Skip to main content
Last updated: 2026-09-03. This page describes how Layers handles your store’s data and your shoppers’ data today, with a focus on the AI features. It is a description of current practice, not a contractual commitment. For the legal terms that govern your use of Layers, see the Terms of Use and Privacy Statement.

​
The short version

  • Shopper data is not sent to any model provider for training, and Layers does not use it to train or fine-tune Layers-owned models.
  • AI models receive only the inputs needed to answer a specific request (for example, a search query or a product title to embed). They do not receive shopper identifiers.
  • All AI processing runs on AWS infrastructure in Layers’ own account. Model providers do not receive your data directly.
  • Analytics events are only captured when Shopify’s Customer Privacy API reports that analytics processing is allowed for that shopper. Layers defers to Shopify for all consent decisions.
  • Layers does not store shopper IP addresses.

​
What the storefront pixel captures

The Storefront Pixel runs on your storefront and records how shoppers interact with search, collections, and blocks. Each event includes:
  • An event type (collection view, product view, product click, product impression, hover or touch, add to cart, search, block view).
  • A timestamp and a unique event identifier.
  • A session identifier, taken from Shopify’s own session cookie where available. When the shopper has not consented, Shopify does not set that cookie and the pixel uses a random identifier that lives only for the current browser tab.
  • The Shopify customer ID, only when the shopper is signed in to your store.
  • The product, variant, collection, block, or search query the event relates to.
  • Attribution tokens that tie the event back to the Layers request that produced the results.
  • Experiment identifiers when A/B testing is active.
When events reach Layers, they are enriched with a country, region, city, device type, browser, and operating system. Location is derived at the edge from the connection and is stored at city granularity at most. The IP address itself is not stored with the event. The pixel does not capture names, email addresses, shipping or billing addresses, payment details, or free-text form input. For the full field-level breakdown, see What the pixel collects.

​
Per-session versus per-store data

Per-session data exists to make a single shopping visit better and to attribute outcomes to it:
  • Session identifier and, for signed-in shoppers, the Shopify customer ID.
  • Products currently in the cart, recently viewed products, and recent purchases, as provided by your theme through the Layers app embed.
  • Aggregate customer signals such as whether the shopper is returning, number of orders, and average order value. These are used for personalization and ranking and are not stored as a shopper profile by Layers.
  • Marketing context such as UTM parameters and ad click identifiers, held in the browser for the duration of the visit and used for Dynamic Linking.
Per-store data is what Layers needs to run search and merchandising for your store:
  • Your catalog, mirrored from Shopify through Catalog Sync: products, variants, collections, inventory, pricing, metafields, and images.
  • Vector embeddings generated from your catalog and from search queries.
  • Your configuration: merchandising rules, sort orders, synonyms, redirects, blocks, and dashboard settings.
  • Aggregated analytics computed from events, such as click-through rate, conversion rate, and popular searches.
Raw events are retained for a fixed window and then deleted. See Data lifecycle for the retention period of each category.

​
What AI models see

Layers uses AI models for two things: understanding language (query interpretation, intent detection, AI-assisted merchandising, and the assistants in the Layers dashboard) and generating embeddings for semantic search. All of this runs on AWS infrastructure in Layers’ own account, whether the model is a third-party foundation model or one Layers hosts itself. Data stays inside that environment and is not sent to model providers’ own services. Model inputs are used only to produce the response for that request. Layers does not opt in to any program that would allow a provider to use request content for model training. For the complete list of vendors that process data on Layers’ behalf, see Infrastructure & Sub-processors.

​
Training and fine-tuning

  • Layers does not send customer or shopper data to model providers for the purpose of training their models.
  • Layers does not use customer or shopper data to train or fine-tune Layers-owned models.
  • Ranking models that Layers operates are tuned on aggregated interaction statistics (for example, click and conversion rates per product) rather than on individual shopper records.
Layers defers to Shopify for consent. The pixel checks Shopify’s Customer Privacy API before queuing any analytics event, and if analytics processing is not allowed for the shopper, no event is captured or sent. Without consent the session identifier is a random per-tab value rather than Shopify’s session cookie. Details, including what happens when a shopper changes their consent mid-session, are on the Storefront Pixel page.

​
Questions

If your security or privacy review needs more detail than this page provides, contact support@uselayers.com.

​
Next steps