The Impact of Field Suggester on IT Teams

When we launched Freddy, our artificial intelligence and machine learning platform, in October 2018, the idea was to harness cutting edge technologies for better customer experience. And that has led to the release of several Artificial Intelligence-based features this year across our Freshworks suite of products.

At the beginning of the year, our product team at Freshservice came up with the idea of using Artificial Intelligence to reduce the time taken to manually assign ticket fields. If the system could learn how older tickets were handled, it could then predict field values for new tickets.

Soon enough we launched the Field Suggester, our first artificial intelligence-powered feature in Freshservice.

In about 3 months, the Field Suggester has made more than 500,000+ predictions and we’re excited more than ever before to bring the power of artificial intelligence to make IT teams more efficient.

At Freshworks, it’s been our mission to make powerful software simple and thereby make it easier to adopt.

Here are some metrics that tell us we’re on track.

Our customers love it

Soon after the launch, we observed a positive trend in adoption. A lot of our customers started trying out the feature and were happy about it. The word spread and until now we’ve seen that close to 30% of customers have used the Field Suggester. We’ve also received great feedback with respect to how the feature had saved a lot of time for IT teams.

Getting it right

When it comes to prediction models, it is important to talk about accuracy. When we launched the feature, we did not want to completely automate field prediction. Instead, we wanted some level of human intervention to train the model and improve its accuracy over time. So, we designed the Field Suggester to recommend fields rather than applying them to incoming tickets. The service manager has to go through the recommended fields and apply them to a ticket only if the recommendations were right.

We’re happy to report that in the first three months since launch, we’ve made 500,000 field recommendations with 84% accuracy. At times when the recommendations were incorrect, the model learned from the service manager’s choices and trained itself to be more accurate.

Faster service, better IT

But, what does recommending ticket fields translate to? During the first three months of Field Suggester, we were able to witness a 40% reduction in the time spent on assigning ticket fields. This is a remarkable achievement since it saves a great deal of time for the IT agents and gives them more time to focus on other important tasks or maybe even have a nice cup of coffee.

What’s next?

The success of Field Suggester is a clear indicator that leveraging the right technologies can empower IT teams and make them more efficient. It also made us realize that we’re moving in the right direction.

We also have plans to extend Field suggester to other modules like Problem, Changes, etc. Our future enhancements will also include predicting field values for custom fields created by our customers.

We’d also love to know what you want Freddy to predict next. This will help us a great deal in making Field Suggester better.

Interested in listening to our experts talk about how Machine learning and Artificial Intelligence are transforming customer experience? Catch our special sessions at AWS re:Invent 2019 that is currently happening at Las Vegas from Dec 2 – 6, 2019

Blog cover by Srinivas Dhotre