Input and experience of higher ed professionals helps a chatbot perform at a higher level over time by:
- Looking at past events
- Determining if desired outcomes were met
- Moving toward more positive outcomes
There are a lot of blogs and articles from software companies claiming that their chatbots are capable of ‘out-of-the-box’ machine learning. The truth is, while actual machine learning is an impressive feature of a few chatbots, it takes a certain level of human effort to accomplish this phenomenon.
With Harmony, Mongoose’s intelligent chatbot, our expert team higher ed experts handles much of the heavy lifting on behalf of your college to help your bot constantly improve. This part of the process gives your staff an incredible advantage in helping your constituents reach their goals.
So, how does Harmony improve over time? Here comes the science.
Chatbots answer questions, collect data, and offer guidance. With Harmony, our team regularly analyzes a college’s conversations, locates areas for improvement, and “trains” the bot to respond better.
Having human beings with higher ed knowledge and the ability to interpret context makes this process efficient. A human element allows us to understand the significance of new topics that keep coming up in conversations with the bot or topics that we expect will become more frequent in the future. By addressing these topics earlier on in the conversation, you can get ahead of questions and be proactive in improving the bot.
A hypothetical example:
Your school is now offering a new academic program. Students keep asking your chatbot about the new option. Your Harmony team identifies this as an area that needs to be addressed now and with the addition of future academic programs.
What happens when the chatbot doesn’t have the answer?
Every Harmony playbook is built with specific objectives and audiences in mind. For example, when a student visits one of the ‘Financial Aid’ pages of a college’s website, Harmony will engage with the student to provide answers to any questions that student might have about Financial Aid.
Because the subject matter is often very specific and well-defined, Harmony can predict with a high degree of certainty what those who engage might need. Proper answers are provided the majority of the time.
There are options to identify and triage the instances when website visitors do not get the answers they need. Chatbot responses to this selection could offer a new menu of items, or ask for the site visitor to type what they are looking for. Harmony gives schools the option to allow the chatbot to book a meeting with a staff member or escalate the conversation to live chat. Harmony also seamlessly captures information for future follow-up.
After providing a response, Harmony can ask a site visitor if it provided the information they needed, offering a ‘thumbs up’ or ‘thumbs down’ option. When a ‘thumbs down’ is selected, the bot can offer to try again, or attempt any of the above fallbacks for human engagement.
‘Thumbs down’ reactions and items that the bot could not understand are aggregated by the Harmony team. These results are used to make Harmony stronger and more prepared for those questions in the future.
So, like humans, machines can learn from their mistakes and grow stronger.
Want to learn more? Schedule a demo to see what Harmony can do for your institution.