There is a strong sense of déjà vu arising from the conversations about whether the new wave of artificial intelligence (AI)-driven chatbots and virtual assistants (VAs) will significantly reduce the need for contact center agents.

After all, labor savings is a compelling market driver for such automation. Staffing accounts for approximately 70% of contact center operating costs and it drives capital, namely facilities costs.

The Frost & Sullivan research paper, The Future of the United States, forecasts that 47% of American jobs are at high risk of being automated in the next 20 years. It lists contact center agents, along with other low-skilled positions such as typists, bank tellers, and file clerks as positions that may disappear.

But back in the late 1990s, email, IVR, and eCommerce via the web emerged as voice agent alternatives. The booming dot-coms tried to avoid having contact centers.

So what happened? Customers disliked the IVR experience, the waiting forever for email responses, and not being able to reach live agents. The dot-coms scrambled, seeking out BPOs to handle customer service. Then came the one-two punch of the tech bubble bust and 9/11, which contributed to a brief downturn.  But when the economy recovered, the contact center industry expanded—much of it nearshore and offshore–throughout the following decade.

There are signs of a similar contact center expansion today even with VAs, and much improved IVR, web, and email customer experiences.  For example, the Site Selection Group reported that in March 2017 there were over 4,300 contact center jobs created in the U.S. alone, compared with just under 1,200 losses.

In reality, the new technologies have slowed contact center growth. There are far fewer centers and agent positions compared to what would have been built had these transformative tools not been deployed and refined.  As examples, online reservations, shopping, and ticketing have displaced contact centers for these functions.

Automation is in a race with the economy.  To create net new jobs, demand and incomes have to outpace automation. But automation, thanks to tools like AI, is catching up while the economy is maturing. For example, the U.S. Census Bureau forecasts a continued slowdown of population growth.

The degree that automation through new tools like chatbots and VAs is successful depends on how well the applications are made and implemented. All new technologies undergo teething pains. For those insights customer contact organizations (and vendors) have to pay heed to the hard lessons of the past and present. Here are several.

  1. Philosophically decide to focus automation to improve customer experiences (CX) in order to build customer loyalty, and attract other customers through social referrals, rather than as cost controls. Going down the cost management hole will inevitably lead to process changes that will annoy and ultimately turn away current and future customers.
  2. Make the applications intuitive to use, ensure seamless connections between channels, and provide single and up-to-date versions of the truth across all channels. Plan ahead to expand automated applications from standalone for specific use cases to become integrated with other and live agent channels.
  3. Educate customers and staff on the benefits of using the new tools. Give them opportunities to experiment before going live: and learn and make adjustments from their experiences.
  4. Determine which types and how many live contacts that the applications will displace initially and over time; AI “learns” from each interaction.  There will always be those that these solutions cannot handle.

Frost & Sullivan published a research report on VAs that also provides several key points to consider.

“When done well, VAs can provide a rich channel of customer interaction that in many cases can handle complete and complex customer inquiries. They provide the chance to enhance a brand’s Customer Experience. They can provide brand differentiation. They also can speed service delivery and reduce costs.”

“However, done poorly they can have the opposite effect. When a customer gets stuck, the VA doesn’t understand the inquiry or gives the wrong response, customer satisfaction plummets…VAs can offload a lot of tasks. But that last 10-30% of an inquiry that requires nuance, intuition, judgment, or in some cases, “tribal knowledge” not found in a database or captured in best practices, can make all the difference in customer satisfaction.”

Brendan Read is Senior Industry Analyst with over 25 years’ experience covering business, communications, staffing, and technology. He has worked in, prepared reports, and blogged on a wide range of topics including customer contact, CX, CRM, IoT, social media, supply chain, and BC/DR. He also has backgrounds in construction, manufacturing, materials, resource extraction, site selection, and transportation. He examines the broad economic, environmental, innovation, political, and social megatrends, and their impacts on businesses, markets, and society.