Intelligent Virtual Assistant Case Study

Challenges

PepsiCo is one of the world’s largest food & beverage companies. With billions of consumers worldwide, PepsiCo was looking for innovative ways to improve the consumer experience, proactively share information on their corporate sustainability efforts, and help their consumers find retailers where they could buy their favorite PepsiCo beverages.

Organizations often opt for customer service representatives to provide a personal experience; however, this can be costly to provide at scale and lead to lengthy wait times during unpredicted peaks in contact volume due to their finite capacity. Alternatively, self-service channels are scalable in nature, but, according to Gartner, only 13% of consumer contacts are resolved fully using self-service solutions. Additionally, consumers dislike traditional self-service channels such as touch-tone interactive voice response phone systems.

 

In order to share more information on their sustainability efforts and help consumers locate retailers where they could buy PepsiCo beverages, the PepsiCo Consumer Relations team sought to leverage artificial intelligence (AI) and machine learning (ML) to exceed consumer expectations with a better experience.

PepsiCo understood that while AI is powerful, it can be very challenging to train and reach the intelligence level needed to speak on behalf of a brand. For example, PepsiCo’s number one consumer contact request is where to buy a specific beverage product/flavor/packaging. There are over 2 million combinations of ways in which people can ask this question. Such challenges require advanced AI-powered tools and expertise. Historically, many organizations have experienced this difficulty and spent thousands of people hours to undergo this process with little success.

We choose XAPP AI as our partner because of their deep expertise in creating intelligent virtual assistants. We had some previous experience with XAPP AI with one of our brands and were hoping to leverage that expertise for our capability. Leveraging their expertise & framework that uses a data-driven approach, we created automated conversations that align with our business objectives and allow us to learn through the process. Their approach overcomes many of the challenges that occur when building complex omnichannel conversation models and helps us to continue to deliver quality consumer experiences. 

Elena Parlatore, Senior Director, Global Digital Consumer Experience.

Solution

PepsiCo partnered with XAPP AI to create an omnichannel intelligent virtual assistant (IVA) that automates conversations previously handled by live Customer Service Representatives. The IVA “Ask & Get” experience understands and responds to consumer questions using natural language across PepsiCo.com, Facebook Messenger, Amazon Alexa, and Google Assistant.

The IVA was designed and built following XAPP AI’s Optimal Conversation Framework™, a data-driven process that continually progresses an IVA closer to its optimal state for achieving business objectives and outstanding customer experience. XAPP AI’s Optimal Conversation™ Studio powers the Optimal Conversation™ Framework and the PepsiCo IVA enabling complex conversation models to be deployed and managed across multiple digital channels. Optimal Conversation™ Studio is powered by Amazon Web Services (AWS), allowing for unlimited scalability of the intelligent virtual assistant as it grows. 

“Optimal Conversation™ Studio makes it easy to deliver scalable intelligent AI powered self-service with very complex conversation models. It accelerates time to value and allows customers to gain access 24×7 to intelligent virtual assistants via chatbots and voice experiences.”
—Curt Kolcun, Executive Chairman, XAPP AI.

XAPP AI partnered with the PepsiCo Consumer Relations team to perform workshops where they analyzed PepsiCo’s challenges, selected use cases, and designed, built, and optimized the IVA.

Use Case Selection & Content Curation

In addition to the previously stated PepsiCo goals of sharing information on their corporate sustainability efforts and helping consumers find where to buy beverage products, XAPP AI analyzed many of the high-volume customer contacts, which were frequently asked questions around product information, nutrition, and other related topics. XAPP AI captured knowledge from existing content sources including  PepsiCo’s website, knowledge base, and over 400 FAQs, which accelerated the process. Additionally, XAPP AI curated paragraph form content about PepsiCo’s innovation with sustainability in the Consumer Packaged Goods (CPG) industry and subsequently taught the IVA how to respond to questions based on it. XAPP AI and PepsiCo collaborated to create a product roadmap for the IVA starting with these use cases, leveraging PepsiCo.com, Facebook Messenger, Amazon Alexa, and Google Assistant as the initial channels to deploy the solution, with a plan to enhance and optimize the IVA over time.

User Experience Design & Information Architecture

XAPP AI’s leading conversation designers curated the solution for each use case, accounting for the nuances of each channel and device type such as mobile devices and smart displays. The solution contains over 40,000 cells of curated data and hundreds of visual designs. The information was architected and documented following the Optimal Conversation™ Blueprint, developed specifically to account for the nuances and complexities of multi-channel/multi-device conversation design and development. It was also important for PepsiCo to allow consumers to connect with a live customer service representative as needed. Because of this, the IVA is designed to have the intelligence to recognize when a consumer needs additional help from a human and hands the consumer off to a live representative.

Conversation Model Training

PepsiCo and XAPP AI leveraged the Optimal Conversation™ Studio to build and fine-tune the conversation model for each channel and device type. The biggest challenge was creating a conversational wrapper around PepsiCo’s product locator API to help consumers find the nearest place to purchase a particular product. This API was not originally meant for conversation and needed to be able to respond to over 2 million different combinations of ways people could ask for a variety of PepsiCo products. 

For example, someone could say “Where can I buy Mango bubly in a 12 pack?”

This Ask & Get conversational interface to our “Where to buy” products API is best in class and helps to address the highest volume request we get from our consumers.”  XAPP AI did an outstanding job tackling our challenges

Donna Laskowski, PepsiCo Senior Director Consumer Relations

Prototype Testing & Analysis

XAPP AI’s Optimal Conversation™ Framework takes a rigorous data-driven approach to each step of the conversational product lifecycle to move the product closer to its optimal state to meet an enterprise’s business objectives and improve the consumer experience. XAPP AI designed and built prototypes while performing usability testing with target end users to test the designs and conversation models. The team iterated on the designs and models based on the data from the test results.

Data Analysis & Optimization

With XAPP AI, PepsiCo can analyze live data from consumer requests and make decisions on high priority content to optimize based on inquiry frequency and the relevance of the current optimized results. By leveraging XAPP AI’s Human-in-the-loop functionality in Optimal Conversation Studio, even nontechnical personnel can identify, prioritize and teach the AI to respond based on how customers communicate with PepsiCo.

Results

“This work and our partnership with XAPP allowed us to really push the envelope on how we can improve the consumer experience. Working with XAPP AI, we created a truly innovative experience that lifts our brands and deepens our relationships with our consumers.” 

Kathy Okrzesik, Vice President Consumer Relations

“The PepsiCo IVA built on Optimal Conversation™ Studio helps transform PepsiCo’s consumer experience to meet the new realities of business.”

Pat Higbie, Co-founder and CEO of XAPP AI.