Support Skift’s Independent JournalismMake a Contribution Now
There’s been a surge of meetings and events that are integrating chatbots into their communication platforms in 2017, powered by artificial intelligence engines to provide automated 24/7 brand engagement.
In Austin this year, South by Southwest’s bot “Abby” answered more than 56,000 questions submitted by attendees in the SXSW Go event app. Other recent event chatbot launches included “Frank” at IMEX Frankfurt and “Betty” at BizBash Live LA.
As was the case with event apps just a few years ago, there’s a lot of confusion today about what chatbots are, and what they can provide for both meeting planners and attendees.
According to Salt Lake City-based Sciensio, which developed IMEX’s Frank and BizBash’s Betty bots earlier this year, “Chatbots are software programs, which can include AI components, designed to interact with people over messaging apps.” Eventually, in all likelihood, all consumer and industry brands will have AI-powered chat platforms to engage their audiences, and as artificial intelligence evolves, they will be able to provide more customized information to the individual user.
Chatbots, then, offer the first glimpse into a future where meeting and event planners can deliver mass personalization at scale.
In effect, chatbots are like live chat boxes you see on many consumer-facing websites, in the sense that they’re live in real-time, just not with a live person. They’re not to be confused with general chat messaging with brands. That’s common with channels like Facebook Messenger, where consumers have the ability to message a company or organization, and then a live representative responds hours or days later.
The rise of AI bot messaging between consumer and brands kicked into high gear in 2016 in the fashion, food, entertainment, media, and retail sectors. In April 2016, for example, Taco Bell launched TacoBot on the Slack messaging platform, and Fandango and CNN unveiled chat channels on Facebook Messenger. And then in September last year, Burberry and Tommy Hilfiger both launched chatbots on Facebook Messenger to drive online and offline shopping sales, supplemented with menus detailing everything from store locations to corporate responsibility mandates.
Search for Tommy Hilfiger in Facebook Messenger to use the TMY.GRL bot, which first asks users a series of personal fashion questions in order to suggest seasonal outfits customized to those preferences, ranging from sporty to boho.
Today, brand chatbots can co-exist across many different channels, and that’s a big reason behind their growing adoption and success, including within the meetings and events industry. Meaning, planners can build chat functionality so attendees can engage with them and their partners via event websites, event apps, mobile text messaging (SMS), Facebook Messenger, Twitter Direct Message, WeChat, Telegram, Skype, and others.
This summer, Sciensio expanded its AI event chat product to work on Twitter, which can be explored here. Twitter opened its platform for chatbot development in April this year, and brands like Shake Shack are taking advantage of that to capture users beyond Facebook.
Presently, WhatsApp doesn’t have a developer platform for chatbots, but that’s expected to change as parent company Facebook seeks to monetize WhatsApp’s massive user base. Also, at its F8 developer conference in April, Facebook launched Facebook Messenger 2.0 with a new “Discover” function to search through the growing numbers of chatbots.
Meeting planners, meanwhile, are now developing AI chat channels across a variety of digital platforms to access a greater variety of audience segments. For example, BizBash designed Betty for its event website, Facebook Messenger, and SMS. You can still see Betty in action and take her for a test drive on the BizBash Live LA website, where she explains that she’s presently gearing up for BizBash Live New York.
IMEX Frankfurt designed Frank for just its website and Messenger, without including SMS due in part to the wide range of cellular messaging fees across Europe. The goal for Frank was to extend the engagement window leading up to the event by answering general questions about Wi-Fi, parking, transportation, registration, etc., with instant responses available anytime, anywhere.
The Future of Event Chat
Sciensio co-founder Chris Colleran, who has developed emerging technologies for a range of blue-chip clients including Microsoft and Home Depot for over 20 years, has now developed chatbots for almost 40 events in the last year. He explained that his company’s bot architecture is integrated into an event’s registration process, where attendees have the option to include their phone numbers and preferred social channels. Once attendees completes their registration, they automatically receive a chat message introducing the bot, which provides a menu of options.
“We have less than half a percent opt-out rate, and less than one-tenth of one percent following the first message,” Colleran said.
Kristi Colleran, another co-founder of Sciensio, explained, “It will say, ‘Hi, I’m the name of the bot, I can help you with the following things for the event.’ It’s a nice time to introduce attendees to the bot because they’re excited, they just registered, so now they have some planning to do, and the bot can help them with some of those things.”
Events are also expanding how they use chatbots. For example, a Sciensio bot was used during Savannah’s Song, a recent fundraising event to benefit St. Jude’s Hospital, and the chat channel helped increase funding levels.
“We enabled the bot so people could bid on auction items, and if you got outbid, the bot would come back to you and ask if you’d like to bid again at the next increment,” said Chris Colleran. “We have a very high engagement rate. Over 80 percent of bids came through the bot, and Savannah’s Song saw a 20 percent increase in average auction prices over the previous year. Overall, we had an engagement rate of 67 percent of attendees who used the bot, which was a lot higher than we were expecting.”
Kristi Colleran added that planners can also create ad hoc messaging with chatbots when they want to address specific issues to event audiences in real time. In one case, at BizBash Live Florida this year, there was a fire alarm in the convention center. So, the planners delivered a message via the Betty bot to alert everyone when it was okay to return.
Leading up to an event, she works with planners to educate them about the chatbot, while at the same time, she programs the bot with information pertaining to a specific event. As an event proceeds, sometimes there are certain questions submitted to the bot that inform planners about unanticipated needs, like additional parking demand or bag check services. In that case, bots can provide real time feedback to help planners provide solutions for different challenges on the fly.
One of the biggest challenges for creating successful bots is identifying beforehand all of the different ways people will ask questions. As Kristi Colleran is preparing the bot with the information required for any given event, that is called the “supervised learning” stage. It’s then when the bot is programmed to understand the basic intentions behind audience questions specific to the event, and the responses they will likely require.
“When the bot goes live and it’s on its own, that’s the unsupervised learning part, or machine learning,” she explained. “It’s receiving messages from users, translating those, and then responding based on what it’s learning with every interaction. We monitor that as well, and look to see where we might need to intervene at times.”
For the BizBash events, about three percent of attendees using the bot wanted live support at some point during their engagement with Betty, but BizBash and Sciensio staff jumped in about 10 percent of the time to offer additional context to questions.
The future of event chatbots relies on the future of artificial intelligence. How event tech companies bundle different AI products together to develop bot platforms is their secret sauce, so to speak. Sciensio uses Microsoft and Google AI components and cognitive services for natural language understanding and processing, among others.
For Kristi Colleran, the future of AI chat development is making the engagement feel natural. The art of that is anticipating and programming as many follow-up questions as possible to keep the conversation moving between users and brands.
“We really focus on trying to give people a human experience, so we talk a lot about conversation design,” she said. “We want you to feel like you’re talking to a human but know you’re talking to a bot.”
“There’s a difference between a human conversation and a dry Google response to an online query,” added Chris Colleran. “If you ask Google for the weather, it comes back with a bunch of links. We try to wrap it in a conversation because that helps with engagement, and that’s what creates good impressions for event planners and their partners.”