A shopper asks an AI assistant for cabin luggage under ₹8,000. They narrow the options to bags measuring less than 55 cm, including wheels, then ask where a shortlisted model is available.
For an eCommerce brand, that conversation raises three practical questions: did the right product appear, were its details accurate, and which seller was shown?
This is an illustrative shopping journey, not a recorded test. It reflects capabilities already documented by shopping platforms. Brands preparing for 2027 can use those capabilities to decide what to measure now, without relying on predictions about how quickly AI shopping will grow.
Start with one priority category. Establish whether your products appear in relevant shopping results, check the exact model or variant, and record the purchase destinations offered. Keep the technical foundation sound, then use the findings to prioritise product information, content and channel improvements.
What changed in AI shopping in 2026?
Several documented updates give commerce teams a concrete basis for planning.
ChatGPT improved product discovery and comparison
On 24 March 2026, OpenAI announced richer product results, conversational refinement and side-by-side comparisons covering details such as price, reviews and features. The announcement also described improvements to product information coverage and freshness.
The practical implication is that a shopping test should follow the conversation beyond the first answer. Check what happens when a shopper adds a requirement or compares shortlisted products.
Source: ChatGPT release notes, 24 March 2026.
Google expanded conversational shopping in India
In April 2026, Google announced shoppable product listings, comparison tables, prices and purchase options in the Gemini app in India, available in English and Hindi. It also announced expanded shopping in AI Mode in Search, including product visuals and comparisons. At that announcement, the AI Mode shopping experience was available in English, with Hindi planned.
For teams serving India, this provides a local reason to include conversational shopping in visibility reviews. Test the languages and experiences actually available to your customers rather than assuming every market has identical features.
Source: Google India shopping update.
Google introduced Universal Cart in the US
On 15 October 2026, Google announced Universal Cart in the US across Search and the Gemini app. The experience brings products from multiple retailers into one place and can alert shoppers to changes such as price drops or renewed availability.
This US launch adds another consideration for relevant markets: a product may remain under consideration while its available offers change. It does not establish equivalent availability in India.
Source: Google’s Universal Cart announcement.
These updates establish useful capabilities to assess. They do not tell us how much revenue any individual brand will receive from AI shopping in 2027.
What today’s AI shopping capabilities mean for 2027?
Planning should begin with what customers can already do. The following are business implications and recommended checks, rather than forecasts.
Check whether products stay on the shortlist as the conversation narrows
A broad discovery question and a detailed comparison ask different things of a product.
In the luggage example, the first prompt asks for options within a budget. The follow-up introduces a dimension requirement. A product could reasonably fit the first request and fail the second.
Record where the product appears and whether it still meets the shopper’s requirements as the conversation develops. Absence alone does not establish a visibility problem. First confirm that the product is suitable for that particular request.
Check SKU accuracy, not only brand mentions
A brand name in an answer does not establish that the right product has been recommended. A stock keeping unit (SKU) identifies a particular sellable item, such as a specific model, size or configuration.
Check whether the answer identifies the expected item and whether its linked page matches. For luggage, distinguish the product’s external dimensions from its internal capacity. For electronics, distinguish storage or memory configurations.
Treat these as accuracy checks against your approved catalogue, not a demand that every eligible SKU must appear. OpenAI’s documentation notes that shopping results do not necessarily include every available product.
For additional context on selection, see GoVISIBLE’s guide to how AI engines choose products to recommend.
Check which seller the shopper is sent to
Product presence and seller presence answer different commercial questions.
OpenAI documents that selecting a product may show merchants offering it, with selection influenced by merchant and product information. A product result therefore needs a second check: where can the customer go next?
A retailer link may support a retail-led brand’s objectives. A D2C team may want to understand whether its own offer also appears. Assess the destination against stock, market coverage and your channel strategy before labelling it a gap.
Source for product coverage and merchant options: Shopping with ChatGPT Search.
Start with the technical essentials
Before interpreting visibility results, confirm that priority product pages are accessible and clearly describe the product, variants, price and availability. Where you use feeds or structured data, check that they represent the corresponding product and offer accurately.
This is a minimum foundation for reliable product information. It does not guarantee a recommendation. Google’s guidance for AI features in Search reinforces established SEO practices, including accessible content, matching structured data and current Merchant Center information, without requiring special AI-only markup.
The detailed implementation belongs in a separate technical readiness checklist. For this first review, resolve obvious access or product-information problems, then focus on how products actually appear.
Source: Google Search Central guidance for AI features.
How do you set up a product discovery baseline before 2027?
Choose one commercially important category, one market and a manageable group of products. Prepare the approved names, identifiers, important attributes and seller URLs so the person checking results has a reliable reference.
Use a short sequence covering discovery, comparison and seller selection:
| Stage | Illustrative Prompt | What to Record |
| Discovery | “Which cabin bags under ₹8,000 should I consider?” | Brands and specific products shown; distinguish text mentions from product cards. |
| Comparison | “Which of those measure less than 55 cm including wheels?” | Exact models retained, dimensions stated and whether the linked item matches. |
| Seller selection | “Where can I buy this model?” | Brand, retailer or marketplace links; displayed price and availability where present. |
These prompts illustrate a method, not actual product recommendations or results. Adapt them to requirements your category buyers genuinely use.
For each run, record the date, platform, market, language, prompt sequence and relevant account or personalisation settings. Save the answer or screenshot with the product and destination URLs.
Repeat the sequence under comparable conditions. If an unexpected result would trigger implementation work, rerun it in a fresh conversation and note whether it recurs. A single response is a sample, not a complete measure of demand or visibility.
Turn the findings into focused actions
Use the baseline to decide what needs investigation and who should own it.
| Finding | First Investigation |
| An expected SKU is repeatedly absent | Confirm that it meets the prompt’s requirements, then review product information and relevant listings for missing attributes. |
| An incorrect variant or specification appears | Compare the answer and destination with the catalogue, product page, feed and seller listings. Identify any confirmed discrepancy. |
| A retailer appears instead of the brand site | Assess whether this supports the channel strategy. If direct sales are a priority, check the brand’s own offer and availability. |
| A displayed price or stock status is outdated | Check the merchant’s current offer and the information supplied through relevant listings or integrations. |
These are investigation routes. A missing product does not prove a crawlability problem, and a retailer destination is not automatically a poor outcome.
Different sellers can legitimately have different prices and stock. Compare each displayed offer with that seller’s listing for the same variant, market and time. Treat a discrepancy between an answer and its linked offer differently from a normal price difference between retailers.
After addressing confirmed issues, repeat the original checks and compare the results with your baseline. Record changes without assuming that your edits caused every movement.
Keep product visibility, website referrals and sales as separate measures. Product appearances describe discovery opportunities; analytics and commerce reporting help assess what happens after a visit. The baseline should support decisions across those stages.
Make your 2027 plan measurable
The useful year-end outcome is a documented starting point: which products appear for relevant shopping questions, whether their details are accurate, which sellers are shown, and who will investigate confirmed gaps.
GoVISIBLE Commerce helps teams examine product presence in AI shopping results, compare visibility across brand sites, retailers and marketplaces, and monitor changes over time. Its shopping-result focus complements broader brand visibility analysis.
Explore GoVISIBLE Commerce to understand your product and channel presence and establish a baseline for your 2027 plan.




0 Comments