Seven Tips for Evaluating Virtual Assistants to Automate Employee Self-Help

eWEEK DATA POINTS: Virtual assistants help provide IT professionals with a better way for routing critical incidents 24/7, lower costs, reduce errors and relieve major stress on the help desk.


During the past several months, IT automation has gone from a “nice to have” to an imperative for enterprises to stay productive. CIOs are now focused on how to prepare for employees returning back to the office, while also providing assistance to those still working from home.

This next phase will entail staggered and limited access to facilities, temperature checks, PPE and social distancing measures, and requirements for automated, self-service experiences. Recent research shows that 59% of Americans agree the COVID-19 pandemic will lead to an acceleration of automated workplace technologies within the next year.

In the current “work from anywhere” norm, IT leaders need solutions to redefine how employees get help, no matter where they’re working. Help desk automation such as virtual assistants have specifically eased the stress on IT teams and their leaders who, at the start of widespread work-from-home mandates, experienced some of the busiest weeks of their careers. 

Virtual assistants help to provide IT professionals with a better way for routing critical incidents 24/7, lower costs, reduce errors and relieve major stress on the help desk. An AI-based virtual assistant especially can enable organizations to reduce their budget through automation, shorten SLAs on critical issues and improve employee productivity to support the changing workforce. Employees also expect conversation-level speed from such technology, but achieving this is not a small feat.

In this eWEEK Data Points article, Fran Fernandez, chief product officer for enterprise service management (ESM) company Espressive, offers tips for helping enterprise organizations seeking to implement virtual assistants for workforce productivity and satisfaction. Fernandez outlines a checklist of functionalities to consider when implementing a virtual assistant to improve the employee self-help experience.

Data Point No. 1: Does the virtual assistant deliver consumer-like experiences?

Today’s consumer mentality around increasingly personalized service and instant gratification has bled into the workplace. Employees are refusing to use traditional tools to get help at work when they have Alexa-like ease of use and immediacy at home.

Research from AppDynamics and Google shows that 80% of users delete apps due to poor performance and 53% of mobile users abandon sites that take more than 3 seconds to load, respectively. Further, an ITSM  survey reported 46% of IT pros said that choosing the right form on an employee portal for help is confusing, and 35 percent said knowledge base articles can be too technical and often out of date, making employee adoption of traditional tools low.

When adopting a virtual assistant, be sure it delivers what employees expect, or it won’t be adopted. Answers need to be personalized based on location and equipment used—for each employee. The virtual assistant must also connect employees with the appropriate expert, rather than employees being sent elsewhere for help and updates.

Data Point No. 2: Does the virtual assistant deliver workflow-based experiences?

Topics like password reset and laptop refresh deserve more than an answer; they require a workflow process to guide employees every step of the way versus requiring employees to fill out a form that they don’t understand. Plus, the virtual assistant has information about the employee so that the workflow can be more of a personal conversation. In addition, the workflow should be able to be completed within the virtual assistant app. That way, employees can give feedback to the workflow and even ask additional questions about it.

Employees should be able to initiate workflows proactively. However, workflows should also be available via push notifications that remind employees that actions need to be taken. If your ITSM tool already has workflows, your virtual assistant needs to be able to leverage them.

Data Point No. 3: Does the virtual assistant deliver easy accessibility for employees?

The virtual assistant needs to be both intuitive and accessible to deliver a true, consumer-like experience. IT leaders shouldn’t assume that employees will easily break old habits when seeking help, such as emailing or calling the help desk. The virtual assistant needs to be accessible to employees from wherever and however they work in order to deliver an easy and efficient way to receive answers and updates.

Virtual assistants should be made accessible on any desktop or mobile device as a native app, deflecting employee issues across a number of different interfaces including email, phone, service portal widget, collaboration tools such as Slack or Microsoft Teams, and native apps. Additionally, virtual assistants should engage with employees even when employees call the help desk to allow IT to focus valuable resources on critical initiatives. When an employee calls for help, the virtual assistant should be there to answer questions.

Data Point No. 4: Does the virtual assistant have a pre-built natural language processing (NLP) model that already understands what employees are asking?

Not all virtual assistants are created equal. Organizations should ask: “What is the time to value for a virtual assistant?”

When implementing a virtual assistant, IT leaders should consider how many employee phrases it can understand on the first day of use. Those phrases should cross all enterprise departments because, even if you roll out the assistant just for IT, employees will ask HR and payroll questions (and more). If a virtual assistant doesn’t understand phrases and topics out of the box, enterprises will need AI experts and linguists, plus a lot of time to get it up and running. A less than stellar experience with the virtual assistant (e.g., the virtual assistant doesn’t understand what employees are asking) will have employees turning back to old habits—emailing and calling the help desk.

It's also not just about understanding phrases. The virtual assistant should come prepopulated with questions and answers for industry wide topics such as Salesforce, Zoom and G Suite. The IT service desk team shouldn’t spend its time answering questions that are the same across all customers.

Data Point No. 5: Does the virtual assistant deliver integration of knowledge articles?

If an enterprise has a robust knowledge base, they’ll want to be able to integrate that content with its virtual assistant. However, if the experience of portals is simply replicated, where employees are served a list of articles to select from, employees will not continue using the virtual assistant.

Virtual assistants, however, should only turn to knowledge base resources when a conversational answer is not available. Machine learning is a key component that enables virtual assistants to build a more accurate search of the organization’s knowledge database while learning from every employee question and answer as it goes.

Data Point No. 6: Does the virtual assistant deliver intuitive and fast updates?

If an organization must rely on IT service management agents or the vendor to update content, its virtual assistant will always be behind, leading to poor employee adoption and higher call volume. IT leaders need to ensure their virtual assistant enables subject matter experts from any department to easily and quickly update content—without the requirement for technical expertise.

SMEs across the enterprise should be able to edit, create and delete virtual assistant content and also set responses to occur for specific periods of time based on role, location, and other factors. It’s also useful if virtual assistants can match answers to a new FAQ response when employees are asking a related but different question. Answers should be able to be previewed before going live.

Data Point No. 7: Does the virtual assistant deliver next-gen artificial intelligence that enables it to continuously learn?

Virtual assistant toolkits and platforms require organizations to hire an ongoing staff of linguists, AI experts and data scientists. Even after long deployment timeframes, the virtual assistant may still struggle to understand what employees are saying. When that happens, employees won’t come back; virtual assistants need to speak the language of employees.

Virtual assistants should come with predefined entities through advanced NLP. It should be able to differentiate between various service tickets, such as IT incidents and IT service requests. Virtual assistants must also have neural networks that allow them to learn new phrases and differentiate between similar phrases with different meanings. As we all know, the same question can be asked in countless different ways and a virtual assistant should be able to identify employee language patterns. Through advanced machine learning, virtual assistants should learn more about generic employee language, improve its understanding, and then share this knowledge with all customers.

Summary: Choose wisely

It’s crucial for organizations to ensure that their AI-based virtual assistant will enable them to dramatically reduce calls to the help desk, reducing MTTR and improving the employee self-help experience by answering questions and resolving problems automatically, or involving an expert when that is needed–all within the app. Write your virtual assistant RFP with these seven considerations in mind, and your virtual assistant initiative will have better success with increasing employee adoption and decreasing IT help desk call volume.

If you have a suggestion for an eWEEK Data Points article, email [email protected].