What buyers ask AI before choosing an ecommerce brand

Online shoppers ask an assistant which product suits their situation, whether a brand is actually good, what size or variant to order, and what happens if it does not work out. They are asking a stranger to reduce the risk of buying something they cannot touch, and the brand whose pages answer that risk is the brand named.

The questions, in the words buyers use

• "best office chair in India for back pain if I sit all day"
• "is this brand genuine or a reselling label"
• "what fabric is best for Indian summer"
• "I am between two sizes, should I go up or down"
• "does it come with a warranty and who services it"
• "is the return free and how many days do I get"
• "does it ship to my pin code and how long does delivery take"
• "made in India options instead of an imported brand"
• "is the cheaper listing on the marketplace the same product"

None of those is a brand question. They are risk questions. Every one of them has a factual answer that you know and have probably not written anywhere a machine can read.

What a good answer looks like

Write the specification as a sentence, not only as a table image. The material named properly, the dimensions, the weight, what it is compatible with, how to clean it. If a person could not answer "will this fit" from your page, an assistant certainly cannot.

Give real sizing guidance. A size chart is the minimum. What helps is a written note on fit: whether it runs small, what to do if you are between sizes, how it sits after one wash. Written fit advice is one of the most quoted kinds of product text.

Publish returns, warranty and delivery as text pages. Number of days, who pays the return shipping, how a refund is issued, how long it takes, what voids a warranty, which pin codes are served. These are the questions asked immediately before a purchase and they are often buried in a policy popup.

Tell people who should not buy it. Too heavy for a small flat. Not suitable for a certain use. It sounds like lost business. It is the line that makes the rest of your page credible, and it cuts returns.

Let reviews be readable. Reviews inside a widget that loads separately are often invisible to a crawler. Make sure the review text exists on the page.

The seven facts a product page must state in text

These are the seven that decide a purchase and that images usually hide.

• Material and grade, named specifically: the steel grade, the fabric composition by percentage, the wood species, the plastic type.
• Dimensions and weight, including the packed dimensions for anything large.
• What is in the box, listed, including whether accessories or fittings are supplied.
• Compatibility, meaning what it fits, what it works with and what it does not.
• Care and durability, washing, cleaning, what shortens its life.
• Warranty terms, the length, what is covered, what voids it and how a claim is made.
• Country of origin and the responsible entity, which the packaged commodities rules and the e commerce rules require you to declare anyway. Under the Consumer Protection (E-Commerce) (Amendment) Rules, 2026, notified on 9 September 2026 with a 1 January start date, a listing also shows the seller's GSTIN or MSME registration number, sponsored listings must be clearly marked, and any discount must show the prior price, which is the lowest price in the 30 days before it.

The three policy pages that answer most pre purchase questions

Most brands have these as a link in a footer, written by a lawyer, in one block. Split them and write them for a buyer.

• Returns. The window in days, whether the reason matters, who pays for return shipping, the condition the item must be in, whether the original packing is required, how long a refund takes and to where, and how an exchange differs from a return.
• Warranty. The period, what is covered and what is wear and tear, what voids it, whether the buyer needs the invoice, where service is done and in which cities, and whether there is a pick up or a walk in requirement.
• Delivery. Which pin codes are served, the expected time for metro and non metro locations, what happens with a made to order or pre order item, who to contact when a delivery is late, and the grievance officer details the e commerce rules require.

What a buyer is really asking when they ask the price

• Is the cheaper version the same thing? Usually the answer is pack size, material grade or an older model.
• Is the marketplace price the same as the website price, and why not.
• Is the return free, because a free return changes the real cost of trying something.
• Is there a warranty with actual service behind it, or only a promise.
• What is the cost per use, which is how buyers of consumables and clothing actually think.
• Is a discount coming, which you cannot answer and can address by saying how you price.

The questions brands will not answer publicly

• Who makes this product?
• Is it imported and packed here, or made here?
• Which of the sellers listing this on a marketplace are authorised?
• What is the actual failure rate, and what usually fails first?
• Do you charge for return pickup, and in which cases?
• Is the product on your site the same batch as the one on the marketplace?

The fourth question sounds impossible to answer and has a good answer: say what part wears first, how long it typically lasts and how to make it last longer. That is genuine product knowledge and it is the most quotable thing on a product page.

A worked example

A man in Coimbatore wants an office chair because his back hurts after a day at a desk. He asks an assistant for the best office chair in India for back pain when sitting all day. The answer names four brands and cites two roundup articles and one marketplace listing. Every source it used contains written specifications: seat height range, lumbar adjustment, recline lock, weight capacity, warranty period.

A brand that sells a good chair and is absent from that answer is usually absent for one reason. Its product page has a hero image, a set of lifestyle photographs, four adjectives and a buy button. The seat height range is in an image. The warranty is in a footer link. The weight capacity is nowhere.

Fixing it is a day of work per product. State the seat height range in centimetres, the seat depth and width, the backrest height, which adjustments exist and which do not, the weight capacity, the castor type and which floors it suits, the assembly requirement and time, the warranty period with what is covered and where it is serviced, and the honest note that a chair helps posture and is not a treatment for back pain and that someone with persistent pain should see a doctor. Then add a written note on who the chair suits: height range, hours of use, and who should choose a different model.

That last paragraph is the one that gets quoted, because it is the only one in the category that admits a limit.

Where you are probably missing

Three gaps are almost universal. Specifications trapped in images. No written fit or use guidance, only a chart. And a returns or delivery policy that cannot be read without clicking through a script. Together they mean the assistant has nothing concrete to say about your product, so it talks about a competitor whose listing does.

A fourth gap is the brand page. Buyers ask whether a brand is genuine, and the answer lives on a page that names the company, the registered address, when it started, where the products are made and who to contact. Many brands have no such page at all, which is exactly the shape of a brand an assistant is cautious about.

What this does not cover

This does not make a product better. A page that honestly describes a product that fails in six months will simply help more people find out.

It also cannot win a question that is decided on price at the lowest end of a category, where the answer is whatever is cheapest on a marketplace that day. That is not a content problem.

And it cannot fix what other people publish about you. Reseller listings with wrong specifications, an old review of a discontinued model, a complaint on a consumer forum: all of it is read, and your own pages sit alongside it rather than replacing it. The most you can do is be the clearest and most complete source about your own product. Nobody can guarantee a place in an assistant's answer.

Common questions

I am between two sizes. Should I go up or down?

That depends on the garment and on how it is cut, which is why a size chart alone does not answer it. What helps is the brand stating whether the item runs small or large, how much it shrinks after the first wash, whether it is a relaxed or a fitted cut, and what to do if your measurements fall between two columns. A brand that publishes that in writing sees fewer returns, which is why it is worth doing for commercial reasons as well.

How do I tell whether a listing is genuine?

Check whether the seller is one the brand names as authorised, check the price against the brand's own site, and look at the packaging details in the photographs against what the brand describes. Brands can make this much easier by publishing an authorised seller list and a short page on what genuine packaging looks like, including any code or seal.

Is the return really free?

Sometimes, and the detail matters. Ask who pays return shipping, whether that changes if the reason is a change of mind rather than a defect, whether the original packing is required, what condition the item must be in, and how long the refund takes and to which account. These are the questions that cause disputes, and a brand that answers all of them in plain text on one page has removed the main reason a cautious buyer hesitates.

Who services the warranty, and where?

This is the question that separates a brand with a service network from one with a promise. Ask which cities have service, whether it is a pick up or a walk in, what the turnaround is, whether you need the original invoice, and what counts as wear and tear rather than a defect. Brands with a genuine network should publish the city list, because it is a real advantage and almost nobody states it.

Does it ship to my pin code?

Most brands can answer this only through a checker on the site, which an assistant cannot use. Publishing a written summary helps: which states and regions are served, the expected delivery time for metro and non metro addresses, which products cannot be shipped to certain locations because of size or restrictions, and what happens with a made to order item.

Why is the marketplace price different from the website price?

Because the costs are different. A marketplace price carries commission, fulfilment and advertising costs, and marketplace discounts are sometimes funded by the platform rather than the brand. Saying this plainly is better than letting a buyer conclude that one channel is selling a different product, and it is the sort of explanation a buyer repeats.

What to do first

Pick your ten best selling products. For each, move every specification into text, add written fit or use guidance, and add a short section on who this is for and who it is not for. Then publish returns, warranty and delivery as their own plain pages, written for a buyer rather than a lawyer, with the grievance officer details on the delivery page.

Then add a real brand page naming the company, the address, where the products are made and the authorised sellers. The companion page on how an ecommerce brand gets recommended by AI covers marketplaces, roundups and community threads, which this page does not repeat.

How we know this

We read the assistants rather than guessing what they do. In our own audit of 1,260 saved answers from ChatGPT and Gemini across the United States, the United Kingdom and India, 82.0% of every citation landed on a site the business itself owned, and review directories took 0.7%. Counted as one domain per reading, 16,516 of 20,143.

That is the whole asymmetry: the pages you own are the pages an assistant quotes back to a buyer. See the dated readings and the method on our proof page.

Find out whether assistants name your business, on your own domain and your own buying questions.

By Aditya Goel, Founder, AI Knows Us. Reviewed 1 November 2026.