Currently the CX advisor for software company Freshworks, Colin Crowley has spent over 12 years managing global customer experience organizations. As customers get more used to doing everything remotely — from working to ordering groceries — they also have new expectations for how they deal with companies, he said. But as he notes, the pandemic has also created some “cultural angst” among consumers who are more anxious about the trustworthiness and delivery of services.
“There has been a renewed focus among many customers on understanding what quality of service really means and investing in the quality of their customer interactions through more robust agent training, agent empowerment and customer-friendly policies,” Crowley said.
Freshworks was a sponsor of CMSWire’s Digital Experience Summit, which took place as an online event on Feb. 16 and Feb. 17. Crowley hosted the session “AI’s Second Cousin: RPA and the Road to AI.” He spoke with Simpler Media Group about how to start using AI in your organization, how customer expectations have shifted in the past few years, and how not even AI can replace the human mind.
The Trick to AI? Take an Incremental Approach
Simpler Media Group: Has customer experience expectations shifted during the pandemic? If so, how?
Colin Crowley: Customer expectations have indeed shifted, partly with the pandemic but also partly without it. With rising technology and ever-faster delivery of services and products, customer expectations are one-way streets, always trending upwards. I think the pandemic has uniquely contributed to the fact that more customers expect companies to have a strong digital presence and [that] customers expect greater empathy from companies.
SMG: As organizations are inundated with pitches for the latest AI solutions, how can decision-makers make sure to make investments in AI intelligently and strategically? What questions do they need to ask to make sure they are investing in a solution that can truly serve a purpose for their company?
Crowley: AI is tricky because it typically comes with a very high price tag and it’s a big black block. You don’t really know what you’ll get out of it until you try it, because the level of efficiency savings can be very particular to your business and your customers. The trick is to take an incremental approach and not feel [like] you have to swallow everything at once before you know what to expect from these systems and before you have built up enough institutional knowledge to manage them well.
I would rarely suggest a company jump into AI too quickly. For instance, it’s probably not wise to leap into the chatbot space with the most advanced chatbot imaginable. Instead, start with a more basic ‘answer bot’ (or ‘info bot’) that simply fetches FAQ articles. Then try a more advanced chatbot with pre-built question-and-answer flows. Then try a chatbot with one API-based workflow that allows customers to actually take action on something through the bot. And so on and so forth. [Track] customer satisfaction and contact deflection along the way.
By the time you get a $50,000/year AI solution, you should be able to ask, ‘What ROI do I expect to receive from using this technology and why?’ with actual data from prior trial experiences — as opposed to hopeful theorizing.
SMG: Tell me about RPA: Is there anything you wish more people would understand about its potential in the workplace? How do you think it could impact the workplace in the future?
Crowley: RPA (robotic process automation) basically involves a bot mimicking the actions of a human as they click their way through a process, all triggered by human agency. So it’s a way to take routine, mundane, high-volume tasks and hand them off to a machine. RPA is therefore a bit like the digital equivalent of the conveyor belt. It’s less flashy than AI and less well-branded, but because of its simplicity, it’s much easier to execute, much less costly, and much less risky.
RPA is a great way to incrementally push towards more advanced solutions like AI — where the bot not only takes the action, but replaces the human in deciding when and what actions are taken. RPA can do a lot across-the-board at any organization to free employees from boring tasks and empower them to spend their time flexing their talents more fully and effectively.
The Potential of Customer Self-Service
SMG: Where do you see the biggest gaps between the optimal digital experience and reality? Why are companies missing the mark, and how can they do better?
Crowley: One area where there is a gap between the optimal digital experience and reality is in the world of self-service, where there’s still a lot that can be done to use technology to surface more targeted self-help options to customers, which studies show 75%-80% of customers regularly desire. But a lot of companies don’t invest as much in this space because the ROI is hard to clarify and there isn’t a lot of institutional knowledge around it.
Another area where there is a gap involves greater personalization of products and services by effectively using all the great data that exists digitally. But using this technology can be tricky because there’s so much of it and organizations frequently don’t do enough to invest in teams that can gather data, examine it holistically, analyze it and develop key learnings and action plans.
Lastly, there’s more that can be done to meld the social experience with the digital experience, since handling customer concerns via social media channels is increasingly no different from handling those concerns via more traditional channels. But this requires close coordination between customer support teams and marketing teams, a different use of technology (the best social media platforms are seldom the best customer support platforms), and a new style of communication across-the-board — which many companies are still trying to figure out.
SMG: What are the most over-hyped assumptions about AI and CX, and what are the most promising? What are you most excited about?
Crowley: The most over-hyped assumption is the idea that AI can be a panacea for all ills and that it will be handling [at least] 50% of all human tasks by next fall. In that regard, it’s not uncommon to have technology vendors throw out extremely optimistic numbers for AI systems, such as chatbots with deflection rates of 60%, 70% [or] 80%. Realistically, that’s unusual and most companies hit a ceiling at 35%-45% if they’re lucky.
Nothing can really replace the human mind. AI has limitations like everything else, and when used strategically and intelligently, it can net you great benefits. But its role will always be a nuanced one and needs to be applied surgically.
I feel the most promising area of AI is in AI systems that are employee-facing or agent-facing rather than customer-facing. Because they impact the customer so directly and can provide a bad experience if not done well, [they] are inherently higher risk. More and more, AI companies are focusing on the agent experience by empowering agents with crucial live support — such as recommended answers to customer questions based on what worked well with customers in the past — as they navigate real-time digital channels like live chat and social messaging. This has the ability to change the agent experience significantly and vastly increase efficiency, standardization and quality of service across digital channels.
Catch up on the Winter Digital Experience Summit on demand here.