Store Intelligence
Guide status: Operating-system guidance. Last reviewed 14 August 2026. Connected sources, fields and freshness labels depend on the installed version and granted Shopify access.
Purpose
Store Intelligence provides the shared context used across Liva 7. It brings together what can be observed about the store, catalogue, customers and storefront so that recommendations are grounded in the same evidence rather than isolated snapshots.
The workspace should help you answer what is known, where it came from, how recent it is and what remains uncertain. It should not turn a correlation, classification or model output into a fact about customer intent.
Before you begin
- Confirm the expected Shopify store, market and reporting window.
- Review source connection and freshness states.
- Understand whether customer-related information is observed, derived, inferred or experimental.
- Use only data and audiences permitted by the store's privacy and consent rules.
Read the intelligence layers
Store context can include trading structure, markets, storefront configuration and relevant operating constraints. Catalogue context describes products, variants, collections, availability and relationships. Customer context can describe observed behaviour or responsibly derived groups. Storefront context concerns the placements and experiences through which a decision may reach a customer.
Keep identity, availability and behaviour separate. A product may be correctly identified but unavailable in a market. A placement may be configured but not exposed during the selected period. A customer pattern may be visible at group level without establishing the intention of an individual.
Check quality before interpretation
- Coverage: are the relevant products, orders, audiences or placements represented?
- Freshness: does the timestamp support the decision being made?
- Consistency: do related sources agree, or is a mapping or timing difference visible?
- Eligibility: are inventory, market, consent and publication constraints applied?
- Evidence class: is the value observed, derived, inferred or experimental?
When comparing periods, check that windows, time zones, currencies and eligibility rules are genuinely comparable. A numerical difference is not automatically a meaningful change.
Safe use and fallback
If a source is late, disconnected or incomplete, label affected conclusions Unknown or Unavailable. Do not silently replace live values with stale ones. If a safe cached value is used, its age and limitations should remain visible. Customer-level data should be minimised, access-controlled and omitted from ordinary support requests unless specifically required and permitted.
You are finished when
- the relevant sources and their freshness are known;
- scope, exclusions and evidence class are understood;
- data-quality limitations travel with any downstream decision; and
- no inferred customer meaning has been presented as observed fact.