Flyin Stopped Using Email For Support Entirely After Moving To A Chatbot

Automation for OTAs

Fady Jabrah manages Customer Experience for Flyin – and in doing so handles customers from the UAE, Oman, Qatar, Kuwait, Bahrain, and Saudi Arabia. And when in 2018, Cleartrip acquired Flyin, the companies managed over 65% of booking in the Middle East and North African (MENA) region. Such a vast international presence required an equally adept and scale-ready partner. The aim was to make customer support for Flyin and its 200,000+ monthly website visitors and 1,000,000+ app users quicker, smarter, and more effective. And so, with the help of Verloop.io’s Head of Forward Deployment, Sujoy Correa, the Flyin team set to creating an OTA conversational automation system unlike any other.

Of course, the first response time has improved. So has the handling time for chats. I also have more visibility on all interactions between agents and customers. This allows me to control agent bandwidth, in terms of the number of chats. I have more data for everything customer related, so that helps us a lot.

Fady Jabrah, Customer Experience Manager, Flyin

1. COVID-19, The Great Equaliser

The COVID-19 crisis wreaked havoc on the global economy. Countries dove into deep recessions, manufacturing came to a standstill and several industries underwent deep turmoil. But one industry received a particularly brutal beatdown – Travel. As governments closed their borders and implemented strict shutdowns to curtail the spread of the coronavirus, revenues for companies in the travel sector plummeted. While companies in other industries could’ve simply shuttered operations during such times, OTAs had a new problem to face – a sudden inundation of support tickets as customers canceled their bookings. This, coupled with a sudden shift to work-from-home meant that traditional human-only support systems started buckling at the knees.

Flyin required a partner that could help automate support ticket resolution while prioritising remote-working, multi-linguistic support, and at-scale operations.

Verloop.io’s conversational automation solutions allowed Flyin to automate queries related to booking cancellations, confirmations, amendments, information, refunds, and general inquiries. Over 84% of the thousands of customers who started a conversation with Flyin’s bot reached the end of the chat flow – meaning that customers had their issues successfully resolved with little drop-offs. The bot collects a user’s name, location, email ID, and booking ID. Post this, it provides users a list of queries and connects them to the respective agent.

2. Fastest First

The problem with support channels like emails is its incredible slowness. When a user sends a company an email, he’s expecting an immediate response. If he doesn’t get one, he’s going to go about his day and only revert to any emails that follow several hours later. By cutting down on the first response time, Flyin discovered that it could also reduce resolution time by up to 80%. Instead of having FRT that ranged from 30-40 minutes via email, with a bot Flyin reduced its first response time to 17-20 seconds. The bot also reduced the average response time to under 40 seconds.

Did quick responses have a direct impact on query resolution times? Absolutely, it did. Support ticket resolution time, which for an OTA like Flyin used to be hours and even days using email, fell to a scarcely believable 32 minutes using the bot. Post these results, Flyin has begun to phase out its email support functionality entirely, moving now solely to a chatbot-only system of support.

3. Tongues of Foreign Lands

Flyin services thousands of customers from a variety of countries every month. They’re incredibly proud of this diversity, but it does come with its own challenge. In fact, 4 in every 5 conversations that were started on the Flyin bot were in a non-English language. Verloop.io serves FAQs in Arabic through SOTA multi-lingual Universal Sentence Encoders and BERT which helps in automating client-specific FAQs. The bot also features a NER system on top of flair which can identify names written in Arabic during conversations. Additionally, the bot also operates in compliance with data and privacy regulations in each individual country.