Online Concierge Search Suggestion Using Conversion Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional online concierge systems struggle to effectively suggest search queries to users based on previous search queries, especially in determining whether a search query resulted in a conversion or not.
Innovation Solution
The online concierge system utilizes an item graph and machine learning search suggestion models to generate rankings of candidate items and suggest subsequent search queries. It differentiates between conversion and non-conversion graphs to provide tailored suggestions based on previous user interactions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional online concierge systems present items matching search terms, then users can find relevant items, but the system cannot effectively suggest subsequent search queries based on conversion status
Solution Approach 1:
The system segments the suggestion mechanism into two distinct pathways: a conversion graph for users who made purchases and a non-conversion graph for users who did not. This segmentation allows the system to provide tailored search query suggestions based on the user's conversion status, resolving the contradiction by preserving conversion information and adapting suggestions accordingly.
Solution Approach 2:
The system changes the parameter of suggestion source based on conversion status. When a conversion occurs, the system switches to using the conversion graph; when no conversion occurs, it uses the non-conversion graph. This dynamic parameter change enables effective adaptation of search suggestions while preserving the distinction between converted and non-converted users.
2Adaptability or versatility
If the system uses a single suggestion model for all users, then the system is simpler to maintain, but it cannot provide tailored suggestions based on whether searches resulted in conversions
Solution Approach 1:
The system divides the suggestion infrastructure into two separate graphs: a conversion graph containing search queries that led to purchases, and a non-conversion graph containing searches that did not result in purchases. This segmentation enables personalized suggestions while maintaining clear, manageable data structures that are easier to maintain than a single complex unified model.
Solution Approach 2:
The system pre-processes and stores search query data into two distinct graphs based on conversion status before suggestions are needed. This preliminary action organizes the data in advance, making the suggestion generation process simpler and more efficient when a user searches, despite the initial complexity of maintaining two graphs.
3Measurement precision
If the system tracks detailed conversion data for all search queries, then it can generate accurate conversion graphs, but the data processing and storage requirements increase
Solution Approach 1:
The system extracts only the essential information needed for conversion tracking: the search query text and its conversion status. By taking out only these critical elements and storing them in specialized graphs, the system achieves high measurement precision for conversion tracking while minimizing data storage requirements compared to tracking all possible search attributes.
Data Source
AI summary
An online concierge system suggests subsequent search queries based on previous search queries and whether the previous search queries resulted in conversions. The online concierge system trains a machine learning model using previous delivery orders and whether initial and subsequent search queries in the previous delivery orders resulted in conversions. When the online concierge system receives a search query to identify one or more items from a customer, the online concierge system parses the search query into combinations of terms and identifies items related to the search query. In response to the search query resulting in a conversion, the online concierge system retrieves a conversion graph and presents a suggested subsequent search query based on the conversion graph. In response to the search query not resulting in a conversion, the online concierge system retrieves a non-conversion graph and presents a suggested subsequent search query based on the non-conversion graph.