Justdial embraces generative AI to switch up local search for businesses and consumers

Artificial Intelligence (AI) is transforming how we connect and access information. Justdial, a long-standing leader in connecting businesses and consumers, has recognised this shift and is leveraging the power of generative AI to enhance its platform and provide an even more intuitive and informative user experience.

In an exclusive interview with Business Today, Shwetank Dixit, Chief Growth Officer at Justdial, delved into the company's strategic embrace of AI, outlining how this technology is being used to overcome user pain points, create richer content, and ultimately shape the future of online business directories and local search.

PD: Justdial has a long history of connecting businesses and consumers. What made generative AI the next logical step in Justdial’s evolution? What specific user pain points or market opportunities did you identify?

Shwetank Dixit: Justdial’s mission has always been to efficiently bridge the gap between businesses and consumers. We have a massive database of over 14.8 crore reviews and 4.3 crore listings, which is one of the highest listings on a single platform in a single geography, so leveraging this vast amount of information effectively became a priority. We have been a leader in Local Search since Web 1.0 and Generative AI was the next logical step to enhance this connection between users and businesses.
AI helps us provide all the necessary information about a business to a user in the simplest and most intuitive form. This includes summarising reviews, guiding users to write effective reviews, generating concise business summaries, providing detailed B2B product information on our B2B platform, JD Mart, and so on. We also ensure the correct categorisation of businesses and products. These features not only improve the user experience but also enhance brand value by offering the best information about businesses available online.

PD: Can you elaborate on how Justdial uses AI to assist users in writing reviews?

Shwetank Dixit: At Justdial, we understand that writing detailed reviews can be challenging for many users. Often, people tend to write brief comments like “great” or “good service,” which, while positive, lack detail. We have also discussed with our users their pain points in writing reviews; the majority of them feel that words do not come to mind when writing and they eventually drop out of the page. We lose potential and genuinely interested reviewers due to this.
Our AI assistant addresses this issue by providing users with attribute-based keywords that highlight different aspects of their experience, like quality of service, amenities, offerings, etc. For instance, when reviewing a restaurant, keywords might include “great ambiance,” “zero wait time,” and “pocket friendly”. A user selects keywords along with questions related to their experience and then uses the magic wand feature in our review section to generate a rich text review instantly. This feature helps them articulate their thoughts more comprehensively and ensures the reviews are more informative. Additionally, users can edit these AI-generated reviews to add any extra information, making sure the final review is fair and personalised. Our goal is to make the review-writing process easier and more effective, resulting in better, more useful reviews for everyone.

PD: Is it about prompting them with questions? Does it help improve the quality of reviews (e.g., grammar, detail)?

Shwetank Dixit: Yes, our AI system prompts users with relevant keywords plus questions, which helps them consider different aspects of their experience for review. This leads to more comprehensive and detailed reviews. With around 17 crore unique users each quarter from diverse regions of India, there is a wide range of understanding of spelling and grammar. Our AI ensures that everyone, regardless of their language proficiency, can produce high-quality, clear, and useful reviews. This makes the review-writing process easier and more effective while retaining the authenticity of reviews, resulting in better, more useful reviews for everyone.

PD: How have you addressed potential biases or manipulation concerns with AI-generated content?

Shwetank Dixit: We prioritise the fairness and accuracy of AI-generated content. We use various prompt engineering strategies to avoid any biases or redundancies in our content. Any content on our platform undergoes rigorous testing before going live. Even after going live, we ensure regular audits by human moderators for the balance and accuracy of the content. User feedback is also continuously monitored along with our Sales and Customer Support teams, allowing us to address any concerns promptly. This comprehensive approach ensures that our reviews and summaries remain trustworthy and unbiased, providing users with reliable information.

PD: What has the impact been on user engagement since implementing AI-driven review features? Are you seeing quantifiable increases in reviews posted, time spent on listings, etc.?

Shwetank Dixit: Since implementing AI-driven review features, we have seen a significant boost in user engagement. Review submissions have increased by 37%, with notable growth in categories where users previously rarely wrote reviews. This indicates that our AI assistance is making it easier and more appealing for users to leave detailed feedback. While the average time spent on review writing pages has decreased due to the efficiency of AI assistance, we have observed increased impressions and time spent on pages displaying summarised reviews and detailed reviews.

PD: The succinct descriptions generated by AI sound promising. Can you walk us through the technical process? What data sources does the AI draw upon to create these summaries?

Shwetank Dixit: We have conceptualised summarising business offerings in less than 20 words, providing users with quick snapshots created using Generative AI. We leverage internal data, including categories, highlights, locations, years of establishment, and other relevant business information. Our primary tool is GPT 3.5, which is well-suited for our use case. We have developed specific prompting techniques for different industry verticals to ensure that the information is relevant and accurate for each sector. For instance, we ensure that amenities and specialities are highlighted for Hotels.

PD: How do you ensure accuracy and avoid creating overly generic or potentially misleading descriptions?

Shwetank Dixit: We have defined a minimum number of parameters or business details for description generation to avoid any generic output. To avoid generic outputs across industry verticals, we also ensure that the results are relevant to the industry and not similar. For example, a restaurant and a hotel will have completely different structures.
We ensure accuracy by using verified business information that we have collected over more than two decades. Regular audits by human moderators help catch inaccuracies, and we continuously update the descriptions with the latest data to maintain their relevance and precision.

PD: What specific strategies or technologies are you employing to minimise Justdial’s environmental footprint while leveraging AI?

Shwetank Dixit: While we are not using on-premise AI infrastructure, we leverage energy-efficient cloud infrastructure using Azure and Google Cloud. To avoid the higher carbon footprint associated with Large Language Models (LLMs), we use Small Language Models (SLMs) like Bidirectional Encoder Representations from Transformers (BERT) wherever possible. LLMs like GPT 4 are valuable for complex content generation, but open-source SLMs are highly effective and efficient for data classification, segmentation, and text entity recognition. These SLMs are easier to fine-tune on our data and provide better accuracy for many internal data operations, helping us maintain efficiency while reducing our environmental impact.

PD: Are there trade-offs you’ve had to consider in terms of AI model complexity or data processing?

Shwetank Dixit: Yes, we have carefully considered the range of complexity in our AI processes. Our models range from the simplest linear regression to advanced LLMs. While LLMs like GPT are powerful, they can be overkill for simpler tasks and come with high computational and monetary costs.

For many tasks, we use SLMs like Phi and BERT, which are more efficient for many of the internal tasks. Additionally, we leverage distilled versions of larger models. These distilled models retain accuracy while reducing model size and computational demands. They are trained on fewer data parameters but at high volumes, making them suitable for most tasks.

Even for AI image generation, we use open-source models. The open-source community has evolved so much in the last two years, providing distilled and efficient models every other day that meet most of our needs. This approach helps us balance performance, cost, and computational efficiency effectively.

PD: How do you see generative AI shaping the future of online business directories and local search?

Shwetank Dixit: Generative AI has great potential to enhance online business directories and local search. However, we also believe the fundamentals of local search will remain the same. AI will make business discovery easier by providing users with more detailed and rich content. It will serve as an enabler, giving business owners and users the tools to be more articulate from their own perspectives.

Evidently, India is evolving quickly by adopting various AI technologies. There is tremendous potential in Tier 2 and Tier 3 cities where web usage has risen significantly in the last decade. Leveraging AI in these areas will give us the most detailed data on listings, helping bridge the gap between regions and improving business discovery for users.
We also anticipate a rise in AI-based solutions for Indic languages, which will enable small businesses to list themselves online in their native languages. This will bring a large number of small businesses onboard, enhancing their online presence and competitiveness.

Additionally, with the rise of voice-based AI, businesses can become more conversational, overcoming the inertia sometimes found in text-based interactions. Voice-based AI would allow users to add details about their business, products, and services in their native language, giving them more freedom to come online and gain a competitive edge.
As more data comes online, users will be able to make decisions quickly. This increase in web usage will lead to more online reviews and business details, ultimately improving the quality and transparency of inquiries for businesses. Businesses can also plan their marketing and inventory by identifying seasonal trends using AI insights from their data in an easy-to-understand language. This will lead to more strategic decision-making and efficient operations.

We are committed to leveraging AI to provide the best user and business experience while maintaining our leadership in local search technology. The future holds exciting possibilities, and we aim to make online business directories and local searches more dynamic, personalised, and inclusive.