Main content

Dreamforce 2019 - Salesforce talks up voice as telephony comes to Service Cloud

Phil Wainewright Profile picture for user pwainewright November 19, 2019
In a tie-up with Amazon, Salesforce brings telephony into Service Cloud, while looking to a future where customers use voice to get data into and out of CRM systems

Bret Taylor (r), Salesforce CPO, with Einstein Smart Speaker DF19 by @philww
Bret Taylor (r) with Einstein Smart Speaker

Salesforce wants to get everyone talking. New voice capabilities are at the center of announcements to open its annual Dreamforce conference today. Most notable is the integration of telephony directly into the Service Cloud console for the first time, making it possible to analyze and act on voice conversations in real time, in the same way that already happens with text-based chat and social messaging. Salesforce is also expanding on voice assistant capabilities first announced at last year's event, although this is more of a futures pitch, with general availability still not expected until 2021.

To highlight the union of machine learning, analytics and voice under Salesforce's Einstein branding, the Dreamforce campus sees the introduction of a cute Einstein Smart Speaker, which thanks to the recent Voice Interoperability Initiative with Amazon and other vendors, is compatible with Alexa as well as Einstein Voice. Unfortunately, however, the device is not currently available to buy.

Three separate use cases for voice

The full spectrum of voice offerings now spans three separate use cases, as chief product officer Bret Taylor explained yesterday in a pre-briefing with selected media.

First there's the use of voice to input data into Salesforce, as launched last year. This is especially useful when people are on the move or otherwise find it difficult to use a keyboard. That's immensely valuable in helping enter data into the system more rapidly, he explains:

It's a huge issue for every sales manager and every CEO in the world. You need to make sure that your data is up to date, and that it's accurate. And if you talk to any sales person, they'll tell you they do not want to spend time updating a piece of software, they want to spend time selling. With these voice technologies, you can walk out of a business meeting, update the data in Salesforce, and all that just happens automatically.

Secondly, there's the use of machine learning to analyze conversations in the call center as they happen, as announced today. This makes it possible for machine learning to work in the background and automatically present relevant information to the agent without them having to go off and search for it:

As a consumer when you call customer service, all the time you get put on hold because the agent on the other side of it is doing what we call swivel-chair integration. They put you on hold, they turn to a different screen that looks some information up. By bringing together the power of voice recognition with the power of Einstein. we're enabling those agents to get these real-time recommendations in these customer service calls.

It's a huge factor for people around contact centres, first of all, because that's essentially money saved because people are spending less time on calls to get to a successful resolution. And success for consumers — if you go to any sort of survey at the end of the call — is 100% correlated with how quickly you got your question answered.

The third use case is being able to ask questions of the system to find information. Today's announcements extend this with the introduction of a point-and-click tool that enables Salesforce administrators to build their own custom voice skills, using voice commands and queries to do any task in the CRM system. This will go into a year-long beta trial early next year. Taylor believes that using voice to interrogate the Salesforce system for information will really open up the use of data more widely:

There's so many amazing statistics about the proliferation of data in this digital age. I think there's an intuition, from almost every executive I talk to that, are we really taking advantage of all that data? Voice isn't necessarily the only way to interrogate that data, but it's certainly the most intuitive. Across almost every domain we see, we think voice will reduce friction and increase adoption of modern technology, and drive that sense of guidance from AI, so it's not just intuition driving businesses.

Telephony comes to Service Cloud

The new Service Cloud Voice, which brings telephony into Service Cloud for the first time, is based on Amazon Connect contact center technology, taking advantage of Salesforce's use of AWS as its primary public cloud provider. Amazon's AI-powered speech analytics provides real-time machine transcripts to Einstein for analysis so that it can surface recommended answers, contextual knowledge articles, and next best actions to the agent within the Service Cloud console. This frees up the agent's attention to focus on the conversation with the customer, explains Patrick Beyries, Vice President Product Management at Salesforce Service Cloud:

We're converting the phone call into data and of course data is what we're good at manipulating on.

Customers can either use the preintegrated telephony packaged with the solution, or connect existing telephony providers such as Genesys. Beyries comments:

It's the first time ever you've had the ability to buy telephony from Salesforce, which is something customers have long wanted to do.

As well as fetching relevant information, the service can also be used for real-time coaching by supervisors, and can automatically provide a wrap-up summary at the end of a call, tagging it for content and follow-up.

My take

Bringing real-time voice analysis into Service Cloud is a table-stakes move that Salesforce has to make to stay competitive with other providers in the marketplace. The convergence of customer channels continues apace, and enterprises need to be able to provide rapid resolution when customers escalate to a voice call. The use case for voice as a data input tool is also well established.

Getting businesses to adopt smart speakers, though, is more of a long-term bet on future directions. Bret Taylor cites the behavior of his own toddlers with Alexa as a guide to how people will quickly get used to voice as part of their user experience. But there are still obstacles to adoption of voice assistants in a business environment — the office is not like a home, there are more people about for a start.

What needs to happen before this technology takes off is for businesses to find some compelling use cases. That's what Salesforce hopes will happen as a result of it launching its new Voice Skills tool into beta. But this is currently very much a case of technology looking for a use case. We wait to see what emerges next year once beta adopters get their hands on it.

A grey colored placeholder image