Nora Kimura
Reverse-engineers how large language models index product catalogs — and which signal families actually survive the crawl.
- Retrieval-augmented generation
- Product catalog indexing
- Schema conformance
- Variant + canonical mechanics
- Posts
- 0
- Guides
- 14
Nora's work sits at the intersection of retrieval-augmented generation and e-commerce search. She reverse-engineers how large language models index product catalogs and which signal families actually survive the crawl. Her guides focus on the low-level mechanics — schema conformance, canonicalisation, variant handling, and the structural choices that make a PDP retrievable in the first place.
- Retrieval-augmented generation
- Product catalog indexing
- Schema conformance
- Variant + canonical mechanics
Guides authored by Nora Kimura
- content10 min
How to write product descriptions for AI retrieval in 2026
The rules changed. Classic conversion copy still works on a PDP, but the same copy that converts a browsing shopper can be invisible to an AI retriever. This is the dual-format playbook — product descriptions that sell to humans and cite to models.
Co-authored with Hiren Bhuva - measurement9 min
How to check if your Shopify store appears in ChatGPT
The 15-minute prompt panel that tells you whether ChatGPT can see your store, whether it recommends you, and whether you are losing ground week over week.
Co-authored with Hiren Bhuva - engine11 min
What is Google AI Mode and how to optimize your Shopify store for it
AI Mode is Google's deeper, research-style answer tab — distinct from AI Overviews. It is Gemini-powered, multi-turn, and draws on a query fan-out that surfaces very different sources than classic organic.
- technical9 min
What is ai-sitemap.xml and how to generate it on Shopify
ai-sitemap.xml is sitemap.xml with AI-specific extensions — content hashes, freshness hints, canonical pointers for AI retrievers. It is not a replacement for sitemap.xml; it is a companion file.
- shopify11 min
Optimizing your Shopify product feed for AI channels
Your Shopify product feed powers four AI channels — Google AI Mode, ChatGPT Shopping, Copilot Shopping, and Perplexity Shopping. Each weights feed fields differently, and Shopify's defaults underfill all four.
- engine11 min
Where do ChatGPT, Gemini, and the other AI engines actually get product data from?
Every AI answer about a product has a provenance. Tracing that provenance is how merchants decide which data pipelines to prioritise. This is the map — nine sources, six engines, and the merchant actions that control each one.
- technical10 min
Using Shopify metafields to make your catalog AI-discoverable
Metafields are the closest thing Shopify has to a native structured-data layer. Used deliberately, they are how you ship rich product schema, answer shopper questions, and give AI retrievers the facts they cannot extract from marketing copy.
- shopify9 min
Shopify collection pages for AI search in 2026
Collection pages are the highest-intent URLs on a Shopify store — and almost always the thinnest. Fixing them is a two-hour content exercise per collection that compounds across hundreds of AI prompts.
- content9 min
Information gain: why AI engines cite original data over summaries
AI retrievers have moved past the era when summarised content could compete for citations. The new ranking primitive is information gain — content that says something the retriever has not already seen. For ecommerce brands, that shifts the brief from 'write about topic X' to 'contribute one original data point to topic X'.
- content10 min
How to write buying guides that get cited in AI Overviews
Buying guides were one of the highest-leverage content formats in classic SEO. They still matter — the format just changed. AI Overviews cite buying guides that answer the shopper's real decision question in a quotable sentence, not the ones that reach 3,000 words of comprehensive coverage.
- content10 min
Product comparison pages that get cited by ChatGPT and Perplexity
Comparison pages are one of the most-cited formats in AI retrieval — when they are done honestly. The ones that win have structured feature tables, named competitors, acknowledged tradeoffs, and a specific 'which is right for you' close. The ones that lose are branded hit pieces dressed up as comparisons.
- content9 min
Answering long-tail shopper questions in ecommerce content
Long-tail shopper questions are the highest-converting and least-competitive queries in ecommerce — and in 2026 they flow almost entirely to AI answers. The brands that get cited are the ones that answer each question in a structured, quotable form, in the right place in their content architecture.
- content10 min
Why Reddit matters for AI visibility — and exactly what to do about it
Reddit is one of the top-five most-cited domains across ChatGPT, Perplexity, and Google AI Overviews. The upside for Shopify merchants is real — but Reddit punishes marketing behaviour fast, so the playbook is about participation and honesty, not link placement.
- concept10 min
The future of ecommerce search: what 2027 will look like
Forecasting is risky, but three trajectories are clear enough to plan against — agentic checkout moves inside chat, structured product data becomes the dominant retrieval surface, and the number of engines merchants must serve doubles. Here is the working model.
Every post and guide from the Surfient research team.
Weekly field notes, structured playbooks, and deep dives on how ChatGPT, Perplexity, Claude, and Google AI Overviews decide which Shopify store gets quoted.