Fang Cheng, co-founder and CEO, Linc Global
By now, personalized marketing powered by artificial intelligence (AI) is all but mandatory for brands to prove their relevance. To truly differentiate their offerings in a crowded field, sellers must apply this one-to-one intelligence to the new realm of conversational commerce, and execute flawlessly to deliver the service shoppers now expect.
Customer reviews, social media, and mobile have contributed to the rise of conversational commerce, in which shoppers and brands engage in a two-way exchange of information. More and more platforms now enable this conversation, from Facebook Messenger to voice search through intelligent agents such as Siri and Alexa.
As a by-product of this phenomenon, consumers now expect instantaneous responses at the speed of live conversation. Of the 39% of U.S. shoppers who use social media to interact with brands, 44% expect a response within an hour, according to Microsoft. On Facebook, brands must respond to 90% of queries within 15 minutes to earn a badge for responsiveness.
To serve shoppers’ need for speed, brands are turning to machine-powered intelligence to handle basic queries, such as requests for store hours or questions about product SKU options. Technology researcher Gartner predicts that by 2021, 7 out of 10 companies will rely on AI to boost productivity. Within the retail sector, 23% of companies are already using machine intelligence to enhance customer service, according to consulting firm McKinsey.
The struggle to apply AI to shopping
In response to this demand, more and more vendors are touting AI-enhanced solutions. But it’s been a bumpy ride so far. Stories of poor AI-driven experiences are commonplace; some are merely annoying, such as persistent remarketing ads for products already purchased, while others are more sinister—such as instances of algorithms displaying racial and gender bias.
Even the tech giants are struggling to apply AI to shopping. Technology researcher Forrester tested the intelligent agents that power “smart speakers” with questions across six commerce-centric topics and found that just 35% of queries were answered. The machines struggled with context and failed to arrive at direct responses.
As the technology improves to close the gaps, brands with quality AI solutions built on foundations of sound data will begin to realize critical gains in 2020. For others, the seams will begin to show—and customers may defect if they lose patience with poor automated experiences. To maximize the success of AI implementations:
Avoid the “black box.”
Merchants should build accountability and transparency into AI-powered offerings—both internally and when it comes to vendors’ technology solutions. That means creating internal guidelines for data governance and ensuring AI-driven results are explainable and provable.
External vendors’ toolsets should accommodate business rules, segments, and other pre-existing constraints, and—given that merging data with analytics is companies’ number one AI data priority, according to consultant PwC—should provide meaningful metrics for success.
Data quality matters.
AI tools are only as good as the data the machines interpret, so merchants must collect, parse, label, and clean up information before it reaches the AI layer. Selecting the right data points is also crucial; sellers should focus on the input that’s most meaningful to the AI-enhanced task at hand.
Test for consistency.
As shoppers become accustomed to voice commands and chatbot interactions, they expect consistent responses. So while the ability to personalize recommendations via chatbot is an exciting prospect, it’s just as important that requests for store directions are delivered uniformly over time, and with the right calls-to-action, such as click-to-call functionality for the local outlet’s phone number and links to GPS mapping.
Be transparent about human/AI transitions.
Merchants should clearly delineate the types of interactions AI-powered services can handle, and then identify and test a variety of escalating situations to ensure a seamless transition to human help. Since shoppers react negatively to machines masquerading as humans, according to SAP, chatbots and avatars should be explicitly identified as such, with handoffs to humans clearly flagged.
Build a foundation of privacy.
As California enacts the nation’s strictest data privacy law in 2020 and the impact of Europe’s GDPR begins to make a dent in company earnings—including a fine for British Airways totalling more than $200 million—integrating privacy controls into AI-powered conversational commerce is a must. That means developing a succinct privacy statement summarizing data collection practices and obtaining consent and using it in live chat and social messaging, with links to more information and controls always accessible.
AI and the ecommerce advantage
As AI technology improves, brands with solutions that deliver accurate, relevant conversational commerce interactions will stand out in a crowded field, engaging shoppers and earning trust, sales, and loyalty.
Linc Global provides a customer care automation platform.
Favorite