Set up in January 2016, Spixii is an automated insurance agent dedicated to making insurance quicker, easier and more personal. The agent predominantly operates through websites, existing messaging applications (WhatsApp, Facebook’s Messenger, etc.) on smartphones and focuses on guiding customers through the insurance buying process. Based on artificial intelligence, Spixii’s white-labelled chatbot learns from and suggests personalisation for each customer interaction.
As a matter of fact, dissatisfied customers are an inescapable actuality of providing a service or product. When it comes to insurance, however, dissatisfaction goes well beyond that: According to Engine, a service design consultancy, 28% of customers surveyed deem the insurance industry to be the worst for customer service. Without a doubt, dissatisfaction with the service of an incumbent does not only hamper loyalty but also a company’s reputation: A survey commissioned by Zendesk, for example, suggests that 95% of customers will share their negative experiences.
Renaud Million, CEO & co-founder of Spixii, shares this perception: “Customer experience is really poor through the lifecycle of the insurance product from purchase, renewal and claims. This poor experience leads to low satisfaction yet insurance empowers people and businesses but it is not perceived as such.” – One of the reasons Spixii was founded.
Indeed, we believe ‘digital’ has an enormous potential to overcome many of the traditional pain points in the insurance service experience, namely pain points like feeling under-appreciated due to generic processes, being passed around multiple agents and being kept on hold for too long – just to name a few. Against this background, chat bots, driven by Artificial Intelligence (AI) are increasingly offering opportunities to enhance the service experience. A chat bot is a software programme that can communicate with humans through various channels (chat, phone, etc.) using AI/ Natural Language Processing (NLP). As an independently operating tool, they enable a streamlined service, 24/7 availability and rapid response times. What’s more, if chat bots are deployed to guide customers through their transactions with an insurance company, they can help in making information easier to find and digest. Taking into account that, according to Forrester research, 53% of customers will abandon an online purchase if they cannot find a quick answer to their questions, they present a powerful tool to increase online conversions. Lastly, chat bots can streamline tedious, repetitive processes and help shift customer service roles away from routine tasks to customer interactions that require deeper, personalised insights and analysis, thereby improving the efficiency of service.
@Renaud, can you quantify the effect that Spixii can have on the business of an insurance client?
“Quote & buy chat bots outperform the current web-forms by up to 30% depending on the product and brand of our clients. The same magnitude can be observed in performance improvement for servicing and claims, for example by saving a considerable number of hours to call centres while delivering superior customer satisfaction. Higher benefits can be observed for clients with no digital capabilities. What is more, our chat bots provide valuable data with actionable insights. Such insights are often unexpected for clients solving problems that could not be solved efficiently before these learnings. In other words, Spixii is helping our clients step closer to their customers through this knowledge”
However, we have seen chat bots failing customers, by simply not being smart enough to digest the variety of requests and human language. Another common problem of (premature) chat bots is routed in the fact that customers felt they were confronted with a ‘machine’ and processed in a ‘production line’ manner – exactly not the individualised personal service experience insurers should be striving for. While AIs get increasingly better at processing human language, they are also improving at imitating human behaviour – like the demonstration of google duplex has recently shown quite impressively.
@Renaud, Spixii claims to ‘make insurance more human’, how do you do this?
“We make insurance more human by designing customer-centric, conversational user interfaces. Designed to for personal interactions with a particular attention to detail. These interfaces are part of the ecosystem of technology integrations and strategic services we developed with our market-proven methodologies, validated with customer feedback. Focus on customer development, our chatbots start and end with customer.”
Looking at the implementation of technologies like chat bots we have experienced several hurdles with our clients: After choosing a specific area of deployment and the respective technology provider, insurers have to look at a variety of technical integration issues, e.g. regarding front ends, data storage and policy administration systems. Before actually going live, the chat bot often needs substantial training on actual data of the insurer. For example, rules need to be refined to enable the chat bot promptly escalate necessary issues (sensitive information, complex queries, fraud, unhappy customers, etc.) to a human representative equipped to handle them.
@Renaud, what are the system requirements for insurance clients to use Spixii?
“None – every client is unique and we pride ourselves on our ability to adapt around their existing set-up and legacy systems. Our end-to-end chat bots are integrated into clients’ backend, which can be heavy or light depending on the use case. The Spixii discovery roadmap identifies the best first use case to explore the benefit of chatbots, aligning such deeply transformative technologies with our client business objectives and strategy without forgetting the value to the end customer.”
What does the implementation process from agreement to go-live look like?
“Blending agile scrum frameworks with Spixii’s own approach to design, an end-to-end chat bot can be designed and launched between 6-12 weeks. This includes connecting to back-end rating engines and claims systems in addition to websites or other channels like Messenger. It can be shorter for simpler use cases with limited back-end integration like sending completed conversations via email instead of being connected to legacy systems. Once Spixii is live, iterating new digital customer journeys in response to customer feedback or behavioural insights, takes an incredibly short time. Ultimately, it empowers companies to switch from a static insurance model to a dynamic one.”
How do you safeguard customer data?
“We have policies in place that we share with clients while going through our clients’ security questionnaires. We consider this topic of utmost importance and put in place relevant certifications applicable to our company and technology.”
Undeniably, insurers have been experimenting with the use of chat bots for several years now. A prominent example being the digital insurer Lemonade, who announced that it is able to pay out claims within 3 seconds, leveraging an AI-driven chat bot. In the case of Lemonade, which is attracting a young tech-savvy crowd, and which is set up and designed for the digital era, the advantages of chat bots are obvious. However, for traditional insurers struggling with legacy systems and insuring potentially a large variety of target groups, the way until a chat bots is eventually improving the customer service can be a long one. Our experience shows, that the gains still can be tremendous, when keeping 3 core success factors in mind: Firstly, the deployment and potential use cases need to be carefully assessed and validated with the target customers. Generic chat bot interactions will drive customers away, particularly if these are laboured and result in extra effort. Secondly, it is important to include fall-back procedures that shift to human support when necessary – while ensuring consistency across the contact points. Finally, it is imperative to keep the application of chat bots simple. The artificial intelligence systems are still a long way from ‘plug and play’. Hence, the deployment should be initially restricted to discrete, specialised and defined capabilities (e.g. Quotation, FNOL), before they can eventually be rolled out across the entire value chain.
Dimensional, “A survey of customer service from mid-sized companies”, 2013
Engine, “Annual customer experience survey 2017”, 2017
Forrester, “How AI will transform customer service”, 2017