Starting in 2019, enterprises will continue to realize ROI from their AI investments but they will change the way they measure those outcomes.
This blog is part of our ongoing series, IPsoft’s 2019 AI Trends, detailing what we believe will be the dominant developments and movements in the Enterprise AI market next year. These blogs will be published regularly through the end of the year.
In 2019, enterprises will continue to realize ROI from their AI investments but change the way they measure those outcomes. There will still be an emphasis on cost and operational savings, but productivity improvements, customer experience and innovation will assume higher importance. Enterprises will also start to measure ROI in ways that are unique to their vertical industry, depending on their strategic and competitive goals.
In addition, as enterprises expand their AI implementations outside of use cases for customer service and IT helpdesk, there will be a greater number of stakeholders involved who will develop their own ROI goals. AI will become more prevalent throughout the enterprise, with improved scalability and autonomy being ultimate measures of success.
Balancing Innovation and Cost Savings
As the AI market continues to evolve, there is ongoing skepticism from some observers about AI’s ability to deliver ROI. Some enterprises have faced challenges in identifying the relevant business use cases or processes in which to apply the technology. And they also need to factor in that onboarding staff to AI, training an AI solution and adapting processes all take considerable time. However, in a recent MIT study, 81% of AI pioneers stated that they have used AI to increase revenues and increasing their investments, so it’s clear many enterprises see AI as worth the effort.
Enterprises need to find the most appropriate and impactful business cases that will allow them to transform quickly while also containing costs. Generally, use cases with complex processes that affect a large number of people, such as cutting IT support request volumes or improving and optimizing staff onboarding will deliver the greatest benefits. In 2019, use cases will become clearer and easier to grasp, partly as the result of more companies watching what their peers and competitors are achieving with AI and attempting to recreate their successes. A larger number of enterprises will be focused on using AI for innovation and revenue growth, which will lead vendors to introduce new packaged AI solutions that clearly map to existing industries and roles, such as conversational banking. This will ensure faster ROI as well as making AI easier to measure along various dimensions.
Enterprises Initially Stick with Tried and Tested Metrics
As detailed in Everest Group’s recent research, Conversing with AI — Intelligent Virtual Agents (IVA) Market Report 2019, enterprises have adopted AI most often for IT, helpdesk and contact center use cases. In these scenarios, enterprises are measuring their AI solutions using familiar metrics. For example, in contact centers, cognitive agents that are used to answer customers’ support queries are measured against standard call center metrics that apply to voice interactions around efficiency and satisfaction, such as:
- Response time – How quickly the virtual agent responds to customer queries.
- Average handle time – The time spent messaging with a client in one session.
- First contact resolution – Ability of the virtual agent to resolve an issue during the first interaction.
- Customer satisfaction – How satisfied the customer was with the interaction and whether their issue was resolved.
- Escalations – The number of times the virtual agent was unable to answer a question and transferred the query to a human agent.
Many of these same metrics also apply when the virtual agent is used to resolve IT helpdesk queries or act as a whisper agent. Additional metrics include:
- Time to resolution – How quickly an IT issue can be resolved by the AI system.
- Employee satisfaction – How satisfied the employee is with the interaction and issue resolution, typically measured using a survey.
- Employee productivity – Whether down time spent waiting for information or broken systems to be fixed is reduced.
However, the virtual agent is not directly comparable to existing processes and human support, and new metrics will emerge over time.
ROI Will Become Easier with Experience
Many AI early adopters have changed the way they measure their AI implementations as they gain experience with the technology. For example, a leading gaming company that implemented Amelia, IPsoft’s cognitive agent, for customer service and account verification found that average handle time for agent-led customer service interactions increased. Typically contact centers try to reduce handle time to decrease costs. However, the gaming company found that with Amelia handling simple requests and eliminating phishers, calls with human agents became higher-value interactions with longer conversations. In the long run, building relationships with customers in this way helps with long-term customer loyalty. The company also found that by using Amelia for account verification, it was able to improve security and adherence to its processes.
In addition to new metrics emerging, the ROI for AI will become easier to measure once enterprises and employees gain experience with implementing, training and using the technology. Those that are willing to learn, grow and adapt with new AI platforms have a distinct advantage as digital-human hybrid workforces become more commonplace.
Three Keys to Success: Adoption, Scalability and Autonomy
As enterprise leaders achieve success with initial AI use cases, they will expand to others and generate further impact. The more widely the solution is used by employees and customers, the greater the benefits for the enterprise. In 2019, as enterprises shift to wide-scale AI implementations, three key measurements can help ensure ROI is achieved:
- Adoption. A solution should be relevant to a large user base, whether customers or employees. A larger number of users will ensure a far-reaching impact on efficiency and productivity.
- Scalability. A cognitive AI agent or automation platform should be able to learn and complete as many tasks as possible, with high accuracy and resolution rates. While an implementation may start small, a solution that can be expanded to resolve additional issues or processes across departments will set the business up for long-term cost savings.
- Autonomy. Any true AI solution needs to be able to take action and resolve issues without human input, as long as it has been trained on processes and given permissions to do so. Productivity will be significantly improved if a virtual agent can take on tasks that were previously completed by humans or make processes simpler.
Vodafone is an example of an organization that has realized success with AI by focusing on these three factors. In 2019, Vodafone is expanding its implementation from IT support into new departments starting with HR processes. It has a goal to provide all employees with access to Amelia to resolve any issue — from IT to HR to finance. Currently for IT support, Vodafone has achieved an adoption rate of 58%, which is the number of employees who have contacted Vodafone’s IT service desk and used Amelia. The solution has been scaled to understand 56 different IT service desk requests, and autonomy is at 53%, which means Amelia handles 53% of all chats without human intervention (the goal is to reach 65-70%).
During 2019, as enterprises continue to expand their AI investments, they will review the potential overall impact on their business, employees and customers, and chart out new ROI goals accordingly.