7 Common Errors when Implementing a Chatbot

Eighty percent of all companies should have a chatbot in place within the next two years (1). It’s a major challenge. But a bot is much more than just the latest trendy gadget: it’s a bona fide tool that must be implemented with care. Here are seven errors you’re at risk of committing if you aren’t properly prepared.

Error #1: Launching a chatbot without a precise goal

Your chatbot is a full-fledged tool that needs to respond to an identified need: relieving helpdesk overload, handling client relations, etc. This is an essential condition for it to be accepted by your end-users, whether internally or externally. Otherwise, it’s nothing but a very expensive decoration!

Error #2: Having a virtual agent that’s too impersonal

Seventy-nine percent of users prefer dealing with a human being when it comes to customer relations. No one’s going to feel like calling a bot that’s devoid of all personality, with a monochromatic voice and no emotion. Make your conversational robot a genuine assistant that’s personal and attentive to its users!

Error #3: Not knowing the bot’s target audience

Your bot is to your users what a product is to the market: if it isn’t adapted for a specific target, no one will want it. Be sure to define the personas upstream so that your chatbot’s responses and vocabulary are appropriate for your current and prospective clients.

Error #4: Sacrificing quality for personality

Your bot has an attractive appearance. It’s also endowed with a strong personality and responds tit for tat… But always misses the mark! Don’t forget about value added: it’s also necessary for your chatbot to correctly inform your users. Because a useful chatbot is an effective chatbot. An off-the-wall virtual assistant is entertaining for about five minutes and then quickly becomes annoying. Train your bot so that it doesn’t systematically reply “Stéphanie of Monaco” every time someone asks it whom to contact!

Error #5: Letting your chatbot freewheel

A bot is a tool in constant evolution. It has to be trained and learn as it goes along under the careful tutelage of a Chatbot Manager. Make sure you entrust this job to someone very familiar with the profession, who can regularly nourish your bot.

Error #6: Failing to anticipate a way out

A chatbot that has reached an impasse (because it doesn’t know how to answer a question) tends to ask other questions in an attempt to get the user to be more specific or to rephrase the inquiry. This could give the impression of being in Westworld on a day plagued by technical glitches! Include exits in your dialogue scenarios so your bot will always know when it’s time to redirect the user to a live operator.

Error #7: Neglecting to track your chatbot’s performance

It’s impossible to know if your chatbot is effective (and profitable) if you don’t track its performance. Identify the KPIs that are meaningful and adapt them progressively, based on the results. Most importantly, scrutinize any sources of dissatisfaction (lack of knowledge on a subject, inadequate responses, etc.) so you can make continuous improvements to your virtual assistant.
To be effective, your chatbot must be in line with your needs, in sync with your end users and in a position to respond (relevantly) to the questions it’s asked. Just like Elsa does for the customers of ÉS Énergies Strasbourg. You can count on Living Actor to guide and support you throughout your chatbot project… And guarantee that you avoid these seven common mistakes!

(1) Cecilia Di Quinzio. “Le marché des chatbots en 10 chiffres”, (in French) Stratégies, June 13, 2017.

Source image: Designed by jannoon028 / Freepik