PIING
Reading Piing: IntelligenceThe fields it writes
Piing: Intelligence
- How Intelligence works
- The fields it writes
- Profile refreshes and legacy units
- Preview your profiles
- Custom fields and prompts
- The storefront pixel
- Run and schedule analyses
- Where your data goes
In this guide
- AI-inferred fields
- Rule-based fields
- Storefront pixel fields
- The summary field
- Adding your own fields
- Next steps
Search or browse documentation
Piing: Intelligence 2 / 8
The fields it writes
The full set of AI-inferred and rule-based customer fields Piing: Intelligence writes to Shopify metafields, plus the storefront pixel fields and the summary field.
Piing: Intelligence ships with 22 ready-made customer fields. Nine are AI-inferred, nine are calculated directly from order data, and four use optional storefront-pixel signals. Scale and above add custom fields and prompts.
Every enabled field accepted by the merchant's controls is written to a Shopify customer metafield in the app-owned namespace $app:piing-intelligence. The value is visible and queryable in Shopify, and metafield-aware tools can use it.
AI-inferred fields
These are inferred from product titles, variants, categories, vendors, prices, and order dates. Each one returns null when the order history is too thin to be sure, rather than guessing.
likely_gender: likely gender, inferred only from the product types and sizes purchased. Null if the mix is ambiguous or unisex.preferred_category: the customer's main product category.preferred_style: the dominant style or aesthetic across their orders.likely_size_clothing: most frequent clothing size from order variants.likely_size_shoes: most frequent shoe size from order variants.brand_affinity: the top one or two brands they buy.customer_persona: a short, plain-language summary a shop assistant can read at a glance.likely_next_category: the category they are most likely to buy next.seasonal_buyer: whether their buying clusters around a season, and which one.
Rule-based fields
These are computed from the order data inside the app, with no AI involved and no token cost.
aov_tier: average order value tier (High, Mid, or Low value).purchase_frequency: how often they buy (Frequent, Regular, or Occasional buyer, or Lapsed).sale_sensitivity: whether they buy at full price or wait for discounts.reactivation_risk: how likely they are to be lapsing (Low, Medium, High Risk, or Lapsed).lifetime_value_tier: total spend grouped into High, Mid, or Low.is_gifter: whether they frequently buy gifts.avg_days_between_orders: the mean gap in days between their orders.preferred_price_range: typical price band (Budget, Mid-range, or Premium).data_confidence: how much order data is available for this customer.
Storefront pixel fields
If you run the storefront pixel, these fields are derived from browsing signals and written alongside the rest:
browse_to_buy_ratio: average product views per completed purchase.recent_browse_categories: the collections they have browsed recently.cart_abandon_rate: the share of cart adds that did not lead to checkout.last_active_at: the date of their most recent storefront activity.
The summary field
Alongside the individual fields, the app writes one intelligence_summary metafield per customer. It lists each field's value, confidence score, source and current status, including values a merchant has confirmed or overridden. This makes the whole profile readable without hiding how each value was reached.
Tip
Each field carries a confidence score. Use a confidence floor in your output rules so weak inferences are held back rather than written.
Adding your own fields
On Scale plans and above you can add custom fields with your own prompts, on top of the set above. See Custom fields and prompts.
Next steps
Related guides
Was this guide helpful?
Next guide in Piing: Intelligence
Profile refreshes and legacy units
New Intelligence plans count successful customer profile refreshes. Existing field-unit subscriptions keep their approved terms.
Still need a hand? Contact us