pointed out.
Analytics has applications across all the major lines. Health and auto are two obvious examples given the ready availability of connected health devices and in-car sensors, which make data easier to capture and, as an extension of this, models easier to feed. This facilitates usage-based insurance (UBI), which we explore in more detail in our forthcoming post on Internet of Things, whereby actual living/driving habits inform policy prices (read ahead straight away by downloading the full Trend Map for free).
Analytics also has obvious applications for predictive maintenance and security in commercial, auto and P&C/general lines, particularly where valuable assets (like property) are in play. Analytics is also growing in home insurance thanks to the increasing prevalence of connected-home devices, with Berg Insight estimating that there were approximately 18 million smart homes in Europe and North America by the end of 2015.
"There will be much more data from structured and unstructured data sources in the future – a huge challenge! 'Past developments are a good representation of future uncertainty' will not be replaced but solutions with AI-tech (big data) in combination with smart data strategies will enable insurances to make decisions based on models and evidence." –Andreas Staub, managing partner at FehrAdviceIs your analytics strategy coordinated across your organization? The uses and advantages of analytics have been obvious for a long time, and we have seen analytics initiatives sprouting up in nearly every corner of the insurance business, from underwriting through to counter-fraud. An ad-hoc approach, often inevitable in the early days of a technology, quickly becomes unwieldy, and the benefits from coordination are substantial. It is encouraging therefore to see 57% of all respondents indicating that their analytics strategy is coordinated across their organization. The trend across our different company types is similar to the one we saw in the investment/focus question above – unsurprisingly, as coordination is vital to gain maximum value from increasing investment and focus, and often represents a large investment in and of itself. We thus see 59% of insurers coordinating, 54% of brokers and agents and 55% of technology partners, with reinsurers once more taking the lead (77%). Are you utilizing external data sources? Plenty of data is available for analytics use beyond that directly captured by insurance companies themselves, both publicly available (like social media) and for-purchase (from third-party aggregators). There is no clear trend across our ecosystem players on this measure, with 77% of insurers, 67% of brokers and agents and 81% of technology partners affirming their use of external data sources (reinsurers had a small though insignificant lead). Segmenting by region, we can tentatively identify Asia-Pacific as trailing on this measure, and our broader research and industry engagement does indeed suggest that the third-party data culture is less-well-developed here than it is in North America and Europe. That said, public sources of data remain available, from unstructured social media through to data generated/collected by incipient smart-city infrastructure (like in Singapore). More details to follow in our forthcoming regional profile on Asia-Pacific, or read on straight away by downloading the full Trend Map here. Do you have a formal data-governance strategy? Insurance companies are being borne along on an exponentially growing tide of customer data, which has brought data governance to the forefront of people’s minds; yet, as of today, only 57% of insurers, 51% of brokers and agents and 58% of technology partners possess a formal data-governance strategy. We expect this figure to rise sharply in the years to come. Reinsurers once again appear to lead (with 77% affirming the existence of a data-governance strategy). Are legacy systems and silos a problem for your business? Capturing data is only the first part of the story to building out an analytics-based business. In many cases, analytics and big-data projects within insurance companies come unstuck not because of a lack of investment or strategic focus but for more prosaic reasons: silos and legacy systems. If infrastructural bottlenecks strangle rather than feed analytical models, preventing them from operating at scale across all the relevant data pools, then the output will be etiolated and limited in use.
We asked respondents whether legacy systems and silos represented "somewhat" or "very much" of a problem for their businesses, and then created a "burden score" based on a weighted combination of these two figures. Insurers clocked up a burden score of 138, brokers and agents 103 and technology partners 105. (Reinsurers score 123.)
There are two key takeaways from this. Firstly, that silos and legacy systems are a problem for the entire insurance ecosystem. And secondly, that carriers are generally harder-hit (comparing insurers and reinsurers to the rest of the industry), which may well reflect their position as the central node of the industry into which all the other players feed.
From descriptive analytics to AI: What's your flavor?
With all new technologies or methods, there is generally a gap or lag between what is theoretically possible and what finds its way into commercial practice. We asked insurers and reinsurers what forms of analytics they were deploying out of a possible six options:
As we can see, every form of analytics has attained at least a modest level of penetration, and we can tentatively construe from this an adoption curve of different analytics formats running roughly from predictive, descriptive and diagnostic (high degree of current adoption) through to behavioral, prescriptive and machine learning/artificial intelligence. So, while most respondents have developed capabilities to describe and predict, only a minority have advanced beyond this toward prescriptive and AI capabilities.
"The rise of insurtech, the analytics explosion and the new face of insurance has created a birth of new roles and impact points across the industry. No longer is analytics and data relegated to just information technology and actuarial — we are now seeing it being integrated into the business culture and DNA of insurance organizations." – Margaret Milkint, managing partner at the Jacobson GroupAnalytics is a very broad category with applications across almost every part of insurance, from underwriting and marketing through to fraud and claims, as well as on the investment side of the business. For the sake of clarity, we have chosen to focus in on two areas of insurance work, underwriting and claims, to capture a snapshot of analytics maturity at the start and at the end of the policy lifecycle.
The donuts above indicate the share of analytics work (as a proportion of the whole) being undertaken by respondents working in the areas of underwriting and claims – this is an intuitive way to compare the prominence of different flavors both within, and between, these two areas.
A larger proportion of the analytics work undertaken by underwriting respondents appears to fall at the early stage of the adoption curve (descriptive and predictive) and a smaller proportion at the later stage (moving toward machine learning and AI), when compared to respondents working in claims. This implies that claims either encourages more advanced analytics than underwriting – which may be oversimplifying things – or that, for whatever reason, it leads underwriting on analytics maturity.
See also: Why to Refocus on Data and Analytics
Join us for our next post, on digital innovation, where we talk about the rise of mobile and the many flavors of digital strategy. Or, if you'd like to access all 11 key themes straight away, simply download the full Trend Map free of charge.
For any inquiries relating to the Insurance Nexus Global Trend Map, this content series or next year's edition, please contact: Alexander Cherry, head of research and content at Insurance Nexus (alexander.cherry@insurancenexus.com)
