By Tom Martin

Forrester Research predicts a year of reckoning for artificial intelligence (AI) in 20191. The market research firm foresees a backlash against automated chatbots, which routinely exasperate customers searching for assistance.

The high expectations placed on AI in customer service will also fall back to earth this year. The intended cost savings through reduced headcount in the contact centre will not outweigh the damage done to customer satisfaction.

What should customer experience (CX), customer care and sales leaders do about the AI backlash?

1. Be realistic about the limitations of AI. Poorly-deployed AI can be damaging to the CX. Specifically, it can compromise trust. “AI is supposed to make our lives better,” said Keng Siau, a professor at the Missouri University of Science and Technology. “Smarter companies are being proactive about this by developing trust.”2

One AI practitioner at a leading financial services provider I spoke to recently expressed that keeping expectations realistic is crucial when working with chatbots. They work hard to level-set with line-of-business and contact centre management, establishing realistic cost-savings targets and making them aware of potential negative impacts to customer satisfaction.

In an April 2018 webinar hosted by destinationCRM featuring Glance Networks and Forrester Research3, it was suggested that AI has an appropriate role in the relationship between the enterprise and the customer. However, CX professionals must begin by understanding the customer journey and determine which interactions are best handled by bots and which should be handled by humans. High-complexity transactions should be given the human touch.

CX professionals should map customer journeys and determine which ones are appropriate for automated support versus human support. For example, in the above graphic, you can see that a Loan Calculator would be an AI function and a Loan Application would receive human support.

Model your buyer behaviour based on important characteristics of the customer’s decision like their financial investment, the duration life of the decision and the customer’s confidence (or lack thereof) in judging alternatives. Low risk/high confidence engagements can be mapped to automated support tools while others may be mapped to high-touch human support.

2. Build better escalation paths from AI to humans. Forrester Research predicts that “a whopping 60% of chatbot deployments in 2019 will not have effective live-agent safety nets attached to web chat sessions”4.

One major reason why customers have grown to hate chatbots is they are often ineffective and offer no recourse in securing solutions to customers’ issues. CEB (now a part of Gartner) reported that, although customers may have a preferred method of communication with your brand, 84% of customers simply want their issue resolved as quickly and easily as possible5 and are willing to be directed to the best possible option to achieve that result. In other words: if AI can deliver a speedy solution to the customer’s problem, the customer will be satisfied.

But what happens when AI fails to solve the problem? It is strongly suggested that you build an escalation path that allows the customer to rapidly escalate from the failure of AI to awesome human engagement that can convey empathy, resolve the issue and save the day.

3. Invest in technology that creates superhuman contact centre agents. Forrester Research predicts that one-third of companies will modernize agent desktops to augment the human workforce6. Instead of using technology to replace humans, this application of technology would improve the agents’ abilities to do their jobs with superhuman quality and efficiency.

For example, the same AI practitioner mentioned earlier told us his financial services company is creating a “banker bot” that does not interact directly with customers, but instead assists human agents to perform their human-to-human interactions better. The banker bot will feed agents customers’ historical records so that the agents can have more contextually relevant conversations with customers. The banker bot can predict additional products customers might find useful and tee up sales support aids for the agents. The banker bot also can predict the questions customers are likely to ask and can tee up corresponding help documents and resources for the agents.

Another example is cobrowsing or screen sharing technology integrated into the agent desktop. This allows agents to see where the customers are stuck in the web sites or apps, giving the agents an immediate understanding of the customers’ problems and enables the agents to guide the customers to speedy solutions. Cobrowsing and screen sharing can eliminate the awkward and time-consuming “game of 20 questions” at the beginning of each call where customers struggle to explain their problems to the agents.

Companies should add to the agent toolbox technologies that can connect with their customers on an emotional level, understand the customer more fully and easily, solve problems faster and communicate more effectively. The ability of a service provider to demystify and personalize the CX is the tiebreaker in industries where service and product offerings among competitors is approaching parity; CX is the ultimate “brand enhancer” for any company.

4. Automate KPI capture in the CRM. Sales reps revile the coming of AI because of its capabilities to collect key performance indicators (KPIs), grade sales reps and penalize underperformers. Forrester Research predicts that salespeople will start to game the system and input falsified data into systems in an effort to save face7.

There is another way to take KPI gamesmanship completely out of the equation, of course, and that’s by building frictionless workflows for agents that automatically capture KPIs in the CRM system. For example, one of our customers, ConstructConnect, a data provider for the construction industry, has integrated screen sharing technology into their CRM system, making it possible for sales agents to launch sharing sessions directly from the agent desktop. The solution made it extremely easy to launch demos. At the same time, the metadata of the sharing sessions are now automatically captured as activities in the CRM, making it easy for sales management to inspect the sales process to ensure that demos are being conducted at the right sales stage. As a result, ConstructConnect has seen a 40% improvement in lead conversion, driven in part by a 66% increase in the volume of demos to new leads and its sales cycles shrink by 60%.

Providing exemplary customer service is a linchpin in capturing and retaining customers. AI can play a role in customer engagement, but ultimately humans are the best resource for closing sales and maintaining high customer satisfaction. AI can do many things, like scheduling and compiling information, but ultimately this technology simply is not equipped to provide empathy and a caring touch to a customer’s problem. Achieving the right balance between technological advances and humanity is the way to succeed in 2019.

Tom Martin is the CEO of Glance Networks where he and his team help enterprises create the ultimate customer experience with smart, omnichannel visual engagement solutions based around integrated cobrowse, screen share, mobile app sharing and one-way agent video. For more information, visit

1 Michele Goetz, Brandon Purcell, Craig Le Clair, Diego Lo Giudice and Mike Gualtieri, “Predictions 2019: Artificial Intelligence”, Forrester Research, report, November 6, 2018.

2 Christopher Null, “Dealing with AI backlash”, ServiceNow, blog.

3 “The ROI of humanizing customer service engagement”, DestinationCRM, webinar, April 2018.

4 Predictions 2019, Ibid.

5 Matthew Dixon, Lara Ponomareff, Scott Turner and Rick DeLisi, “Kick-Ass Customer Service”, Harvard Business Review, January-February 2017.

6 Predictions 2019, Ibid.

7 Predictions 2019, Ibid.

Previous post

Turning a mistake into success

Next post

Making sense of data trails



No Comment

Leave a reply

Your email address will not be published. Required fields are marked *