Skip to main content

CX leaders know artificial intelligence must be part of their technology strategy—if for no other reason, to gain competitive advantage. But “artificial intelligence” is somewhat a loaded term. It encompasses a variety of applications, and it may be challenging to convince internal stakeholders of AI’s value.

That’s why it’s important to treat the business case development in a structured and methodical way. It’s different than building a business case for, say, video conferencing. Most people have used the technology and have an inherent understanding of its value. With AI, that’s not always the case—and sometimes, the opposite is true. They have used it, but had a bad experience.

Metrigy recommends to its clients that they follow these steps to building the business case—and allow enough time to do it right.

  1. Identify business metrics
    Before even exploring what you’re going to do, explore why you’re even considering it. Don’t implement AI just because everyone else is doing it. There should be a business goal at the heart of the project—and I’d be hard-pressed to find any company that doesn’t have a CX-related business goal that AI can’t address. Are you trying to generate revenue, save operational costs, improve customer ratings, increase employee productivity? Something more specific, such as First Call Resolution (FCR) or Call Handle Time (CHT)?
  2. Identify problems and opportunities
    In discussing the business metrics, the conversation will naturally turn to the specific problems or opportunities AI can address. For example, let’s say the business metric you want to improve is customer ratings, because they’re lower than competitors and you’re losing business as a result. Now you need to drill down and figure out what’s causing that problem before you can suggest a solution. That may mean holding some customer focus groups, conducting post-call, custom surveys, or evaluating unstructured, open-ended comments. Let’s say in this case, the issue is that the customer cannot find answers in the company’s self-service knowledge base and it’s difficult to escalate to a live agent. So, the business metric is customer ratings. The specific problem is to address a poorly functioning self-service knowledge base.
  3. Connect the dots to AI
    With the business goal and problem or opportunity defined, it’s time to determine what technology will be best. Business-technology liaisons are great resources in this stage. These are people who understand the business issues and the available technologies—and enough knowledge to figure out what to pair together. (Register for this webinar to learn more about the specific flavors of AI.) In our example, a few key AI technologies maybe effective:
    – AI-enabled virtual assistant to interact with customers trying to find an answer in the knowledge base, with strategically placed hooks linking to live agents when the virtual assistant can’t direct the customer to answers.
    – AI-assisted intelligent routing to route frustrated customers to the best possible agent at that second-in-time for the issue at hand. This will increase the likelihood of a respectable FCR, which will help meet the business goal of improved customer ratings.
    – Predictive analytics to build a machine learning knowledge base that can predict what customers will do next, what the AI virtual assistant should recommend, or when to escalate to a live agent to improve how they rate their experience.
  4. Estimate costs and return on investment (ROI)
    There is no shortage of technologies that companies can implement with an unlimited budget. Since that’s never the case, you’ll need to estimate costs and ROI prior to making your case to stakeholders and evangelists. In our research, we have found that the most successful companies (based on measurable before-and-after figures on revenue, cost, customer ratings, and/or agent productivity) are spending 4% of their revenue and roughly $6,000 per employee per year on CX technology (contact center, CRM, and other customer- or agent-facing apps), about 13% of which goes to AI applications.
  5. Solicit evangelists, stakeholders
    Armed with the “story,” now is the time to identify key people within the company that need to be part of the project. Stakeholders have budget, while evangelists have influence. You’ll need both for a successful project. Explain to them the business goal, the problem or opportunity you’re trying to address, the recommended technology, and the financials: “If you fund this project for $3 million, we expect customer ratings will increase by 53%, generating an additional 4.2 million in revenue within 18 months.” Make sure to address the previously identified business goals—and tie it to revenue whenever you can.
  6. Educate employees and customers
    Many companies fall flat here. The reason? IT staffs generally run these projects, and they are not by nature marketers. But in order for a project to be successful, invest time and resources in explaining to employees and customers what change is coming and why it will make their lives easier (vs. how to use the technology). Evangelists are crucial to success at this stage to intrigue the ultimate users of the technology. Because of their track record and influence, employees and customers pay attention to the impact of the change and become excited to leverage the technology if explained right.
  7. Evaluate technologies and providers
    Parallel with the education phase, the IT staff should be evaluating the specific technologies and associated providers. Cast a wide net and don’t simply sign with the providers you already use (though that may be what ends up being the best option). Conduct a Request for Information (RFI) to effectively compare offerings from the different providers to see which is best matched to your goals.
  8. Select technology provider
    Once the RFI is complete, narrow the wider pool to a short list. Then conduct a more detailed Request for Proposal (RFP). After a detailed analysis, select the provider.
  9. Measure baselines
    Prior to starting any implementation, determine what success metrics you’ll measure—and then measure the baseline. What are the figures today, prior to the implementation of the new technology. So, for example, maybe the CSAT score is 43%, while FCR is only 40%. Then determine the intervals you’ll re-measure the metrics.
  10. Implement the technology
    Start the technology implementation, with regular updates on the progress to stakeholders.
  11. Conduct training
    Train employees on how to use the technology, and how it will change their processes. Create FAQs and how-to videos for customers so they will understand how to use the technologies.
  12. Measure success and revise
    Based on the previously determined metrics and intervals, regularly conduct new measurements to document success or failure. If the figures aren’t as high as you’d like, make adjustments to how you’re using the AI applications or technologies. By documenting success, you’re more apt to secure funding from the same stakeholders for the next project.
Robin Gareiss

Robin Gareiss is CEO and Principal Analyst at Metrigy, where she oversees research product development, conducts primary research, and advises leading enterprises, vendors, and carriers focusing on customer experience and engagement, digital transformation, and contact center.