Gen AI Shopping Is Here: How Indian SMBs Should Prepare

Last month, a jewellery retailer in Jaipur told me something that stuck with me. He'd spent ₹3.5 lakh over two years building a decent website, running Google Shopping ads, and keeping his catalogue updated. Then he watched his 26-year-old niece buy a birthday gift for her mother without touching Google or Amazon at all. She asked ChatGPT to suggest a Kundan necklace under ₹15,000 that ships to Pune in three days, and it gave her three options with links. None of them were his products. Not because his products were worse, but because his catalogue was invisible to the AI.

That's the shift nobody sent you a memo about. A 2024 survey by Bain & Company found that a large majority of Indian consumers are willing to use generative AI tools for shopping decisions, and adoption is climbing fastest among the 20-to-40 age group that drives discretionary spending. When people ask an assistant "find me a good CRM for my clinic in Kochi" or "which local caterer does South Indian weddings under ₹800 a plate," the AI answers based on what it can read and trust. If your business isn't structured to be read, you don't exist in that conversation.

This post is a practical field guide to gen ai shopping for small business owners in India. I'll walk you through what actually makes a product discoverable to AI assistants, a real example of a business that fixed this, a comparison of where you should focus first, and a checklist you can hand to your web developer this week. No hype. Just what works.

Key Takeaways

• AI assistants recommend products they can read and verify. Clean, structured product data matters more than ad spend now.
• Implement Schema.org structured data (Product, Offer, Review, LocalBusiness) on every product and store page. This is the single highest-leverage move.
• Get listed in the sources AI models trust: Google Merchant Center, Bing Places, Justdial, and marketplace catalogues with consistent naming.
• Make your prices, GST details, stock status, and delivery timelines machine-readable in rupees, not buried in images.
• WhatsApp is where Indian AI-driven commerce actually closes. A Business API setup with a product catalogue is now essential, not optional.
• Start with one channel done properly rather than five done badly. Structured data first, then reviews, then conversational commerce.

Why is AI-driven shopping suddenly a problem for Indian SMBs?

For fifteen years, the game was SEO. You optimised for Google's crawler, chased keywords, and fought for the first page. That game still exists, but a second one has opened alongside it, and the rules are different.

When someone asks ChatGPT, Gemini, or Perplexity to recommend a product, the model doesn't return ten blue links. It returns an answer, often with two or three specific recommendations. That's a brutal funnel. Being on page one of Google gives you a shot. Being the fourth product an AI considered gives you nothing, because it only mentioned three.

The models pull from a mix of sources: their training data, live web search, structured product feeds, and increasingly direct commerce integrations like OpenAI's partnerships and Google's Shopping Graph. What they consistently reward is clean, verifiable, structured information. A product page where the price, availability, brand, and reviews are marked up in code that machines parse cleanly will outperform a prettier page that hides all of this inside images and JavaScript.

Here's the uncomfortable part for many Indian SMBs. A lot of catalogues were built for humans browsing on mobile, with product details baked into JPEGs, prices written as "Contact for price," and no structured data at all. That's fine for a person. It's a dead end for an AI.

What does an AI actually look for in a product listing?

Let me be concrete. When an assistant evaluates whether to recommend your product, it's trying to answer a series of questions quickly and confidently. If it can't, it moves on.

• What is this, exactly? Clear product name, category, brand, and a text description, not just a photo.
• How much does it cost? A specific number in rupees, marked up as a price, not "starting from" or "DM for rate."
• Is it available? In-stock status and realistic delivery windows.
• Can it be trusted? Ratings and reviews, ideally with structured review markup.
• Where does it come from? A legitimate business with a consistent name, address, and GSTIN across the web.
• Can the customer act on it? A working link, a WhatsApp number, or a checkout path.

The technical answer to most of this is Schema.org structured data, usually implemented as JSON-LD in your page code. This is the shared vocabulary that Google, Bing, and the AI crawlers understand. When you mark up a product with the Product, Offer, and AggregateRating types, you're literally handing the machine a clean data sheet instead of forcing it to guess.

Pro Tip: Don't just add Product schema and stop. Add LocalBusiness schema on your contact and about pages with your exact GST-registered business name, address, and phone number, and keep it identical to your Google Business Profile and Justdial listing. AI models cross-check these. When the details match across sources, your trust score goes up. When your address on the website says "Sector 44" and Justdial says "Sector 45," the model gets cautious and may leave you out.

How one Coimbatore business made its catalogue AI-ready

A friend of mine runs a home textiles business in Coimbatore, about 40 people, selling bedsheets, curtains, and table linen. Annual turnover a little over ₹6 crore, mostly through their own website and a couple of marketplaces. Traffic was flat and they were losing younger buyers who "just weren't finding them anymore."

We ran an audit. The findings were painfully common:

• Product prices were embedded inside banner images, so no crawler could read them.
• Zero structured data on the entire site.
• Their Google Business Profile listed an old address from a previous location.
• Product descriptions were three words each: "Cotton double bedsheet."
• Reviews existed on the marketplace but were not surfaced or marked up on their own site.

Here's what we did over about six weeks, on a budget of roughly ₹1.4 lakh including developer time:

1. Rewrote product data as text. Every product got a real title, a 60-to-100 word description with materials, thread count, dimensions in cm, wash-care, and use case. Prices moved out of images into actual text fields in rupees.
2. Added JSON-LD structured data to all 320 products using the Product and Offer types, including price, currency INR, and stock availability.
3. Fixed identity consistency. Updated the Google Business Profile, Justdial, and website footer to the same current address and GST-registered name. Added LocalBusiness schema.
4. Pulled reviews onto their own site and marked them up with AggregateRating, so the star ratings became machine-readable.
5. Set up a WhatsApp Business catalogue so an interested buyer could go from an AI recommendation straight to a chat and place an order.

Within three months, they saw a measurable rise in direct traffic from AI referrers (Perplexity and ChatGPT search were both showing up in analytics), and more importantly, a jump in WhatsApp enquiries that mentioned "I saw your bedsheets recommended." It wasn't magic. It was making themselves readable. We handled the structured data and the WhatsApp setup as part of a custom software development engagement, but a competent local developer can do most of this if you brief them properly.

Where should a small business focus first for gen ai shopping?

You can't do everything at once, and you shouldn't. Here's how the main channels stack up on effort, cost, and payoff for a typical Indian SMB.

Channel / Action	Setup effort	Typical cost	Impact on AI discoverability	Priority
Schema.org structured data on products	Medium	₹15K–₹60K one-time	Very high	Do first
Google Merchant Center feed	Medium	Free listing, dev time	High	Do first
Consistent NAP + GST identity across web	Low	Mostly time	High	Do first
WhatsApp Business API + catalogue	Medium	₹1K–₹5K/month + per-message	High (closes the sale)	Do second
Structured reviews / ratings	Medium	₹10K–₹30K	Medium-high	Do second
AI voicebot for enquiries	High	Varies	Medium	Do third

The pattern is clear. The unglamorous work of structured data and identity consistency is what moves the needle first. The shiny stuff like voicebots comes later, once buyers can actually find you. If you're unsure where your gaps are, this is exactly the kind of audit our IT consulting team runs, and it's a short engagement rather than a big project.

How do I make my product catalogue machine-readable? A step-by-step

Here's a walkthrough you can execute or hand to a developer. It assumes you have a website with product pages.

1. Take prices and specs out of images. Every price, size, and spec must exist as real text on the page. Images are for humans; text is for machines. This one step often unblocks half the problem.
2. Add Product JSON-LD to each product page. Include name, description, brand, image URL, and an offers block with price, priceCurrency set to INR, and availability. If you have reviews, add aggregateRating.
3. Set up Google Merchant Center. Create a product feed, verify your website, and submit it. This feeds Google's Shopping Graph, which several AI assistants tap into. Keep GST-compliant pricing shown as final price to the customer.
4. Verify and clean your Google Business Profile. Correct address, category, hours, and phone. This is your identity anchor.
5. Make identity consistent everywhere. Your business name, address, and phone should be identical on the website, GBP, Justdial, IndiaMART, and social profiles. Use your GST-registered legal name where it fits.
6. Ensure a bot can read the page. If your product content loads only via heavy JavaScript, some crawlers won't see it. Server-side rendering or static content is safer. Ask your developer to test with the page's raw HTML.
7. Check your robots.txt. Make sure you're not accidentally blocking AI crawlers you actually want. This is a surprisingly common own-goal.
8. Connect a conversational path. Add a WhatsApp click-to-chat link and a catalogue so a buyer who arrives from an AI recommendation can transact immediately.

Common Mistake: Businesses add structured data and then never validate it. Broken or mismatched schema (say, a price in the markup that differs from the visible price) can get you ignored or penalised. Run every page through Google's Rich Results Test and Schema Markup Validator after implementation, and re-check whenever you change pricing. Fifteen minutes of validation saves months of invisibility.

Why WhatsApp is the closing venue for AI-driven sales in India

Discovery might happen inside an AI assistant, but in India, the sale usually closes on WhatsApp. That's just how buyers here behave. Someone gets a recommendation, clicks through, has two quick questions about delivery to their pincode, and wants to talk to a human or a decent bot before paying.

This is why setting up the WhatsApp Business API with a product catalogue is one of the highest-return moves you can make. You can share products, send order confirmations, take payments, and answer queries in the channel your customer already lives in. Pair it with automated replies for common questions like return policy and delivery timelines, and you free up your team for the enquiries that actually need a person.

For businesses fielding a lot of repeat questions, an AI voicebot or a WhatsApp bot can handle first-line queries at any hour, which matters when an AI sends a customer your way at 11 pm. If you want the deeper picture on where this is heading, we broke it down in Meta's Business AI on WhatsApp: What It Means for Indian SMBs. And for high-volume order updates and OTPs, layering in bulk SMS services gives you a reliable fallback when WhatsApp isn't delivered.

What about compliance, GST, and getting the basics right?

AI discoverability sits on top of your business fundamentals, so don't skip these. Your prices shown online should be GST-inclusive final prices where you're selling to consumers, and your invoices must carry your GSTIN. If your registered business address is unstable or you're operating without a proper registered address, that inconsistency will hurt both your compliance and your identity signals online.

A lot of small e-commerce sellers operate from home and don't want that address plastered across the internet. That's a reasonable concern. A virtual office address for GST and company registration gives you a legitimate professional address you can use consistently for GST, your Google Business Profile, and your website footer, which is exactly the kind of consistency AI models reward.

One more thing worth flagging: as your team starts using AI tools to write product descriptions and answer customers, keep an eye on which tools they're actually using. Unmanaged tool use is a real risk, which we covered in Shadow AI at Work. If your staff already lives in Microsoft or Google, the built-in AI features can do a lot of this drafting safely, as we explained in 6 Hidden AI Features in Microsoft 365 & Workspace SMBs Miss.

How do I pick the right tools without ending up with a mess?

The temptation is to sign up for six different AI and commerce tools in a burst of enthusiasm and end up paying for tools nobody uses. I've seen it too many times. Pick a small, coherent stack and make it work before adding more. We wrote a full argument for this discipline in AI Tool Sprawl: Why SMBs Need a Strategy, Not More Apps, and if you're specifically choosing an assistant, Claude, Copilot, or Gemini compares the practical options.

For most SMBs, a sensible starting stack is: a website that renders clean structured data, a Google Merchant Center feed, the WhatsApp Business API, and either Google Workspace or Microsoft 365 for the AI-assisted content and email work your team already does. That's enough to become genuinely discoverable and to close sales, without drowning in subscriptions.

Frequently Asked Questions

How do I make my products show up in ChatGPT shopping results?

Focus on structured product data (Schema.org Product and Offer markup), a Google Merchant Center feed, and consistent business identity across the web. AI assistants pull from live search and shopping graphs, so machine-readable prices, availability, and reviews on crawlable pages are what get you considered.

Is SEO dead now that people shop through AI?

No. Traditional SEO and AI discoverability overlap heavily, since both reward clean content, fast pages, and trustworthy signals. Think of it as SEO plus structured data plus identity consistency. The work you do for one largely helps the other.

How much does it cost to make a small business catalogue AI-ready in India?

For a typical SMB with a few hundred products, expect roughly ₹15,000 to ₹1.5 lakh depending on your current site's condition and whether prices need to be extracted from images. Adding structured data and cleaning identity signals sits at the lower end; a full catalogue rework and WhatsApp integration sits higher.

Do I need a new website to become discoverable to AI?

Usually not. Most existing websites can be upgraded with structured data and cleaner content. You only need a rebuild if your site renders product data entirely through heavy JavaScript that crawlers can't read, or if it's built on a platform that won't let you add JSON-LD.

Will WhatsApp Business API help with AI-driven sales?

Yes, significantly, because Indian buyers tend to close purchases in chat. A WhatsApp catalogue plus automated replies lets a customer who found you through an AI recommendation ask questions and buy without leaving the app. It's the bridge between AI discovery and an actual order.

How do I keep my business name and address consistent for AI trust signals?

Use one exact business name, address, and phone number across your website, Google Business Profile, Justdial, IndiaMART, and social profiles. Where possible use your GST-registered details, and if you work from home, a virtual office address gives you a stable, professional address to use consistently.

Which AI shopping tools are Indian consumers actually using?

Adoption is spread across ChatGPT (including its search and shopping features), Google's Gemini and Shopping Graph, and Perplexity, with Meta's AI on WhatsApp growing fast because so much Indian commerce already happens there. Rather than betting on one, get your structured data right so you're readable by all of them.

The bottom line

The shift to gen ai shopping for small business owners isn't a distant trend you can watch from the sidelines. It's already deciding which products get recommended and which stay invisible, and the deciding factor is boring and controllable: whether your catalogue is clean, structured, verifiable, and connected to a way for customers to buy. The jewellery retailer in Jaipur wasn't beaten by better products. He was beaten by better data.

Start with structured data, fix your identity signals, get on Google Merchant Center, and connect WhatsApp. Do those four things properly and you'll be ahead of most of your competition, because most of them haven't noticed the game changed yet.

If you'd like a hand auditing where you stand or implementing any of this, our team does exactly this kind of work. Take a look at the full eDarpan services overview, explore our cloud and managed services if your infrastructure needs to scale, or just get in touch and we'll tell you honestly what's worth doing first.

Image credit: Reflections on the new Machine Age — technology, inequality and the economy by jurvetson via flickr (BY 2.0), sourced through Openverse.