Solutions · AI shopper readiness
AI shopper readiness for Shopify brands
A growing share of shoppers ask an assistant before they search, and your analytics calls most of them ‘direct’. This page is about the part of that you control: whether an assistant can read your store, describe your products correctly and send a shopper to a page that exists.
- 6 scan checks
- Reviewed 6 October 2026
The store cannot be read
Product structured data that is missing, invalid or has no offer and availability leaves an assistant guessing at your price and stock. A robots file that blocks GPTBot, ClaudeBot or PerplexityBot, or a store with no llms.txt, is invisible by its own choice. The scan parses the JSON-LD on each product page it crawled and fetches robots.txt and llms.txt.
The answers are not where an assistant looks
Shipping cost, returns and sizing are the questions a shopper asks an assistant about a store. If the answers live only in a PDF, a popup or three clicks down, the assistant answers from somewhere else, often wrongly. The scan checks whether each is reachable as plain text within one click of the home page.
The links an assistant gives are dead
Assistants cite URLs they saw months ago. A collection you renamed or a product you retired becomes a cited link that lands on the home page or a 404, and the shopper the assistant sent you leaves. The scan asks for your brand the way a shopper would, collects the URLs cited, and fetches each one.
Shopify’s own Storefront MCP lets a shopping agent search your catalogue and ask about your policies. The scan asks it for your crawled products by name and for ‘how much is shipping’ and ‘what is your return policy’, and reports what came back: a product not found, returned with no description or as unavailable, or a question with an empty answer.
The checks that look for this
The free scan runs these on your own pages, desktop and mobile, and shows each finding with its evidence. Each check has its own page saying how to see the problem without installing anything.
Planned and not run by the scan yet: A1 (URLs cited by AI answers redirect to home or 404), A5 (AI-referrer share unknown (no pixel)). The scan runs 35 of the 39 checks it defines.
What Liftable does about it today
What is live today. Liftable finds the opportunity and our founding-team programme implements and measures the first fix with you. Automated testing is not live on any merchant’s store yet.
- Recommendation for your developerA2 · A3 · A4
A recommendation with the change written out for your developer. Liftable never edits theme code, live or otherwise.
- Shopify redirectA1
A URL redirect added in Shopify, once you approve it.
- In Shopify adminA6
A change your team makes in Shopify admin; Liftable points at it and does not make it.
- PixelA5
Turn on the Liftable pixel, so the question can be answered from your own sessions.
- Available now the scan, the evidence, the drafts, the approval
- With the founding team implementing and measuring the first fix
- Coming automated tests on a merchant’s store
What you get
- Six AI-readiness checks with the evidence an assistant would see: the parsed structured data, the robots and llms.txt fetched, the cited URLs and their status, and what Shopify’s Storefront MCP returned for your products.
- An AI-shopper readiness score from 0 to 100, weighted by what matters most (structured data, then redirects, then plain-text answers, then llms.txt, then shipping clarity).
- A drafted fix for each: repaired schema, a FAQ block, an llms.txt, a redirect in Shopify, product descriptions and store answers to fill in Shopify admin.
- Once the pixel is on, the share of your sessions that arrive from an assistant, which no analytics tool labels for you.
Who it is for
A good fit
- Brands whose category is one a shopper asks an assistant about: supplements, skincare, specialist equipment, gifts.
- Stores that have renamed collections or retired products in the last year.
Not the right tool
- A store looking for an ‘AI SEO’ content programme. Liftable checks whether your store can be read and linked; it does not write articles for you.
Questions
Is this the same as SEO?
It overlaps. Structured data and resolving URLs help search engines too. What is specific to assistants is llms.txt, plain-text answers an assistant can quote, a robots file that admits AI crawlers, and the Storefront MCP that shopping agents use to search a catalogue.
How does the scan know which URLs an assistant cites?
It asks the kind of question a shopper would (‘where can I buy [brand] [product]’) through Liftable’s own harness against the assistants’ APIs, collects the URLs in the answers and fetches each. It reads your store as a shopper and an assistant would; it does not log in to anything.
What does an llms.txt contain?
A plain-text description of the store, what it sells, its policies and the pages that matter, at /llms.txt. It is a convention assistants’ crawlers read. Liftable publishes its own at liftable.dev/llms.txt, and drafts one for a store from what the crawl read.
Does Liftable measure how many AI-referred sessions I get?
Once the pixel is on, yes: sessions referred from the known assistant domains are labelled. Without it, the scan can only say the share is unknown, which is why the sixth check is information rather than a finding.
See which of these your store has. The scan is free, takes about a minute, and needs no install.
Scan your storeGuides
- AI shoppers · 4 minHow AI shoppers buy, and why your analytics call them directChatGPT, Perplexity and Gemini send shoppers to stores. How that traffic behaves, why analytics file it as direct, and what to fix so assistants can find you.Read
- AI shoppers · 6 minWhat autonomous agents actually do for a storeNot a chat box on your dashboard: work split into checkable stages, run on a schedule whether or not anyone looked, that stops short of changing your store.Read
Terms
By industry
- SupplementsWhere supplement stores lose orders — the subscription toggle, shipping at checkout, the discount field, a missing delivery promise — priced at a $71 average order. Illustrative example.Read
- BeautyWhere beauty stores lose orders — the discount field, shipping at checkout, the subscription toggle, slow review widgets — priced at a $62 average order. Illustrative example.Read
- PetWhere pet stores lose orders — the subscription toggle, delivery dates, the code field, variant confusion — priced at a $58 average order. Illustrative example.Read
- FitnessWhere fitness stores lose orders — add to cart below the fold, sizing, delivery on heavy items, slow pages — priced at a $92 average order. Illustrative example.Read
Integrations
- CommerceShopifyOrders, products, themes read; a theme app extension and Functions for tests you approve.Read
- CommerceLiftable pixelThe steps a shopper took, under 30 kB, silent until they consent. No video, no typing.Read
- AnalyticsGoogle Analytics 4Landing page by channel, read-only, so each opportunity says who it affects.Read
Other solutions
- Cart and checkoutShipping surprise, discount-code hunting and the cart drawer: where orders are lost after the add to cart.Read
- Product pagesThe ten things a product page gets wrong before the shopper reaches add to cart.Read
- Mobile conversionThe part of the mobile gap that is yours to close: fold, taps, speed and popups.Read
- Site speed and appsThe apps that slow the store, the ones doing nothing, and the ones doing the same job twice.Read
Last reviewed against the product on 6 October 2026.
Find what's holding your Shopify store back
ShopifyLiftable reads your storefront as a shopper would and shows what looks wrong on your own pages, desktop and mobile: friction, slow pages, missing trust and apps doing nothing, with the evidence for each. Free, in about a minute.
Scan your store.
See where your store is losing revenue and what to fix first. It is free, takes about a minute, and needs no install.