The ugly truth about chatbots

Image credit: iStockphoto by Getty Images

Fast response, operational efficiency, and better engagement – these are the promises of using a chatbot.

It makes perfect sense. Having a program to answer most of the inquiries can help to reduce staff issues, who can then focus on more value-adding or complex tasks.

Chatbots readily collect data, allowing marketing teams to understand if a situation is developing or a trend is emerging.

Chatbots do not sleep. For a global e-commerce firm, it allows them to respond to customer queries all day long.

Lastly, they learn quickly and etching their wisdom in your corporate memory forever. So, you do not have to spend time re-training staff or depend on a few individuals who have the knowledge and expertise that the entire firm will need to rely on. 

In some functions, like marketing, chatbots are transforming processes and strategies. It offers a conversational approach to developing a lead, enables hyper-targeting that cuts through the social media noise and allows the offer of hyper-personalized products at the right time. For example, products like Alexa is now bringing marketing and sales into living rooms—and consumers love it.

“Chatbots are transforming the way of marketing because it's not providing a new channel but a new way of interaction and a new type of brand experience,” Xiaofeng Wang, senior analyst serving B2C marketing professionals at Forrester.

Why are there so few successes?

While the case for chatbots is very clear, especially for marketing and customer service, successful use cases are few.

According to the Forrester report, which Wang authored and entitled Chatbots are transforming marketing, the reason is purpose—or rather the lack of one. 

Often firms create chatbots as a means to an excite their end customers. They build one, launch it, and expect it to permanently replace their human peers while luring end customers to their novelty.

The problem is that chatbots depend on defined repositories of answers. AI-driven assistants can “learn” over time and build on these repositories to answer complex queries, but they need time. This where most chatbots fail. Often their defined purpose is too wide or unclear.

Forrester cites Singapore’s POSB as an example. They rolled out a chatbot to handle general queries on their products and services. Although it got some press coverage, the chatbot was unable to fully answer queries, reducing the engagement value and fail in converting potential interest into leads.

OCBC’s chatbot, on the other hand, had a very focused purpose—generating home loan leads. And it helped them secure S$10 million in new loans within three months.

Context is another problem for chatbots. While you can program answers to key queries, they are often unable to understand situation to advise on life insurance. AI will help, but this, again, will take time.

However, the biggest problem is firms launch chatbots too early. Forrester noted that when they examined a variety of chatbots, they found that they made mistakes, were unable to solve problems or overcomplicated simple queries that a Google search could have handled. Without a proper human escalation protocol, where someone living and breathing takes over complex or specific queries from chatbots, customers will discard them when the novelty wears off.

“Creating chatbots is not once-for-all effort. Just like developing marketing strategies, marketers should plan a roadmap for chatbots. There will be different phases and generations of chatbots, of course, it won't be static. However, the purpose of a chatbot should be rather stable, for example, can’t be generating leads for home loans today, and answering questions for car insurance tomorrow,” Wang said.

What are the steps for chatbot success?

Forrester advises firms to take a structured approach to developing and launching chatbot marketing strategies. It calls it people, objective, strategy, and technology (POST). 

POST begins by understanding whether the target customers whether they are ready to engage chatbots through their messaging apps. Some nurturing may be required. Next, you need to define the right objective with clear marketing goals. Then you can decide the strategy and technology to use.

If the approach is too complicated or resource-intensive, consider working third-party chatbots. Although they may not work exactly as an inhouse designed one will, it allows you to build on their repository of knowledge and gain valuable experience. Forrester highlighted BBC Player’s market launch campaign on the bus schedule chatbot Bus Uncle as a good example.

“There are a lot of good marketing use cases in the report. Marketing is always evolving regarding new technologies, but fundamentals, like understanding customers' needs and using POST (people, objective, strategy, technology) method to create strategy, are always the same,” Wang said.

Where is the best place for a chatbot to live?

Best of all, we are living in the best region for chatbot evolution—Asia. While China is taking a different route with WeChat, chatbot technology across Asia is quite advanced due to the popularity of messaging apps and a fast growing mobile-savvy customer segment.  

“We have lots of chatbot examples from the Asia market…marketers here are more willing to experiment, [and] the market is more competitive. How to move faster than your competitors, and how to differentiate with them are more likely to be challenges for Asian marketers,” Wang said.

Further reading:

Shaping the customer journey with AI

Mobile foundation vital for B2C marketing success

Will artificial intelligence replace CMOs?