# Product recommendations and purchase tracking Show suggested products with store-backed cards and associate recorded recommendations with later tracked purchases. *Release 3.3 · operator, supervisor* **Settings › AI › Features** · **Settings › General Settings › Admin** ## What it is **Product Suggestion** helps you find products relevant to a conversation. When a suggestion has one clear match in the connected store, INSIDE can present it as a store-backed product card. Recorded recommendations can later be associated with tracked purchases of the same SKU. ## Who uses it and why - Operators review suggestions and use relevant products in conversations. - Supervisors review recommendation activity and associated purchases in commerce reports. ## Where to find it Operators use the product suggestion controls in the conversation. Users with the **Settings** or **Developer** role configure the AI feature through **Settings › AI › Features**. Powerfront manages **Product Suggestion Attribution Days** in **Settings › General Settings › Admin**. Ask Powerfront to change this period for you. ## How it works The configured AI feature uses the conversation and its product tools to suggest products. Existing operator product recommendations can also be recorded. Commerce enrichment looks for an exact, unambiguous product match in the current website and subsite. A match can use a product identifier or a recognised product alias. One match supplies the product card. No match, several matches or no configured commerce provider leaves the existing recommendation available without that enrichment. INSIDE records recommendation use with its visitor, conversation, operator and product context. When it processes a purchase, it looks for recommendations for the same website, visitor and SKU within the attribution period. Recommendations already linked to a purchase are excluded. A completed purchase supplies the final order and purchased-item association. ## Configure it Ask a colleague with the **Settings** or **Developer** role to configure **Product Suggestion** and its product tools. A commerce provider can supply store-backed cards, but existing recommendations can remain usable without one. Ask Powerfront to set **Product Suggestion Attribution Days** if the default seven days does not suit your reporting. This period is separate from the conversation sales attribution window. ### Settings | Setting | Where | Applies to | Default | Who can change it | What it does | |---|---|---|---|---|---| | Product Suggestion Attribution Days | Settings › General Settings › Admin | website | 7 days | Powerfront, on request | Sets how far back INSIDE looks for matching recommendations when processing a purchase. The control accepts 1 to 60 days. | ## Limits and good to know - Enrichment checks at most 50 distinct product identifiers at a time. - Generating a suggestion, recording its use and linking a later purchase are different events. - Purchase association uses SKU and visitor matching. Different store products that share a SKU need care when interpreting the result. - More than one recorded recommendation can qualify for the same purchased SKU. Recommendation counts do not necessarily equal distinct purchases. - A linked purchase is evidence of an association. It does not establish that the suggestion caused the purchase. - An incomplete basket can help resolve product identity, but it does not create the final purchase association. ## Depends on and works with ### This feature uses - **Configured in** Admin settings and feature switches (managed by Powerfront): Powerfront manages the recommendation attribution period. - **Configured in** [AI product suggestion settings](AI-027.md): Product Suggestion settings choose the AI instructions, search functions and eligible conversations. - **Works better with** [Messages and rich content](CHAT-002.md): Sharing recommended products in chat can record their use. - **Works better with** [Commerce provider setup and selection](COM-001.md): An available provider supplies authoritative product information for matching suggestions. - **Works better with** [Product options and variant selection](COM-008.md): Product lookup enriches unambiguous recommendation matches. - **Works better with** [Product card configuration and shopping actions](COM-014.md): Matching suggestions can use product cards. - **Works better with** [Tracked product search and integrations](COM-018.md): Existing product integrations can continue to supply usable suggestions. - **Works better with** [Purchase tracking and sales totals](COM-020.md): Tracked completed purchases let INSIDE link recommended products to later sales. - **Works better with** [Product recommendation and purchase reports](REP-009.md): Commerce reports use recorded recommendation activity and purchase associations. - **Related** [Sales attribution to conversations and operators](COM-021.md): Recommendation purchase links use different matching rules and periods from conversation sales credit. ### Used by - **Required by** [Product recommendation and purchase reports](REP-009.md): Recorded recommendations and purchase linkage supply the recommendation measures. - **Enhances** [AI product suggestion settings](AI-027.md): Recommendations can carry commerce information and purchase linkage. - **Enhances** [Product card configuration and shopping actions](COM-014.md): Recommendations can contain commerce product cards. - **Enhances** [Tracked product search and integrations](COM-018.md): Existing product results can also support recommendations. - **Enhances** [Purchase tracking and sales totals](COM-020.md): Tracked purchase items can be associated with recorded product recommendations. - **Related** [Sales attribution to conversations and operators](COM-021.md): Recommended-product purchase associations are separate from conversation and operator sales credit. ## FAQ and troubleshooting ### Why does one suggestion have a rich card while another does not? Only an unambiguous commerce match receives a store-backed card. Unmatched recommendations can still be used. ### Why is a purchased product not linked to a recommendation? Check that the recommendation use was recorded, the purchase completed through INSIDE tracking, the visitor and SKU match, and the purchase falls within the attribution period. ### Why does recommendation revenue differ from conversation sales? They use different matching rules and attribution periods. Conversation credit is not the same as a recommended-product association. ### Can I change the seven-day period myself? Powerfront can change **Product Suggestion Attribution Days** for you, within the control's 1 to 60 day range.