Table of Contents
Do not index
Do not index
In 2007, I moved to the United Kingdom.
This amazing life adventure opened my eyes to cultures, languages, and opportunities I had never experienced in America.
Upon arrival, it was clear my American English was not going to cut it in the UK. I immediately started hearing words I didn’t understand, and I’m sure my word choices created some confusion.
I tried to understand these differences and why they existed, and I committed to speaking to my co-workers and neighbors in a language that was comfortable for them. Within six months, I was speaking the Queen’s English, a choice that contributed to many meaningful professional and personal experiences in the UK.
While I lived in the UK, I traveled to many European, Asian, and African countries. Unfortunately, I couldn’t easily adapt to all of the different languages and relied heavily on locals who could translate. I was grateful that so many people were willing to try to communicate in my language. It was a humbling experience.
During my five years in the UK, I often yearned for a piece of technology that could translate in real time—one that would enable more interesting, relevant, and meaningful conversation.
Since then, a new age of technology has dawned: machine learning, AI, and bots. With these innovations, I am starting to see my desire come to pass. As I learn more about these capabilities, my thoughts naturally gravitate toward my profession, customer service and success.
This technology is creating opportunity for companies around the world to break down language barriers. It reminds me of the Babel Fish, the universal translator in Hitchhikers Guide to the Galaxy . That type of innovation could help us in the CX industry as we engage and help customers who enjoy our products every day. That would be simply amazing, right?
The Experiment
I decided to double-click on this idea and did a beta test of what I call “Bots and Babel Fish.” The simple idea was to build and train a bot to understand a different language from the one in which the customer wrote their email—and then to create a set of workflows that would enable that bot to handle a communication exchange between two humans who write in two different languages.
It was not enough just for the exchanges to happen. My team and I needed the bot to be 98%+ accurate in translation, and customer satisfaction could not suffer. With all of these things considered, we built a business case to answer the below questions with the bot language translation beta:
● Will customer satisfaction stay on par with native language support?
● Will using bots improve response and resolution time?
● Does it provide any cost savings back to the business?
What We Learned?
In short, we found that all questions were answered yes, but not for all languages and not right away. A key takeaway from the many examples offered at the 15th annual Frost and Sullivan Executive Mind Exchange in Marco Island in April was that bots and AI are a long play. They take time to set up, time to teach, and time to deliver value. We found the same to be true with Bots and Babel Fish.
Overall, it took us about six months to set up the project and another six months to have enough credible data to test our hypothesis. Here are the five big “Aha” things we learned:
1. Asian languages are super hard: We were not able to get them above 65% accuracy, so we chose not to include them in the beta. In time, I believe this capability at 98%+ accuracy will be possible, but more innovation is needed.
2. Spanish is not Spanish everywhere: Spain’s Spanish is not the same as Argentina’s or Mexico’s. We had to train our bot differently based on the country of the customer. Once we made this tweak, we saw 98%+ accuracy and were able to operationalize.
3. Common issues get the highest customer satisfaction: Our beta covered all of our customer issue types—from the most mundane to the most complex. We found that the more complex issues were not able to hit customer satisfaction levels on par with a native writer.
4. It is best to start small, iterate, and then scale: We focused our beta on one language to start, then we tried five, and scaled after that. This increased the timeline of the test, but it ensured we maintained a stable customer operation and mitigated unnecessary commercial burdens.
5. The Bots and Babel Fish works: However, not yet in all cases. We resisted the urge to use our “shiny ball” everywhere: we implemented the technology where it worked well but not in areas where it would not achieve all of our CX goals and visions. In other words, we found great cost optimization in this project, and we were not willing to let cost dictate our decision making if the quality was not there.
Where To Go From Here?
Now your next question might be, “Who did you use, and how did you do it?” It is a great question. There are many different providers and CRMs that can help you duplicate our success. If you’re interested, I’ve listed a few companies you may want to reach out to:
2. Bold 360ai
3. WhosOn
These options alone, newer startups, and/or all of them used together with an established CRM can put you on your path to testing it out for yourself. If interested further, feel free to reach out to me via email.
Emerging Trend
As you consider your AI language bot options, you may also want to consider an emerging language support trend: gig-economy customer service and language support. The magazine Language Wire shares some great insights in an article with the same name.
The great advantage of gig-economy workers is that they are native speakers who can deliver the same quality as your center agents at less cost, as they are not burdened by the same overhead cost structure. Additionally, they are more flexible and are willing to work in a way your customer demand curve dictates vs. standard minimum payments to a language support vendor.
All in all, Bots and Babel Fish beta test was a great learning experience. It taught us a lot about bots, AI, and language translation. It took patience for us to test our hypothesis, and it took great discipline to only implement high-quality use cases.
Over the next few years, we are going to continue to see disruption and innovation in the language and communication area. I will always savor my time in the UK. And yet, I am insanely jealous of the capability that will exist in 20 years. We will see relevant, meaningful, and enriching conversation, regardless of the languages we speak.
One thing is for sure...there is a great future ahead of us with bots and humans working together!
Written by
.jpeg&optimizer=image&quality=80&width=280)