If your company stakeholders are expecting immediate return-on-investment (ROI) from implementing AI systems, you should manage their expectations carefully. AI systems do provide massive ROI, but it’s possible that those results won’t be instantaneous. Read this article to learn more.
We’ve written at length about the training and customization required to fully implement an AI system. In short, AI systems don’t begin yielding return-on-investment (ROI) overnight. They require significant investments of time, labor, and financial resources before you see a dime of ROI. In this article, we provide details on how IPsoft implements an AI system, how you can convince your company stakeholders that such a project is worthwhile, and how to properly manage expectations around project ROI.
When Will The AI System Yield Results?
The training process could be as short as several weeks, or as long as three months, depending on your business goals. During this time, IPsoft and your technology team work together to upload an assortment of documents designed to enrich Amelia, our proprietary cognitive AI system, with new knowledge. This could be as simple as your company’s business vocabulary or employee handbook, or it could be as complex as all of your company’s historical customer support data. You’ll provide IPsoft with your industry-specific semantic network, and any logic framework you’d like Amelia to follow in her decision-making. You’ll also upload pre-trained classifiers and analytic modules that will help her to conduct business according to your industry’s requirements and regulations.
From there, the two teams will work to auto-annotate Amelia’s conversational framework so she can understand utterances and common phrases — essentially, the teams help Amelia speak as humans do. This allows her to complete user interactions, regardless of the topics she might ultimately discuss or how users respond. During this process, Amelia will actually advise the teams on the best technique to help her learn and understand user conversations. You’ll also work with IPsoft to teach Amelia to follow standard operating procedures, so she becomes intimately familiar with how your business operates.
At this point, Amelia is ready to begin performing test interactions with your employees. We recommend setting up a Super User group to help test and run quality assurance on Amelia’s interactions. The Super Users will ask Amelia common questions and try to get her to solve standard issues. They may even try to stump her to see just how far she can take a process before requiring assistance. If Amelia passes this test, she’s ready to begin interacting with customers or internal users.
“Amelia needs training to perform her work, analogous to a new employee,” said Georg Huettenegger, IPsoft Implementation Architect. “We need to integrate with all the relevant back-end systems and tune her for the company in question. That takes time and some effort. This is crucial to deliver real business benefits.”
The First Stages of AI Systems-Driven ROI
Once Amelia is ready to begin handling queries on her own, she will inevitably get stumped by unique or custom questions and issues. This is where her ability to learn on her own and work with human employees comes into play. If Amelia is unable to resolve an issue, or finalize a transaction, she will escalate the conversation to a human agent. As this happens, she will stay connected with your employees to determine how the issue was resolved. She will then learn from her human colleagues, applying that knowledge to any similar conversations so as to avoid unnecessarily escalating future interactions.
This means that the early stages of implementation will likely only produce ROI for the basic cases for which you’ve trained Amelia. However, as Amelia continues to monitor, learn, and process information, she’ll be able to widen her scope to take on more complicated tasks.
During this time, your company will realize ROI on several fronts: 1) Amelia takes on tasks typically conducted by human workers 2) Amelia is online 24/7, so your company can handle massive volume 3) Your customer satisfaction improves and customers are recommending your service to friends and 4) Your human workers are now focused on money-making or unique problem-solving tasks.
The Second Stages of AI Systems-Driven ROI
Now that Amelia has proven her value in a sole use case, your company may want to think about expanding her purview. If Amelia was hired to handle account verifications, password resets are the next logical task. Once she masters password resets, she can be trained to start recommending products.
Training Amelia to handle secondary use cases requires less time and effort than training her for her first use case. IPsoft can help clients train Amelia for secondary and tertiary use cases, but the process is so much simpler the second and third time around that we’re not required to be part of the training process.
The more use cases in which Amelia takes part, the more she’ll be able to handle complex and multi-step issues. Whereas in the first stage of your implementation Amelia handled only isolated use cases, in the second stage she’s able to take on secondary and tertiary ones, all in one interaction. She can now deliver ROI across customer service disciplines and at scale.
ROI is now being generated in the following ways: 1) Amelia takes on the roles of human workers on multiple fronts 2) She can handle multi-step processes on her own, without escalating to human workers 3) She can recommend products and services to customers and 4) Even more human workers can focus on money-making or unique problem-solving tasks, now that Amelia is doing more for the company. Because training Amelia for secondary use cases requires less time and effort, you’re achieving ROI much sooner than you did for the initial use case.
One other important point to note: Amelia never forgets her interactions with customers. When she interacts with a customer a second or third time, she brings the historical information from those interactions to the new conversation. Here’s why that’s important: If Amelia assisted a customer with account verification back when she was able to handle only one use case, she still has that account information. So, when the customer attempts to reset his or her password, Amelia doesn’t need to repeat the entire account verification process. She remembers the interaction, resets the password, and moves onto a new task. Amelia is not only capable of taking on more tasks, she’s also improving the speed at which she accomplishes them thanks to her information recall.
At this point in your AI system deployment, the hard work is done, and the focus should be on repeating this process for additional use cases that will solve business problems. Success with Amelia’s initial use case will help set expectations within your company about the kinds of ROI Amelia delivers. Clear communication on Amelia’s ability to generate ROI in stages as she takes on tasks — first in an initial use case, then in other use cases with multiple tasks and procedures — can help manage those expectations going forward. With this step-by-step approach to implementing use cases for ROI, and by setting the appropriate expectations, your company can reap the rewards of digital labor, turning human workers who have historically handled menial tasks into change agents and revenue generators. We can’t promise it will be easy, but we do promise that it’s achievable and manageable.