Justdial's Missing Machine Layer: 160 Million Reviews No AI Engine Can Read

A trust ledger most people underrate

Justdial started in 1996 as a phone-based local search service, built around a single Mumbai number a caller could dial to ask what plumber, doctor, or restaurant was nearby. That origin matters because it shaped the company's entire data model: listings and, later, reviews were built around a verification step tied to a real phone number rather than an anonymous web account, according to Justdial's own company history. Three decades on, the company reported 160.5 million ratings and reviews and 56.1 million active business listings for the quarter ended June 30, 2026, alongside 192.9 million quarterly unique visitors, in its Q1 FY27 results.

Reliance Retail acquired a 66.95 percent controlling stake in Justdial in July 2021 for roughly 34.97 billion rupees, folding the company into its broader commerce ambitions rather than starving it. The platform has kept growing since: active listings were up 13 percent year over year and ratings and reviews up 4.4 percent year over year in the same Q1 FY27 filing. This is not a legacy directory winding down. It is a maintained, actively expanding first-party dataset with real financial backing behind it.

None of that is a proprietary claim. It is what Justdial discloses to its own shareholders every quarter, and it is worth stating plainly because the scale is easy to underrate next to Google, which dominates the conversation about local search. Justdial's corpus is smaller than Google's global index in absolute terms, but it is denser in a specific place: Indian tier-2 and tier-3 categories where a phone-verified listing has historically carried more local trust than an anonymous Google Business Profile entry, particularly in categories prone to review fraud.

What phone verification actually buys

Justdial's differentiator was never search technology; it was verification labor. The company built a large telecalling and listings operation to confirm that a business at a given address actually existed and operated the way it claimed, a model inherited from the pre-internet yellow-pages era and carried into its online and app products. Reviews inherit a version of the same discipline: Justdial states that submissions are screened by automated systems intended to catch content that does not reflect a genuine transaction, and users completing certain actions go through code-based verification steps tied to their account.

India's regulatory environment has since caught up to part of this instinct. The Bureau of Indian Standards published IS 19000:2022, a voluntary standard for the collection, moderation, and publication of online consumer reviews, which requires review authors to have their identity verified by the platform and bars reviews written or purchased by parties with a commercial interest in the outcome. A platform whose review model was built around phone verification from the outset is, structurally, closer to that standard than platforms retrofitting identity checks onto an anonymous review pool.

This is the asset side of the story, and it is real. The problem is not the data. The problem is what happens to that data once a buyer stops typing a query into a search box and starts asking a generative engine for a recommendation instead.

The wall: what a direct check of Justdial's own infrastructure shows

Two checks were run directly for this piece against Justdial's live infrastructure in August 2026, and both point the same direction.

Justdial's AI investment is real, and it is aimed at the wrong reader

It would be inaccurate to describe Justdial as ignoring generative AI. In 2024 the company's Chief Growth Officer described a working generative AI stack built on GPT-3.5 and smaller models, used to write concise business summaries from listing data, help users draft richer reviews through an attribute-based prompt and "magic wand" rewrite feature, and improve categorization across a platform serving users with wide gaps in language proficiency. The company reported that review submissions rose 37 percent after the review-assistance feature shipped, concentrated in categories where users had previously rarely written reviews at all.

That is a genuine, measured product outcome, and it demonstrates Justdial understands generative AI can deepen its own data. But every one of those features is aimed inward: it improves what a human sees inside justdial.com or the Justdial app. None of it is a structured feed, an API, or a markup layer built to let an external answer engine cite a Justdial-sourced fact the way it might cite a Wikipedia page or a press release. The company has built a better factory for its own storefront while leaving the loading dock closed to outside buyers.

This is a distinction worth being precise about, because "Justdial uses AI" and "Justdial is visible to AI answer engines" are different claims, and only the first one is currently true.

The open question that remains unmeasured

No public, audited dataset yet measures how often ChatGPT, Perplexity, Gemini, or Copilot cite Justdial specifically when answering an India-local query, category by category, city tier by city tier. That is a real gap in the evidence, and it should be named as one rather than papered over with an invented percentage. What can be said is the surrounding structure: the demand side of the query is large and growing in exactly the geography where Justdial is strongest. India reached 958 million active internet users in 2025, per the IAMAI-Kantar Internet in India report, with rural India's roughly 548 million users growing at nearly four times the pace of urban India, and 44 percent of users overall reporting they have already used an AI-enabled feature such as voice search or a chatbot.

A reasonable, falsifiable method for closing this gap exists and is not exotic: sample a fixed basket of India-local queries across several city tiers and categories, run them against ChatGPT, Perplexity, and Gemini through their standard consumer interfaces on a fixed cadence, and record which sources are cited or named for each answer. Compare the citation rate for businesses with a strong Justdial-only presence against businesses with a strong Google Business Profile presence and against businesses structured with their own schema markup. Nobody has published that study for the Indian market at the time of writing. Until someone does, the claim is directional, not quantified: a platform explicitly blocking one major crawler and structurally invisible to several others is very unlikely to be cited at the rate its underlying review volume would suggest, and the gap is plausibly largest precisely in the tier-2 and tier-3 categories where Justdial's corpus is deepest.

The contrast: what an open protocol does differently

India already has a live example of the alternative design philosophy sitting next to Justdial's walled model. The Open Network for Digital Commerce, a government-backed initiative from the Department for Promotion of Industry and Internal Trade, standardizes cataloguing, inventory, and order data through open, published protocol specifications rather than a single company's proprietary database. The explicit goal, per the initiative's own framing, is that a seller's catalog data should not have to be recreated separately for every platform it appears on. Whatever ONDC's commercial traction turns out to be, its data model is built to be machine-readable by design, in a way a single platform's closed, JavaScript-rendered listings database is not.

That contrast is the point of this piece, not a detour. Justdial proves that India already has the trust data an answer engine would want to cite: phone-verified listings, 160 million-plus reviews, three decades of continuity. What it does not yet have, on the evidence gathered here, is a structured, crawlable, permission-granted path from that data to the systems increasingly standing between an MSME and the customer asking for a recommendation.