paint-brush
New CX Benchmark Report Highlights the Hidden Costs of Poor AI Integration in Businessby@ehecks
169 reads

New CX Benchmark Report Highlights the Hidden Costs of Poor AI Integration in Business

by Eleanor HecksJune 21st, 2024
Read on Terminal Reader
Read this story w/o Javascript
tldt arrow

Too Long; Didn't Read

A new report shows that not every use case of AI is equally beneficial. Decision-makers can anticipate the best results by understanding their organizations' most pressing needs.
featured image - New CX Benchmark Report Highlights the Hidden Costs of Poor AI Integration in Business
Eleanor Hecks HackerNoon profile picture

Artificial intelligence has become a buzzworthy topic and — as I’ve recently discovered by adding it to my workflow — some of the excitement is warranted. I use AI tools as a web designer and writer to complete routine tasks like sending marketing emails, creating content, and answering repetitive questions. These applications save time and increase my adaptability during pressure-filled days.


However, a new AI report also reminded me that people must be thoughtful about integrating the technology. Not every use case is equally beneficial for every worker or identified need.


While AI works well for me in many situations, business leaders should be optimistic yet cautious when trying it. Decision-makers can anticipate the best results by understanding their organizations' most pressing needs and determining how — or if — AI could address them.

New Benchmark Report Highlights AI Use Also Has Its Downsides

One of the downsides of the artificial intelligence boom is that decision-makers may rush into using it, feeling pressured by peers who have already adopted it in their organizations.


Forethought is a company specializing in tools that automate parts of the customer experience process. It recently released its first-ever “AI in CX Benchmark Report.” Many of the study’s takeaways show that specific choices significantly affect how well artificial intelligence works for companies.


The report examined cost-per-resolution, defined as the average cost of resolving customers’ problems. When companies trained AI tools with historical internal data, they were almost three-and-a-half times more likely to reduce those expenses.


Although internal historical training data resulted in the most cost-effective AI tools, in-house development of those resources resulted in 77% customer satisfaction, the lowest of all methods studied.


Deflection rate was another critical metric in this study, and it represented the percentage of customer needs solved by self-service resources rather than support agent contact. Tools trained on a company’s data gathered over time achieved a 37% deflection rate. In-house-developed products showed the worst net promoter score (NPS).

4 Best Practices for Implementing AI Wisely

The training data and chosen tools substantially affect how much artificial intelligence pays off for the companies using it. Besides paying attention to these factors, decision-makers can do other related things to improve their odds.

1. Personalize Customer Interactions

A notable advantage of combining internal and historical data and artificial intelligence is that companies can provide responses targeted to the individual customers contacting them. That’s a strategic way to increase loyalty, especially since research shows 71% of customers want such personalization, and 76% feel frustrated by not getting it.


Consider incorporating details such as someone’s favorite products, order history, or the number of interactions with the company during an average month to personalize the responses someone sees when engaging with an AI tool. Train customer service representatives to provide tailored service at every opportunity. Those efforts emphasize that companies are relevant to customers and their needs.

2. Prioritize Efficient Responses

AI tools can answer customers outside of business hours. In the best scenarios, they immediately provide the necessary information, eliminating the need for the person to speak to a person.


Take a tiered approach that gives people instant responses to the most straightforward queries and tells them when to expect answers to more complicated questions. Then, customers always get quick answers. Alternatively, not receiving a response could make people more likely to look elsewhere for the required product or service. Letting an AI chatbot handle out-of-hours needs is also an excellent way to grow a company’s customer base so it extends to other countries or time zones.

3. Design Tools for Different Groups

Company leaders thinking seriously about their AI integrations should aim to create various products or applications to address customer segments. Consider the example of a home health agency. It would likely receive communications from these groups:


  • People currently receiving the agency’s services
  • Those interested in hiring the company for themselves or loved ones
  • Individuals interested in working at the business
  • Concerned family members or other loved ones


Creating AI tools for each group keeps the interactions maximally relevant for everyone involved. Leaders from the Chinese e-commerce company Alibaba Group applied this tip when building five chatbots for people associated with the Taobao shopping platform, including one to train customer service representatives.


The latter tool has shortened training by more than 20%, supporting the 1,500+ workers using it daily.

4. Choose a Specific Process to Improve

Leaders must also determine which workflows they hope AI will enhance. Figuring that out ensures that organizations use the technology for well-defined reasons that support long-lasting success.


As a case study, take inspiration from a Nordic insurance company where leaders wanted to modernize the claims management process by reducing the manual, time-consuming responsibilities it required. Besides hiring external experts to guide this tech transition, decision-makers recognized the importance of keeping humans at the heart of the change.


The completed solution digitized and organized unstructured data, allowing workers to process claims data in near real-time. Additionally, the tool correctly interpreted 70% of the data fed into it, saving users time. Leaders should also select metrics before implementing AI and track how they change over time. Remember that people need time to adjust to new processes, so executives should expect gradual progress.

Learning From the Study’s Data and Best Practices

Much of Foresight’s study focused on the benefits of using internal historical data when training AI tools, and the results revealed the advantages of relying on commercial tools rather than in-house products. Stay mindful of those takeaways and the above tips to have the best chances of applying AI effectively to meet the needs of modern businesses, their service agents, and customers.