The Return of the Chat Bot (And This Time They Work!)
Those of you who have seen me speak over the years know that I love to give examples of bad self-service technology, and nothing has received more of my stage time than the terrible early products for automated chat dialogs, aka chat bots, response robots, or avatars. Call them what you will, they were bad. In fact, one of the largest US wireless carriers did a huge launch a few years ago with a certain chat bot with a female personality, which was derided in the press as a “repulsive cockroach-like alien” because of its inability to answer even the most basic question.
So much for making service a differentiator!
Why were the early chat bots so bad? Mainly because the underlying technology was so poor. They used “answer pairs,” which required that you fill a database with every question a customer might ask, along with the correct response to the question. If the customer didn’t word their question in exactly the right way, they were out of luck. In fact, I have been using the online robots from 2 companies as examples for at least 5 years, asking this question: Do you offer factory tours? Neither one can answer it, telling me to rephrase my question and try again. (I will avoid giving the name of the companies or the providers of the underlying technology in this posting, but shoot me an email if you want to know!)
Well, the times they are a-changing. I have been hearing from sales teams at customer service vendors, and from SSPA members, that demand for these tools is rebounding. Why, you may ask? Here are some reasons:
- Customer experience. Companies trying to create a fantastic experience for customers are finding that having a friendly personality greet the customer on the main webpage and ask, “How may I help?” goes a long way in quickly guiding the customer to whatever it is they need.
- Idiot proof. Chat bots are great for novice customers who may be afraid of self-service options like diagnostic trees, but they also work well for experienced customers who ask detailed, specific questions.
- Ease of navigation. Let’s face it, today’s company websites have too much information. Having a friendly bot guide you to the correct area right away saves minutes, if not hours, of navigation time, and the customer is likely to return to the site again.
I had a visit this week with Marcel Smit, CEO of Q-go, who was in town from the Netherlands for meetings with partners. Q-go is well known in Europe, but so far they haven’t concentrated on breaking into the US market. And that is a real shame for US companies.
I was first introduced to Q-go about 3 years ago while at a conference in the UK, and subsequently met with the Q-go team in Amsterdam, and even did a presentation at a customer event. I wrote a case study while at Forrester about their success with the UWV, the the Institution of Employee Insurance in the Netherlands (similar to our Social Security Administration). In the first year, Q-go answered over 650,000 customer questions (yes, the UWV calls them customers, not ‘constituents’), and customer adoption was growing every week. (If you are a Forrester client, you can read the full article here.)
Q-go offers an intelligent NLS (natural language search) engine that understands the intent of customer questions, can make inferences from questions based on usage patterns, and–here’s the really cool part–they do this for 8 languages! (US based NLS engines tend to only provide NLS for English, reverting to text search for other languages.) Also, Q-go has built vertical versions that already understand common terminology in the following industries: banking and insurance, healthcare, telco and logistics. An impressive list of customers includes ING Bank Nederland, Deutsche Telekom, Deutsche Post, Telefonica, and DHL.
Moving beyond Web self-service, Q-go is investigating products for additional channels, including voice and email, so stay tuned for a multi-channel Q-go!
Do you have a chat robot story to tell? Add a comment or send me an email!