Many experts predict chatbots will revolutionize marketing, public relations and customer service.
Some say the revolution is already underway. The chatbot market, valued at $190.8 million in 2016, is estimated to reach $1.25 billion by 2025, according to Grand View Research.
Organizations in just about every industry are using software applications that mimic human interactions and answer questions on websites and on messaging services such as Facebook Messenger, WhatsApp, Viber and WeChat. Chatbots are showing huge promise for marketers—especially in terms of customer service.
Unlike human customer service agents, chatbots are always on duty and ready to respond. Also, many customers—especially younger, digital natives—prefer to communicate through apps rather than by phone or email.
“Implementing a well-developed chatbot auto-response system for the most common questions from your niche or industry can help to improve response time and customer satisfaction, and ensure customer retention,” notes Gary Vela, CEO of digital marketing agency Web Daytona, in Forbes.
How to build a chatbot
There are two types of chatbots, explains software developer Anadea. “Simple” chatbots are programmed to respond to questions based on specific keywords. If the user’s question lacks a recognized, pre-programmed keyword, the chatbot will not understand the query.
“Smart” chatbots use artificial intelligence (AI) to analyze customers’ questions to learn and improve over time. However, improvement takes time—and you run the risk of alienating customers during your chatbot’s learning phase.
There are other potential pitfalls, too. Chatbots often fall short of human-like conversation or misunderstand questions. Also, some customers might flat-out refuse to communicate with a robot.
Before deploying a chatbot, Anadea recommends these steps:
1. Define the goals. What should your chatbot do? Clearly establish the list of functions your chatbot needs to perform. (In programmer lingo, these are the “requirements and specifications.”)
2. Choose a channel. Be where your clients prefer to communicate — your website, mobile app, Facebook Messenger, WhatsApp or other messaging platform.
3. Choose the creation method. Anadea recommends business owners hire a programmer experienced in creating chatbots. Others suggest that staff programmers can use tools, such as Dialogflow or Wit.ai, to build a bot. (Adventurous non-programmers can also give it a whirl.)
4. Create, customize and launch. Tweak the bot’s algorithm, develop a database of answers, and test your chatbot’s work before showing your creation to actual customers or prospects.
Chatbots typically answer common customer service questions, but organizations are developing highly creative uses for AI software.
Skincare brand Pond recently launched an AI chatbot that matches people to the right skin products. Pond found that its chatbot delivered better results than its existing digital channels.
In 2017, in the aftermath of Hurricane Harvey, the nonprofit Direct Relief built a chatbot to handle the deluge of Facebook Messenger requests. The bot dramatically cut response times, and it helped flag urgent requests that required human assistance. (Direct Relief’s bot is now available free to other nonprofits.)
Flower delivery service 1-800-Flowers created a chatbot that helps users send flowers and gifts directly from the messaging app, instead of through the company’s online store, notes Square Capital. The bot also makes gift suggestions, processes orders and sends shipping updates.
In partnership with a company called ModiFace, beauty retailer Sephora created a Facebook Messenger chatbot dubbed ColorMatch that helps customers find lipstick that matches the shade in a photo they’ve uploaded. Customers can also upload a selfie photo to “try it on.” Sephora Reservation Assistant books appointments to reserve a makeover at Sephora stores. The appointment bot has achieved an 11% higher conversion rate compared with other channels used to book in-store makeover appointments.
Swedish fashion retailer H&M, meanwhile, quizzes customers about their style preferences and then offers clothing recommendations tailored to their tastes. Users can also share a piece of clothing they like, and the bot will use it to select a complete outfit.
How about you? Have you successfully deployed a chatbot for your business? Please include your recommendations—or reservations—in the comments below.
A version of this post first appeared on the Glean.info blog.