The voice assistant is an extension of a brand and an opportunity to create even stronger relationships with customers.
Brand-owned voice assistants let organizations ensure the voice assistant experience is crafted in alignment with the overall brand strategy. They offer the ability to create a unique experience for customers while also giving you complete control of the data flowing through the system.
Mike Zagorsek, chief operating officer of SoundHound, pointed out the three key advantages of brand-owned voice assistants:
- The ability to own user experiences and retain customer relationships
- The visibility into critical data and analytics
- The capability to generate real revenue
From his perspective, insights gained from aggregated data collected through conversational intelligence provide brands with the critical information required to understand customers’ needs better, improve the voice experience and inform product and customer service roadmaps.
Plus, brand-owned voice assistants are a boon for data privacy, a hot topic for many consumers. “When brands own their technology and control the data and how it’s used, customer confidence can be further developed when data collection policies are communicated with assurances that data is anonymized and not shared with other entities,” Zagorsek said.
On the topic of revenue, he added, “Only when brands own their voice experience can they open the door to monetization opportunities that may include partnerships with other entities — such as restaurants or content domain providers — and begin to collect real revenue from the voice assistant.”
Related Article: Examining the Core Components of an Effective Voice Strategy
Tech Stack Should Include ASR and NLP
Custom voice assistants typically involve a few layers in their tech stacks, including:
- Wake word engines that listen for key words or phrases
- Automatic speech recognition (ASR) that parses speech to text
- Natural language processing (NLP) to extract meaning from an utterance
- Text-to-speech to deliver back responses
Eric Turkington, vice president of growth at RAIN, explained that there is a range of different enabling technologies for each piece of this stack.
Some companies specialize in one slice (e.g., pureplay synthetic voice companies focused on text-to-speech voices), while others provide tooling that spans the spectrum but can also be employed a la carte.
Turkington added that machine learning techniques can be applied to automate the training of speech recognition models for improved accuracy and speed and adapt to different microphones, noisy environments, new languages, vernacular and even regional accents.
“These techniques are also used to improve language processing and understanding tasks such that systems can get better at discerning user intent to take the right action.”
He said the first step in building a custom voice assistant is identifying the meaningful tasks to complete for customers across their journey.
This process includes finding what he often calls “assistive moments,” where users are looking to know a piece of information, do a specific thing or buy something. These become moments where a voice assistant can do the job faster or better than the status quo.
Custom voice assistants can have a narrow utility to start (a “crawl” stage), but whatever they do should be frequently needed and tangibly valuable to the end-user, forming a solid basis on which to build into “walk” and “run” versions.
“To flourish, custom voice assistants demand a true product mentality of ongoing development and optimization, both as products in their own rights but also as part of the fabric of whatever device or software they are integrated within,” Turkington said.
Related Article: How Amazon SageMaker Studio Assists Marketers With Machine Learning
Voice Assistants Can Discover Trends
Tony Lorentzen, senior vice president and general manager of intelligent engagement solutions at Nuance, explained that while these assistants are an extension of the brand, they also provide huge value to the overall financial performance of an organization.
“Voice assistants can mimic a live agent by completing transactions, providing recommendations and enabling purchases,” he said. “Lastly, voice assistants can give brands early indication of emerging trends in a business.”
They understand and classify intents and can alert business owners of changes. For example, a new product that was recently shipped has an increase in customers contacting support.
“This not only is helpful from a privacy and compliance perspective but also lessens the chance of other technology companies leveraging your customer data to achieve their own business objectives,” Lorentzen said.
AI Offers Real-Time Insights
Conversational AI that can manage speech and natural language and respond quickly and accurately, even with background noise and accented language, supports a robust voice assistant, explained Zagorsek.
He said the best voice AI technology offers easy, intuitive and intelligent voice AI interfaces that process speech the way humans do — in real-time.
However, he pointed out that many challenges and opportunities exist when deploying a brand-owned voice assistant.
“The first step is to have a strategy and to carve out the time with the right stakeholders in the organization to determine the goals and the needs the voice assistant will address. Organizations must then consider how they’ll design and develop their voice experience.”
Zagorsek said that deploying a brand-owned voice assistant is either a build-it-yourself or a build-it-with-a-partner strategy.
In making this decision, companies must determine the level of investment and expertise in technology and engineering they can provide to decide whether it’s right for them to build their own voice interface or partner with a voice AI platform provider.
Flexibility of the Platform Is Key
Lorentzen noted that the key to a unique voice assistant experience is flexible technology that fits with existing systems while offering opportunities to customize deployment and, ultimately, the consumer experience.
“The best voice assistants are those that are seamlessly integrated into a brand’s cross-channel experience, meaning customers can access the information they need and complete transactions without having to leave the application,” he said.
Voice assistants need to integrate into a CRM to retrieve and update customer data and use that data to personalize interactions.
Here, AI and machine learning can help predict caller intent and customize end-user interactions — both of which can maximize live agent time by providing next-best responses and recommendations to agents based on the end-user interaction.
Related Article: Intent Data Explained: Types, Sources, Tracking and Scoring
The Bottom Line
According to Lorentzen, the most successful voice assistant deployments happen when a cross-functional team is invested in the implementation.
Often this requires representation from the customer engagement or customer service team, the IT team and those from marketing and security.
“Together, this group can ensure the voice assistant is meeting the market need while fitting cohesively into the overall company strategy and aligning with overall compliance requirements,” Lorentzen said.