India-based online travel platform MakeMyTrip has been working to integrate generative artificial intelligence into its business strategy. The company kicked off its implementation in May with a deal with Microsoft to use its GPT technology - created from OpenAI’s GPT language models.
Group chief technology officer Sanjay Mohan outlined three use cases for AI the company has honed in on including:
- Text analysis - to analyze text including user-generated content, descriptions and reviews.
- Content synthesis - to synthesize content based on what it identifies as “context or triggers” such as user profiles to identify similar travelers to improve user experience
- Language translation - for voice-to-text translation to make it easier for users to input information as existing keyboards for regional languages are “clunky.”
The biggest challenge, so far, said Mohan, has been language translation. “We are still working [on] getting a good colloquial translation into any of the Indian regional languages,” said Mohan, noting the Indian government is also working on the issue.
The company also offers “hybrid” interaction - voice and visual [screen interaction], which Mohan said isn’t quite “seamless” yet.
Subscribe to our newsletter below
MakeMyTrip started incorporating machine learning a decade ago to help with cross-selling, fraud prevention, discovery and more and is working to do the same with generative AI as the technology progresses. The company, which has 74 million users, believes AI can serve as a growth driver for offering personalized experiences.
Mohan answered questions as part of PhocusWire’s initiative to check in with some travel companies that were the early adopters last year of generative AI as it generated buzz. Responses have been edited and condensed for clarity.
What have you learned since implementing AI into your business plan?
In our 23-year tech journey, a crucial lesson has been the paramount significance of data quality. The success of AI initiatives too, therefore, is intrinsically tied to the quality of the data we work with. Our relentless efforts have been directed toward ensuring top-notch data quality, which we recognize as a foundational element for the efficacy of our data science and AI/machine learning (ML) models. Over the last couple of years, our AI-driven product features have garnered significant success, with offerings such as Zero Cancellation and Fare Lock in the flights funnel, shielding customers from cancellation penalties and last-minute fare fluctuations. Similarly, our personalized/contextual hotel rankings have become immensely popular among users.
Our recent introduction of GenAI into our product has not only helped make user journeys on the platform shorter and more effective but also given our supply ecosystem additional tools to attract the right consumer. We have leveraged GenAI to extract key points from the hundreds if not thousands of reviews that are enlisted on our platform for every property to arrive at a crisp summary contextual to the user’s need-case. This saves precious minutes for the user, helps them make an informed decision while also helping property managers highlight key USPs of their listings all at once. Similarly, our voice-assisted chatbot booking feature enabled in Indian languages have helped make booking flights, etc., more accessible and entirely conversational. We are also using GenAI to curate thematic collections for properties on the platform, which again enhances hotel discovery, thereby bridging demand and supply more effectively.
How has your strategy changed since you first started?
While our overarching strategy has remained consistent, we've made nuanced corrections along the way based on continuous learning. The evolution has been marked by strong emphasis on data quality, an aspect we consider foundational. As we gradually tasted success, the proliferation of AI/ML models became evident. With a more profound understanding of customer preferences, our strategy now extends to superior personalization during both the shopping process and post-sales support. We've recently rolled out GenAI in voice for English and Hindi, with plans to expand into other regional languages, although this is still in the early stages.
What surprised you about using AI?
The realm of AI/ML has witnessed notable growth, especially in the GenAI space, propelled by the advent of "transformers" in deep learning architecture. In 2023, the maturity of ChatGPT and open-source models, combined with the prowess of GPUs and cloud scalability, took the industry by surprise. Particularly staggering was the maturity exhibited in various capabilities, including generating high-quality code, facilitating human-like conversations, understanding content nuances, simplifying complex data into summaries, and comprehending both images and text to offer meaningful and persuasive decisions. The extent of personalization and efficiency achievable through AI also stood out as a remarkable highlight.
What percentage of users engaged with the AI integration you implemented?
Since the implementation of AI/ML, we've witnessed a significant uptick in engagement, with over 95% of our users actively interacting with our AI-powered tools. These tools, ranging from personalized rankings to product recommendations, have not only elevated the user experience but have also had a tangible positive impact on both the CSAT [customer satisfaction score] front and the company's bottom line.
How does your conversion to booking compare to overall user conversion?
The conversion to booking for users engaging with our AI features has surpassed the overall user conversion rate significantly. This notable difference underscores the effectiveness of AI in not only enhancing the user experience but also driving substantial business outcomes. The increase in conversion rates among users leveraging our AI integrations currently hovers around the early to mid-teens. This can be attributed to the personalization and simplification of the shopping process. In turn, this allows customers to swiftly find relevant offerings, solidifying our position as a travel super app – a comprehensive one-stop-shop for all travel-related products.
What types of queries are coming through?
Our queries encompass a broad spectrum, ranging from fundamental product inquiries to more intricate matters such as travel advisories, visa policies and destination discovery. Users also engage in extensive conversations, seeking personalized travel recommendations and assistance with itinerary planning. While our GenAI journey is still in its early stages, a diverse array of queries is being addressed directly by the GenAI bot. We're also experimenting with post-sales support, although it's currently in its nascent phase.
What are your plans for the next three to five years using AI?
Over the next three to five years, our vision is to deepen the integration of AI across all facets of our business operations to simplify what is otherwise very complex and nuanced. On the customer side, we anticipate significant strides in product and content discovery, with a focus on enhancing free-text search, personalized travel planning, route planning and consistently enriching content. For our suppliers and MakeMyTrip operations, AI will play a pivotal role in refining inventory and price management. Within the next two to three years, we aim to build comprehensive content and discovery systems for all lines of business, fostering seamless interactions between suppliers, customer support personnel and customers.
What’s your strongest AI-related prediction for 2024?
AI will become an indispensable tool and a key differentiator for all companies. Within the travel industry specifically, these tools are set to play a crucial role in creating hyper-personalized travel experiences, tailored itineraries, curated local experiences and ultimately differentiating companies in a competitive market. Purchasing a travel product is rather nuanced and engaging – and the planning and exploratory cycles are long-drawn and complex. We expect generative AI technology to ease the discovery and exploratory cycles, combining the specifics of the products with the context of the place (location) and the person (user).