Artificial Intelligence Takes Off in the Customer Interaction Space
Intelligent Assistants and smart bots have been stealing the headlines. Yet, Artificial Intelligence (AI) is finally making headways in the broader Customer Interaction Management space. This market has historically been prudent with adopting new technologies. Customer service departments have a lot of technology options to choose from to better their productivity and the customer experience. These organizations are measured using traditional Service Level Agreements (SLAs) and other activity related metrics. It incents them to invest in software allowing incremental improvement of these performance indicators. This has led to a conservative approach with breakthrough technologies such as AI. This is changing though, and I would like to go over the state of AI adoption.
Virtual Customer Assistants (VCAs) sometimes called Virtual Agents (VAs) and customer service bots continue to be the most vibrant AI-enabled segment of the Customer Interaction market. They were voted the top channel growth in the DiData 2017 Customer Experience Benchmarking report. In my latest market scan, I found more than 50 participants. Technology maturation has accelerated lately, leading me to believe this application of AI has crossed the chasm. The development of customer assistants and customer service bots is driven by the massive adoption of mobile messaging.
The popularity of messaging has not reduced the importance of calls. AI is changing how they are handled, in particular with voice self-service. Machine Learning has boosted the accuracy of speech recognition from a stagnant 80% to 95%, as good as humans. Furthermore, it has allowed new players to emerge in a market once dominated by Nuance. Machine Learning is not only improving speech recognition, it also dramatically reduces the engineering and professional services required to implement solutions. Because voice is a very compelling modality for interacting with mobile devices, all the technology giants have also invested in the technology, contributing to its commoditization. Amazon with Alexa brought it into the home. I was expecting these developments to trigger a wave of Interactive Voice Response (IVR) upgrades and modernization. Indeed, IVR remains a huge source of customer frustration. Speech combined with Natural Language Processing (NLP) can transform the customer experience like at Royal Bank of Canada. It hasn’t happened though. I don’t have good explanations for this slow adoption but expect it to change.
Assistance is not just for customers
Agent productivity and providing customer service representatives with the proper tools have always topped the agenda for customer service organizations. The development of self-service has actually further increased its importance. With self-service taking over routine and easy interactions, agents are left handling the more complex ones or those that couldn’t be resolved using self-service. It has become critical to help agents with better tools to juggle the various systems they have to deal with and access knowledge bases or other sources of information. Robotic Process Automation (RPA) has brought some responses to these issues but as the name suggests focuses mostly on automation. With AI, customer service representatives can get recommendations on how to best handle customer questions using assistants from the likes of Genesys (Kate) or Pegasystems.
One tedious task for agents is to find the most relevant information to resolve cases and tickets. In addition to AI improving knowledge bases, Machine Learning can dramatically accelerate the response process. Companies such as DigitalGenius or SmartAssist (formerly Wise.io) harness it to analyze customer service requests and responses. They can optimize the classification of customer inquiries, automate their triage, and recommend responses. AI shines in this scenario because of its ability to scale, unlike traditional approaches that use manual rules or macros.
Routing of interactions can rapidly become complex and is also a perfect candidate for AI. This use case has been pioneered by Afiniti and Mattersight. The two companies have approached it from a different angle. Afiniti focuses on sales motions. It uses big data, combining information from in-house applications and third-party databases into rich consumer profiles. It then applies AI to find the best pairing with an agent. Mattersight built its solution on the Process Communication Model behavior theory. It uses speech analytics to map callers and agents to six personality types and match them together. Although these machine-based approaches enjoy impressive ROI–Afiniti says it helped T-Mobile generate $70 million in incremental revenue per year; their market adoption has remained limited. Again, I expect this to change.
Making sense of data
Interaction management and customer service are among the most tracked and measured processes within enterprises. Organizations enjoy a plethora of metrics and reports. However, they all struggle to make sense of all this data, extract actionable information, and identify what to focus on to improve their operations. Moreover, the industry has been relying on activity metrics such as Average Handle Time (AHT), Service Level Agreements (SLAs), and First Contact Resolution (FCR). These measures can have loose and sometimes misleading linkages to desired business outcomes and customer experiences. AI should shine with these large sets of data, uncovering patterns and correlations. Alas, I am not finding many solutions in that realm. Recent announcements from Medallia or Qualtrics suggest Voice Of the Customer (VoC) vendors have a head start. I expect AI to have a big impact, in particular on performance management and customer segmentation.
I tried to plot the current stage of AI penetration in the various product categories.
Despite a slow start, AI for Interaction Management is starting to take off. What do you think?
Disclosure: I have been helping several companies mentioned above, including Afiniti, Genesys, and Wise.io.