What does digital comms think of chatbots?

Jason Sumner reveals what corporate digital communicators think about this much talked about emerging technology

I don’t have many conversations with chatbots on corporate websites yet, but I’m certainly talking a lot about them at the moment.

Digital corporate communicators, perhaps having just seen a shiny pitch from a software company, are asking us – do they actually work? Is anyone else using them? Should we believe the hype?

Actual chatbots in use on corporate websites are few and far between. They are so unusual, and there is so much interest, that we try to write about them when we see one.

Enel, the Italy-based power company, has a Facebook Messenger news bot on its corporate site, as we noted in this recent piece. It is called ‘Elen’ and helps visitors find news, press releases and stories, and introduces the company’s sustainability activities. The bot shows some promise, but those wanting detailed information are better off sticking with the corporate website.

Tying the bot to Facebook Messenger could be risky too, since not all corporate website users will have it installed. Enel may just be following the path of least resistance, however, by relying on Facebook Messenger, which had 33,000 active chatbots as of January 2018, according to IBM.

We’ve seen other examples, focused on careers: T-Systems, a Deutsche Telekom subsidiary, has a German-language bot called ‘KATY’ on its website careers section. UPS, the US logistics company, has one on its careers site, as does US telecoms giant Verizon. These three are not very helpful and a little cold (the Verizon one keeps answering questions by referring people to the careers website, but at least it does that much); they do not perform nearly as well as a good online FAQ, for example.

That’s what we’ve found, but what are digital communicators saying about them? There are more questions than answers, but our clients see the potential to save money on admin and deflect queries, with the following caveats:

  • Chatbot success depends on the quality of tagging, metadata and taxonomy working in the background. Artificial intelligence should allow systems to ‘learn’ and improve results. Eni’s sophisticated search engine also relies on complex tagging in the background, and indeed, chatbots can be seen as another form of search (It is worth pointing out that sites have had years to get tagging and metadata right for internal search, but it still doesn’t work very well on most sites.)

  • They work best in a limited context, so the possible range of questions and answers are contained; in corporate communications terms, this means interacting with a specific set of stakeholders with easily predicted queries (hence the focus on jobseekers) or telling customers where a petrol station is; rather than an all-purpose ‘ask me anything’ bot.

  • If chatbots become more widespread, the technology will raise new reputation risks. Companies, for example, will need guidelines for appropriate tone of voice, race and gender.

Finally, no one we have spoken to has launched a chatbot, yet. But there are some interesting things happening internally – eg, one has launched chatbots for employees, with a view to rolling them out externally when the technology is proven.

Jason Sumner

The theme of our annual conference on June 18th and 19th 2019 in Berlin will be ‘Online communications, tomorrow and today’. We’ll be doing many more pieces on the future of online communications in the next weeks and months.