Qualtrics generative AI gives surveys a real-time twist

Today, Qualtrics International Inc. is introducing three new features to its line of customer experience measurement applications that incorporate generative artificial intelligence capabilities.

They include a feature that makes it possible for anyone in an organization to easily ask questions and get feedback on customer opinions, surey customization based on responses and intelligent summaries that provide feedback without compromising the confidentiality of individuals.

The capabilities are embedded in Qualtrics’ three product suites – XM for Customer Experience, XM for Employee Experience and XM for Strategy + Research. Qualtrics said they’re intended to enable companies to predict customer needs, address employee attrition, customize shopping experiences and automatically act on feedback.

Survey questions on the fly

The most notable new feature, called conversational feedback, uses AI to analyze survey responses and generate personalized follow-up questions while the survey is in progress. Unlike conditional branching, which generates pre-formulated follow-ups, the feature uses generative AI to create entirely new questions based on responses received. For example, if a survey-taker says a car needs to be more comfortable, the generative AI feature can create follow-up questions that probe for specifics.

Qualtrics said experiences reported by early users of the feature indicate that respondents contribute additional information about 40% of the time and that responses are more verbose and descriptive than standard survey answers. It said customers didn’t detect any change in dropout rates or increases in respondent frustration when asked to elaborate on responses.

“Think about this as really the first major transformation in surveys in decades,” said Brad Anderson, president of products, UX and engineering at Qualtrics. “We’re using generative AI to understand the context and ask the individual for more in-depth and actionable data during the survey. It’s not branching logic; it literally is understanding the context of the conversation and then encouraging the survey-taker to go deeper.”

Anderson said Qualtrics built extensive guardrails into conversational feedback to protect against unanticipated problems like hallucinations, intrusive questions and offensive language.

Giant training data source

Qualtrics said it has more than 20,000 customers globally who collectively log 3.5 billion conversations every year in more than 20 languages. Its database includes over 15 billion experience profiles.

“Our AI is able to understand structured data like survey responses, which general-purpose AI cannot,” Anderson said. “Add into that the years of experience we have building what we call enrichments into the data to give our generative AI solutions much more accurate insights faster and at a lower cost than others.”

One new feature, Qualtrics Assist, allows people to ask questions about gathered feedback using natural language prompts. The large language model has been tuned to understand the distinction between customers, prospects and employees and to transform data into simple insights, recommendations and suggested responses. It can also automatically trigger programmatic actions and workflows in Qualtrics’ xFlow, an XM-specific workflow engine that integrates with various third-party platforms.

“The easiest way to think about Qualtrics Assist is that it is like our Microsoft Copilot,” Anderson said. “It gives you a chat interface with an agent so they can ask questions and ask for recommendations.” Specially trained models are available that understand contexts like hotels or restaurants.

Another new feature, intelligent summaries, digests feedback into concise overviews that automatically surface and analyze insights and offer recommendations for response. It can anonymize responses to preserve confidentiality and generate compilations of video responses in minutes, a process that Anderson said can take days when compiled manually.

Users of Digital Experience Analytics are getting new heatmap capabilities that help quickly identify and resolve web technical, design and product issues through a visual diagram of customer interactions with a company’s websites. New funnel analytics identify drop-off points during customer sessions on a website and help identify why customers fail to complete their tasks by combining experience data with behavioral clues. New real-time frustration intercepting capabilities enable businesses to identify customers who are having problems online and offer targeted support.

Relief for frontline workers

For workers in retail and online environments, the new Frontline Locations Assist analyzes feedback from surveys, mobile apps, contact centers, online conversations and reviews to create insights and recommended actions to improve customer experience such as replying to customer reviews, checking in on open tickets and rewarding employees.

The features will be available as upgrades to the existing suites. Anderson said Qualtrics recently simplified its packaging and pricing, consolidating multiple point products into three suites and pricing based on interactions rather than user seats.

Qualtrics Assist is available in private preview and will be available in public preview in the second half of 2024. Conversational feedback capabilities are now available in public preview. Intelligent summaries will be available in public preview in the second half of the year. Automated workflows are generally available today.

Photo: Unsplash