Menu
Chatbots have taken the center stage as AI wordsmiths replacing IT Service Desk support agents. The purpose is simple: facilitate service requests for end-users without having to rely on human support agents.
And it makes sense in the digital world where products and services are technology driven, and any IT incident directly impacts end-user satisfaction and customer loyalty.
However, Chatbots are also seen as an automation dilemma: when the chatbot fails to correctly identify and help resolve a ticketing request, your customer satisfaction score plunges even before a first interaction with your IT Service Desk agent.
So what exactly makes chatbots frustrating for end-users?
In this blog, we will review the underlying cause of a dysfunctional chatbot support system:
The first step is to understand the disconnect between technology and the end-user. Chatbots are programmed to deliver generic responses that are optimized for specific keywords. The response is often static: without adequate information and context, chatbots repeat the same (and often, incorrect) response unless the end-user is able to provide information exactly as required by the automation program that powers the chatbot communications. In practice, users may not be able to use the right keywords and ticket categories, and so the chatbot response is inherently limited by the scope of intelligence that powers it.
The IT Service Desk in particular is considered as a contact center that connects end-users with resolution teams – each acting independently. The result is a rapid increase of repetitive ticketing requests, frequent ticket hops between resolution teams and a slow resolution process.
Even when organizations transform their operational workflows and adopt the principles of a modern Agile and DevOps organization, traditional ITSM technologies provide limited support and functionality to achieve the goals of continuous improvement, collaboration and automation.
In order to maximize the ability of a chatbot system to interpret a service request and deliver an optimal response, chatbots need to be well aware of the context. While many chatbot systems rely on end-users to provide sufficient relevant information, the technology should explore other avenues to extract the most relevant information and shift-left the resolution process with limited end-user engagement. Context is often available within historical ticketing archives, network log big data and ongoing patterns of ticketing requests.
In practice, this context is more useful in identifying a problem’s root cause or solutions as the request lifecycle follows common patterns, and solutions may already be known to the IT Service Desk.
Every ticketing request potentially generates new pieces of information that can help the Service Desk understand the performance of their IT environment and applications, as well as patterns that lead to potential issues. These patterns should be learned on an ongoing basis and the IT Service Desk should adopt a mechanism of proactively conducting risk mitigation efforts.
As part of a hyperautomation intelligence strategy, these efforts require close collaboration between the IT Service Desk, engineering teams as well as business decision makers. Repetitive issues can be attributed to specific technology choices that are made from a business and operational perspective. For instance, a cost-effective proprietary solution may be exposed to compatibility and security issues, and any lasting resolution may require a shift to open source alternatives. Therefore, hyperautomaton intelligence is not only about connecting a ticket request to the right solution, but also about providing a long-term resolution.
While chatbot technology is adopted to save cost, time and efforts required to resolve overwhelming volumes of ticket requests, the associated cost and value metrics are still in their infancy. It’s not always clear exactly how to measure the true cost performance of a chatbot system. IT Service Desk needs to evaluate whether the technology is actually saving cost and improving support efficiency, or frustrating users, increasing the volume of ticketing requests with incorrect categorization, or making the entire support pipeline more complicated. This issue becomes more challenging due to the lacking mechanism to accurately quantify intangible KPIs such as end-user sentiment and satisfaction, brand loyalty and the additional workload required to resolve requests. In this context, advanced NLP technologies should be adopted to learn user behavior and chatbot performance across every customer interaction.
Chatbots and automation are only one part of a holistic hyperautomation strategy that is focused on the technology, process and people.
At its core, a chatbot solution is designed to perform tasks that follow predefined workflows. Most organizations adopt ITSM frameworks such as ITIL to guide their operational workflows. An effective hyperautomation intelligence strategy builds on the ITIL principles of focusing on value, iterative improvements based on end-user feedback, and continuous optimization.
This is where hyperautomation intelligence technologies can do the heavy lifting, identifying the operational bottlenecks and finding opportunities for effectively replacing repetitive manual tasks with automation. The chatbot technology then works as an interface to a wider intelligence platform that accounts for the processes and people across various stages and functions of the IT service delivery lifecycle.
The modern ITSM organization already collects vast information on customer interactions through the IT Service Desk. ITSM solutions are well placed to execute the operational workflows within the ITIL framework tailored to meet their own unique business needs. The next step is to embed intelligence into the process and build chatbot interactions based on insights from the user, network and service performance, existing ticketing requests and a range of information sources most relevant to the service request.
