The artificial intelligence (AI) revolution is fundamentally altering the nature of enterprise software. And nowhere is this truer than in the customer service space. 

Traditionally, the responsibility for maintaining enterprise software systems rested largely with the vendor. However, with the rise of AI-powered software, this responsibility has shifted from being primarily on the vendor to being shared between the vendor and the customer. Consequently, the customer now has to undertake significantly more work than they would have previously done.

This blog was written by Adrian Swinscoe, Customer Experience Advisor, Author, Speaker, Workshop Leader and Aspirant Punk at Punk CX.

The artificial intelligence (AI) revolution is fundamentally altering the nature of enterprise software. And nowhere is this truer than in the customer service space. 

Traditionally, the responsibility for maintaining enterprise software systems rested largely with the vendor. However, with the rise of AI-powered software, this responsibility has shifted from being primarily on the vendor to being shared between the vendor and the customer. Consequently, the customer now has to undertake significantly more work than they would have previously done.

In addition, maintenance now encompasses not only design, testing, monitoring, management and tweaking, but it also includes all aspects of data sourcing and cleaning, as well as content generation, which are the lifeblood of any AI system. This shift also necessitates additional resources, and the investment required to support these extra efforts.

These changing dynamics are causing issues within enterprises regarding their ability to prove return on investment (ROI) and to scale AI solutions beyond the pilot or proof-of-concept stage. This is supported by recent research from both Accenture and IBM, which indicates that enterprises are currently experiencing mixed results with their AI initiatives and investments.

For example, Accenture examined over 2,000 generative AI projects and consulted more than 3,000 C-level executives, finding that only 36% stated they have scaled generative AI solutions, while only 13% of executives report having created “significant enterprise-level value.” Meanwhile, IBM surveyed more than 2,000 CEOs globally and discovered that only 16% have managed to scale their AI initiatives across the enterprise. And only 25% report their AI initiatives have delivered the expected ROI over the past few years.

While this data may highlight some of the broad challenges that companies face when implementing AI systems in customer service, it doesn’t address some of the specific challenges that customer service and CX leaders face when implementing AI.

Let’s look at four stories from different organizations that illustrate some challenges CX leaders have encountered when incorporating AI into their customer service and support departments, what we can learn from each of them, and what it takes to successfully harness the power of AI and your people.

AI Resistance in the Financial Services Industry

The first story comes from a financial services firm that implemented an AI-powered chatbot designed to help customers self-serve. Like many other AI-powered chatbots, this one included a human-in-the-loop (HITL) or human oversight capability to help train the model and monitor the quality of responses provided to customers. This would help to ensure they were on brand and aligned with the company’s policies.

It would also allow agents to step in when the AI encountered a query it couldn’t handle. However, the project failed as the existing customer service representatives refused to engage with it, feeling threatened by the new technology and fearing that they were effectively training their replacements.

Fear That AI Training Will Shift Agent Time and Focus 

The second story involves a customer service operations director of a retail brand who was facing pressure from the leadership and board of directors to implement AI into their day-to-day operations to help drive improvements in service outcomes and productivity. However, doing so would have required them to redeploy some agent resources away from the frontline to assist in designing, training, testing, managing and monitoring these new tools.

Consequently, this would have jeopardized their ability to meet their existing KPIs and service-level agreements (SLAs), putting not only their own performance at risk but also the rewards that agents would have received if they had met their targets. As a result, the customer service operations director decided to do nothing until they received the budget and additional resources required to undertake this new work.

Clear and Inconsistent Documentation Upends AI Value

The third story comes from a customer service department in a business services firm that wanted to deploy agent-assist technology to help improve the productivity and efficiency of their agents. They procured some technology and proceeded to upload a slew of policy documents, training materials and FAQs to train the AI model and to make sure any suggestions were on brand.

However, following the implementation and training period, they discovered that the output from the agent-assist technology was largely ineffective because their documentation wasn’t clear or consistent. This was an issue their agents faced daily, but they had managed to provide effective and appropriate responses based on their understanding of the business.

AI Empowers a Greater Focus on Customers

The fourth story comes from a customer service department at an eCommerce firm that successfully implemented an AI-powered chatbot to enable customers to self-serve, along with agent-assist technology that included an automated messaging capability. This allowed them, using AI and automation, to respond automatically to a large volume of simple and straightforward queries that took up the majority of their team’s time.

As a result of implementing this technology, they were able to turn the phones on, something they hadn’t previously managed to do because of capacity and resource constraints. This enabled them to spend more time talking to customers, solving tricky problems, building connections and adding value.

1. Bring Your Agents With You

You must bring your agents with you – and have a story to tell them about what you want to build, why you want to build it, what will be your agents’ role, how that will change and how you will help them adjust. If not, then there is a distinct possibility that your initiative will fail or, at the very least, won’t meet its objectives. 

This is supported by research from McKinsey that shows that over 70% of all transformation projects fail, largely due to under-investment in three crucial areas: communication, behavior and training. That might not be a surprise to many people who have their own personal experience of going through a transformation project. However, it seems like it is a lesson that is often worth re-learning.

2. Align Metrics and Resource Your Implementations

It bears repeating that, with the arrival of AI-powered software, the nature of enterprise software has changed. If we are to successfully harness AI’s potential, we must ensure that we resource it appropriately and align our internal targets, metrics and incentives to give ourselves the best chance of success.

3. Ensure You Have Quality Data

In computer engineering, there’s an adage that says: “Garbage In, Garbage Out.” This applies not only to your quantitative data but also to the qualitative data that you use to train your chosen AI solution.

As the third story above illustrates, just because you have written policies, procedures and FAQs doesn’t mean they’re effective. Before you blindly assume that your official company documents are worthwhile, ensure you work hard to make certain your data — particularly any documentation you will be relying on — is clear and consistent. Remember: Your AI solution will only ever be as good as the data it’s trained on.

4. Be Clear About the Choices You’re Making

Many brands are choosing to implement AI with the aim of improving efficiency and productivity, as well as reducing operating costs. While it may not always be evident from the outset, this often results in a reduction of headcount due to the efficiency and productivity gains achieved through leveraging AI.

However, there are others, like the company featured in the fourth story, that are making different choices with the increased agent capacity that they now have available. They’re choosing to leverage the vast amount of agent intelligence that has now been freed up to not only solve complex customer problems but also to invest time in talking to and building better relationships with their customers. They’re secure in the belief that if they do so, it will yield positive returns for the business in terms of increased loyalty, revenue and profitability.

Conclusion

Implementation of AI-powered software doesn’t happen in a vacuum. Be clear about the choices you are making and why.

Despite the current level of hype surrounding AI, it’s still early days in the AI revolution. Therefore, if you’re looking to implement AI to enhance the customer service and experience that you want to provide, you should consider these lessons if you want to give your efforts the best chance of success.

To learn more, check out Adrian’s blog, download a podcast episode or email him.