Voice AI Joins the CRM Conversation
As artificial intelligence continues to evolve, it is already an integral part of every business, no matter the size or industry. And with CRM technologies as the central system of record for most of these businesses, it makes perfect sense for businesses to want to integrate AI into as many aspects of the entire customer experience as possible. It’s not just innovative, it’s essential for streamlined customer data management, personalized interactions, deeper customer insights, and greater operational efficiencies.
And when bringing AI into CRM systems, perhaps the most valuable information can be found in the conversational AI and voice AI systems that are so prevalent today.
“Voice is still the main channel that many customers use,” says Dvir Hoffman, CEO of CommBox, provider of an AI-driven omnichannel customer engagement platform. “If you want to build an autonomous enterprise, you need to integrate the technologies.”
The integration of voice, conversational AI, and CRM systems is the first step in deriving maximum value from voice technologies, adds Nikola Mrkšic, CEO and cofounder of PolyAI, an intelligent virtual assistant systems provider.
Hoffman and Mrkšic are not alone in their thinking. Across the industry, there is little doubt that integrating conversational and voice AI with CRM systems is a great idea. Conversational AI provides not only quick summarizations of any conversation but, when integrated with CRM, overviews of all interactions between the customer and the company across departments and throughout the entire customer journey.
“CRMs provide the key channel to collect all sorts of information that can be tracked, like last contact date, the reason for contact, etc.,” says Brian Yang, senior director of project management at Verint.
“Identifying user behaviors helps improve the customer experience because you know who they are and what they are contacting the company about.”
Connecting conversational AI and voice with CRM makes it much easier and faster to service a variety of customer needs, says Sean Evers, vice president of sales for Pipedrive CRM, a sales platform provider.
In most instances, voice interactions are easiest for the customer—no instruction manual is needed, so integrating voice, conversational AI, and CRM enables customer service tasks (including transactions, product information requests, etc.) to be completed more easily and efficiently, according to Evers.
Integrating voice and conversational AI helps with scalability, according to Zac Sprackett, chief technology officer of SugarCRM. “In today’s world, we’re all being asked to do more with less. The technology helps handle more of those inquiries in a way that leads to positive outcomes without having to invest in the same number of people to handle all of those requests.”
Mrkšic points out that many companies are struggling just to man the phone lines, so the integration of these systems is becoming a necessity from a scalability and customer service perspective.
Integration Considerations
And while seeing the need to integrate the technologies isn’t complicated, getting there is a little more dicey.
Companies that have yet to integrate their CRM and conversational AI systems should first seek to understand the outcomes they hope to achieve, Sprackett advises. “What are the challenges that your business has today that are not being well-served by the systems that you already have in place? Is there a gap between your different business units where you’re not handing off data effectively or handing off customers effectively? Is there a place where you’re looking to become more efficient, or where you feel like you’re missing valuable insight around that customer?”
If companies can’t articulate the problem that they’d like to solve, then any results are difficult to quantify, according to Sprackett. “If you can’t measure those results, then it’s difficult for you to understand whether you’ve actually achieved any of the value.”
Evers agrees: “If you want an AI agent to interact with customers on support tickets, the only way to leverage this is by connecting your support and knowledge base with a large language model. So the first thing to think about is, if you want to digitize in an AI format, via voice, via text, how to connect to the necessary systems and tools, and how to deliver messages back to the customer.”
On the technical side, the success of the integration depends on systems having open APIs, which eases connections with other technologies, several experts agree.
“As long as you have open APIs that are accessible from the cloud, AI can adapt to what you need to do,” Yang says. “The solutions on the market today are designed to meet companies wherever they are at in their journeys. You can build more complex capabilities.”
With available open APIs, the integration of conversational AI, voice (most conversational AI includes the voice element), and the CRM system can be done all at once, but most companies have legacy versions of these technologies, so they tend to integrate in stages, even if all of the technology is from a single vendor, according to Sprackett. Each stage has certain performance measures to be met before advancing to the next stage.
To avoid adverse effects on other systems, such as enterprise resource planning (ERP), you should examine core systems to see how any integration of voice, conversational AI, and CRM might impact them, recommends Nikolaus Kimla, founder and CEO of Pipeliner CRM.
The security of all systems to be integrated also needs to be up to date. The security reviews can be the longest part of the integration process, Mrkšic says. “We do a lot of work with finding the right version of the process that works, including the right version of the technology.”
Evers adds that the integrated system needs to have guardrails so it doesn’t perform tasks outside of company rules. For example, if the company has a 30-day return policy, the integrated system won’t attempt to process a return outside of that time frame.
The integrated technologies also need to be able to work with structured and unstructured data, Hoffman says, pointing out that many older chatbots failed because they didn’t integrate with enough of the business machinery.
Though there is no technical reason to perform the integrations separately, sometimes it makes more sense from a practical standpoint, according to Yang.
“There is nothing out there that helps you do everything right out of the box,” Kimla adds. So he recommends rolling out the integration with one or more select groups first, then expanding the rollout.
The integrated technologies should immediately enhance a company’s productivity, Evers says. Conversational AI will quickly summarize many of the interactions, and the integrated CRM will provide personalized emails, SMS messages, etc. with customers.
Yang agrees about the immediate benefits. To that end, Verint offers 90-day risk-free programs. “The focus has been on how do we get value quickly and early. Ninety days is the time frame that we think that customers should be able to complete the integration in and be able to get value out of it. Most of the implementations start small.”
Yang adds that most integrations take 30 to 90 days.
Evers recommends starting with simple tasks, like scheduling meetings and sending emails. Ensure the integrated technologies are performing those tasks as expected before moving on to more complex tasks, he says.
While Yang agrees that starting small makes sense, some companies start too small. “There can be too much focus on FAQs or simple questions and answers. It’s an easy trap to fall into because it’s very safe. But people usually call in with a question because they are trying to do something.”
Going through the time and expense of the integration just to handle FAQs might offer some efficiency benefits but doesn’t deliver on the promise of the integrated technologies, according to Yang.
Additionally, while having FAQs and answers to basic questions can help in terms of call containment, it doesn’t help customers reach a live agent, Yang adds. Even with the most advanced automation and the latest AI, there are still times a human agent is needed to meet the customer’s needs.
The integrated technologies will be able to perform an increasing number of tasks as they continue to evolve, but they won’t replace humans, Kimla says. “The service industry will never be done. You can’t do a haircut with a machine.”
Look at Change Management
Beyond the technology changes, companies need to consider change management from a business perspective as well, Sprackett adds. Employees and managers need to actually use the technology, not sidestep it when they can.
So training is essential, Sprackett says. “You have people involved as well as the technology. You need to build out training for the people who will be using the technology. Users play a key role. Getting them excited to adopt and use the technology is really important.”
The training should detail what users of the technology should expect, where the value will come from, how value will be measured, and other details.
“You need to know what your North Star is—what you’re trying to achieve—or you will not wind up at the place that you are trying to get to,” Sprackett says. “You need to have a plan and vision about what you’re trying to do. Once you have that, understand that, you build out the plan, navigate the complexity to get there, then get the training right.”
Generative AI can still hallucinate, so companies have to ensure that any answers it provides to customers are indeed factual, Evers cautions. “So it’s super important to think about this when designing the system. A big mistake is not spending enough time thinking about that.”
Beyond the productivity improvements, the integrated technologies will provide customers with superior experiences and improved connections with the brand, according to Evers.
Integration in Action
Several companies are using these integrated systems on a trial basis, according to Evers. For example, some healthcare organizations are using the integrated technologies to book exams, appointments, and follow-ups. Insurance companies are using integrated systems to make changes to policies.
“The really cool things that we have seen people do are recognizing repeat interactions or knowing a customer’s prior interactions to drive a more streamlined experience,” Yang says. For example, to the customer who habitually uses voice/conversational AI to request a statement, the integrated technologies can present a page displaying the statement. Customers who typically ask for something else would be presented with a different landing page.
A PolyAI customer in the restaurant industry uses the technology to prompt booking reminders for holidays at the end of other customer calls, Mrkšic says. The system also collects reservation information to determine any slow times, particularly for holidays, enabling the integrated technology to proactively reach out to regular customers to offer special deals for making a reservation for a slow time.
Transformational for the Future
As Sprackett puts it: “Companies should look at this as an organizational transformation project and not just a technical upgrade.”
“Don’t expect it to be perfect,” Evers adds. “If you’re waiting for perfect, your competition has already likely implemented it. Just get started with it, build out some boundaries or guardrails, and then start experimenting with it. Not every experiment is going to work, but the more you get started with it, the more you will learn.”
Mrkšic adds that some of PolyAI’s customers have very ambitious plans for the integrated systems once they master many of the simpler capabilities.
Even though a company might start slow and many of these integrated systems are still in their trial stages, the combination of conversational AI, generative AI, and CRM will provide outstanding benefits as the technologies evolve and the integration reaches its full potential, according to Evers. “It will kind of give us superpowers and we will be able to provide customers with an amazing experience.”
“First, you need to have all of your ducks in a row,” Kimla says. “But you have to get started. The later you get started, the more that you fall behind. You have to bring all of your applications together and integrate them into your CRM or you’re lost for the future. If you don’t have digital processes in place, you are done, because everything that is repetitive has to be automated.
Phillip Britt is a freelance writer based in the Chicago area. He can be reached at spenterprises1@comcast.net.