30 Aug How to maximize your contact center AI ROI?
Posted at 13:27h in Contact Center AI
Contact centers are at a crossroads. The pandemic has forced them to relook at their overall resourcing strategy. Besides the forced remote working and constrained access to talent, contact centers face external competition. Different industries in the US are competing for the same contact center labor pool, and that is likely to continue upward pressure on resource availability and wages in both the short and medium-term. Many contact centers are therefore deploying technology that reduces overall need for human resources, especially for customer facing roles. All this has accelerated focus and adoption of contact center AI solutions.
Robust contact center AI strategy is a big challenge
89% of CX professional believe in the importance of using AI in contact centers and acknowledge its role in improving CX. Yet, most organizations lack a coherent and exhaustive contact center AI solution. There are multiple reasons for the lack of success that plagues AI deployment in contact centers.
- Limited clarity on objectives – Most organizations approach AI deployment without clarity on the objectives and what AI is meant to achieve for their contact centers. Hence the solution development and its integration with the existing workflows are impacted. Moreover, they aren’t sure how to measure success of their program and which metrics to use – AHT, ASA (or wait time), FCR, deflection rate or any other metrics.
- Creation of channel siloes – While chat has been the primary channel using self-service, for rest of the contact center, voice continues to be the preferred channel. By applying A automation in pockets, the contact center AI journey creates channel siloes, with each channel competing against others and delivering different levels of experience
- Lack of domain understanding – Customers expect AI to be domain-aware and provide contextual, intelligent answers. The days of dumb chatbot are gone but majority of contact center AI solutions operate rely solely on speech recognition and natural language understanding which are useful when it comes to generic intents, but fall short when complexity and domain intensity goes up.