Hamza JameelField sheet Nº 012

Free field sheet · Nº 012
Issued 28 Sep 2026

KeywordFEED

Getting Your Products Ready for AI Shopping Surfaces

A checklist for your feed, product pages and Merchant Center so an AI assistant can read, price and recommend what you sell.

Checks
27
Do first
5
Parts
5
Cost
Free
Hamza Jameel
Prepared byHamza Jameel10+ yrs · $8M+ ad spend managed · 150+ page-one rankings
Checked 0/27Top

AI assistants are starting to read product data directly and act on it, including early tests of buying inside Gemini. If your feed is thin, your prices disagree between page and feed, or your stock status is stale, an assistant will skip you or quote you wrong. This document walks through what to check, where to click, and what a good result looks like.

Part 1 of 50/4

What changed

Facts from the TechCrunch report, worded no stronger than the source.

  1. Reported by TechCrunch on 26 September 2026. Flipkart is owned by Walmart.

  2. Treat it as an experiment, not a live channel you can sign up for today.

  3. Planned, not confirmed as shipped. Dates in tests like this often move.

  4. That is the direction of travel the report points to. Whatever surface comes to your market, it will read structured product data first.

Part 2 of 50/6

Check your product data is machine readable

Do this on your five best selling products first, then scale.

  1. Paste a product URL into Google's Rich Results Test. A pass with no errors under the Product item means an assistant can parse price and stock without guessing.

    Wheresearch.google.com/test/rich-results

    Copy this
    https://[yourdomain.com]/product/[your-best-seller]
  2. These identifiers are how an assistant matches your listing to the same product elsewhere and decides whether to show you. Missing GTINs are the single most common reason a product is ignored or mismatched.

    WhereMerchant Center › Products › All products › click a product › Attributes

  3. Check three products by hand: open the page, open the feed row, compare. A mismatch triggers a Merchant Center item disapproval and makes an assistant distrust your data.

    WhereMerchant Center › Products › Needs attention

  4. Open yourdomain.com/robots.txt and look for Disallow rules covering /product/ or your image folders. Blocking images is a quiet killer because listings without images get filtered out.

    Copy this
    https://[yourdomain.com]/robots.txt
  5. In Chrome, open DevTools > Network > right click > Block request URL on your main JS bundle, then reload. If the price disappears, put it in the HTML or in JSON-LD so it survives a simple fetch.

  6. Size and colour picked by a dropdown with one shared URL means an assistant cannot recommend a specific variant. Use item_group_id in the feed to tie variants together.

Part 3 of 50/6

Fix the Merchant Center basics

  1. Sort by number of products affected and fix top down. Good looks like zero disapprovals and under 5 percent of items with warnings.

    WhereMerchant Center › Products › Needs attention

  2. This lets Google correct your feed from your page markup when the two disagree, which prevents whole-account suspensions for price mismatch.

    WhereMerchant Center › Settings › Automatic improvements › Automatic item updates

  3. A scheduled fetch once a day is the floor. If you sell out of things within hours, add the Content API or a supplemental feed for availability only.

    WhereMerchant Center › Products › Feeds › click your feed › Settings › Fetch schedule

  4. Assistants and shopping surfaces show delivery time and returns window next to price. Blank fields mean your listing looks worse than a competitor with the same product.

    WhereMerchant Center › Settings › Shipping and returns

  5. An unverified or mismatched domain blocks your products from most free listing surfaces entirely.

    WhereMerchant Center › Settings › Business information › Website

  6. Free listings are the pool many AI shopping answers draw from. There is no cost, only a data quality bar.

    WhereMerchant Center › Growth › Manage programmes › Free listings

Part 4 of 50/5

Write product copy an assistant can quote

Assistants summarise. Give them something worth summarising.

  1. Usually size or fit, material, and what it is for. Assistants truncate long descriptions and will quote the opening.

  2. "Ultra durable" cannot be matched to a query. "1680D ballistic nylon, 2 year warranty" can. Aim for at least six named specs per product.

  3. Use a real question as the heading and answer it in two sentences underneath. Mark it up with FAQPage structured data if your platform supports it.

    Copy this
    Does the [product name] fit [common use case]? Yes. [One sentence of specifics: measurement, compatibility, or limit.]
  4. Compare your product titles to your top search queries. If people search "running shoes wide fit" and your title says "AeroFlex 3 W", add the plain words to the title.

  5. Brand, then product type, then the two attributes people filter on, under 150 characters. Keep it identical between page title and feed title.

    Copy this
    [Brand] [Product type] - [key attribute], [size or colour]
Part 5 of 50/6

Test it like an assistant would

Run these once now and once a month. Save the answers so you can see drift.

  1. Anything it gets wrong or says it cannot find is a gap in your data, not a flaw in the model. Fix the page, then rerun.

    Copy this
    Open this URL and tell me, using only what is on the page: the exact price and currency, whether it is in stock, the return window, the delivery estimate, and three specifications. If any of these are not stated on the page, say "not stated" rather than guessing. URL: [your product URL]
  2. This shows whether you appear at all. If you do not, note which brands do and what data they publish that you do not.

    Copy this
    I am looking for [product category] under [price] for [use case] in [country]. Recommend five options with price, why each suits me, and where to buy. List the sources you used.
  3. Read the comparison table it produces. Every blank cell against your name is a missing attribute you can go and add today.

    Copy this
    Compare [your product URL] with [competitor URL 1] and [competitor URL 2]. Build a table with price, materials, warranty, delivery time, returns policy and sizes. Mark any cell where the information is not published as "missing".
  4. Shipping thresholds and return windows drift between the product page, the footer and the FAQ. Assistants pick one at random, often the wrong one.

    Copy this
    Check these URLs for any contradictions about price, shipping cost, delivery time or returns policy, and list each contradiction with the two conflicting statements: [URL 1], [URL 2], [URL 3]
  5. Look at the Merchant listings or Products report. Valid items should roughly match your catalogue size. A big gap means pages are not being parsed.

    WhereSearch Console › Enhancements › Merchant listings

  6. Feeds break silently after theme updates, price changes and plugin upgrades. Recheck disapprovals, run the three prompts above, and compare to last month's answers.

Want me to look at yours?

If you have gone through this and want a second pair of eyes, book a free call. We open the account together and look at what is actually happening. No deck, no pitch.

More free sheets

Sources
  1. https://techcrunch.com/2026/09/26/google-tests-buying-from-walmart-owned-flipkart-through-gemini-and-ai-mode-in-india/