Abstract: A machine-learned predictive model is trained to predict potential for customer complaint. The model is part of an online concierge system. The online concierge system accesses a customer order that includes one or more items. The online concierge system determines input data for an item of the one or more items. The online concierge system determines a prediction value associated with potential for customer complaint for the item by applying the machine-learned prediction model to the input data. The online concierge system provides the prediction value to a picker client device associated with a picker who is assigned the item. The picker client device presents an alert to the picker based in part on the prediction value, and the alert includes a message that is customized to mitigate a cause of potential customer complaint for the item.

Abstract: The disclosed computer-implemented method may include matching transportation requests. By collecting and batching match requests over an extended period, a dynamic transportation matching system may identify more efficient matches (e.g., may match transportation requests with greater overlaps). In addition, by dynamically setting and/or extending the upper bound of time that a transportation request may remain batched with other transportation requests, the dynamic transportation matching system may account for contextual information thereby situationally improving matching efficiencies made possible with higher upper bounds while avoiding requestor dissatisfaction, lost conversions, or other inefficiencies that may result from upper bounds that are too high.

Abstract: An online concierge system receives information describing the progress of a picker servicing a batch of existing orders and predicts a first likelihood the picker will finish servicing the batch within a threshold amount of time based on the picker's progress and information describing the batch. If the first likelihood exceeds a threshold likelihood, the system accesses a machine learning model trained to predict a second likelihood the picker will accept a batch of new orders for servicing while servicing the batch of existing orders. The system applies the model to inputs including a set of attributes of the picker and the picker's progress to predict the second likelihood. The system matches batches of new orders with pickers based on the second likelihood and sends one or more requests to service one or more batches matched with the picker to a client device associated with the picker.

Abstract: An online concierge system receives information describing the progress of a picker servicing a batch of existing orders and a service request for an order. The system identifies picker attributes of the picker and order attributes of the order and each existing order of the set and accesses a machine learning model trained to predict a likelihood the picker will accept an add-on request to add the order to the batch of existing orders. To predict the likelihood, the system applies the model to the picker attributes, the progress of the picker, and the order attributes. The system determines a cost associated with sending the add-on request to the picker based on the likelihood and assigns the order to a set of orders based on the cost. The system sends the add-on request to the picker responsive to determining the order is assigned to the batch of existing orders.

Abstract: A concierge system sends batches of orders to pickers that they can review and accept in a batch list on a client device. Each batch in the batch list is presented with a hide option that enables the picker to hide a batch that they do not intend to accept. In response to receiving a hide signal, the system extracts features associated with the batch and stores those features with a negative indication of the picker towards the batch. The hide signal provides the system with a higher quality signal indicating the picker's negative intent regarding an order, as compared to simply ignoring the order in favor of fulfilling another order. This higher quality signal is then used to train models to better predict events related to the pickers' acceptance of orders, such as for ranking orders for pickers or for predicting fulfillment times.

Abstract: Embodiments provide techniques, including systems and methods, for determining matches of requestors and providers based on a dynamic provider eligibility model. For example, a request matching model uses an estimated arrival time for a requestor and estimated travel times for available providers to a pickup location to determine eligible providers for matching to a ride request. The matching model determines those providers that are far enough away from the request location to allow the requestor time to arrive at the pickup location without matching providers that are too far away, causing delay for the requestor and lowering the efficiency of the system by taking provider system resources from other service areas and increasing provider downtime upon matching. Additionally, embodiments provide more efficient matching processing leading to fewer canceled matched requests, fewer requests for a successful match, and fewer system resources necessary to meet requestor demand.

Abstract: Techniques for predicting a wait time for a shopper based on a location the shopper's client device are presented. A system identifies a shopper's current location and uses a machine learning model to predict a wait time until the shopper will receive one or more orders. The machine learning model is trained to use input features including a number of orders received during a current time period for fulfillment near the current location, a number of other shoppers available for fulfilling orders during the current time period near the current location, historical information about a presentation of a plurality of orders to a plurality of shoppers near the current location, and historical information about the shopper and the other nearby available shoppers. The system then sends the predicted wait time to the client device for presentation to the shopper.

Abstract: An online concierge system accesses and applies a model to predict likelihoods of acceptance of a service request for an order by pickers. The system accesses timespan distributions for accepted service requests and identifies sets of pickers based on the order. Based on the likelihoods and distributions, the system generates simulated responses of the sets of pickers to the service request and trains an additional model based on attributes of the order, the simulated responses, and information associated with corresponding sets of pickers. The system receives a new order, identifies additional sets of pickers based on the new order, and applies the additional model to predict responses of the additional sets of pickers to an additional service request for the new order. Based on the predicted responses and a delivery time associated with the new order, a minimum number of pickers to send the additional service request is determined.

Abstract: An online concierge system receives location information associated with pickers and actual orders associated with a geographical zone. A model trained to predict a likelihood an actual order associated with the zone will be available for servicing within a timeframe is accessed and applied to forecasted orders. Each picker is matched to an order for servicing by minimizing a value of a function that is based on a difference between a location associated with each picker matched to an actual order and an associated retailer location, a difference between the location associated with each picker matched to a forecasted order and an associated retailer location, and the predicted likelihood. Recommendations for accepting an actual order, moving to a retailer location associated with a forecasted order, or checking back later with the system are generated based on the matches and sent for display to a client device associated with each picker.

Abstract: Embodiments provide techniques, including systems and methods, for determining matches of requestors and providers based on a dynamic provider eligibility model. For example, a request matching model uses an estimated arrival time for a requestor and estimated travel times for available providers to a pickup location to determine eligible providers for matching to a ride request. The matching model determines those providers that are far enough away from the request location to allow the requestor time to arrive at the pickup location without matching providers that are too far away, causing delay for the requestor and lowering the efficiency of the system by taking provider system resources from other service areas and increasing provider downtime upon matching. Additionally, embodiments provide more efficient matching processing leading to fewer canceled matched requests, fewer requests for a successful match, and fewer system resources necessary to meet requestor demand.

Abstract: An online concierge system iteratively makes a batch of one or more orders available to an increasing number of shoppers to choose to fulfill. Each shopper may choose to accept or reject a batch for fulfillment. To improve batch acceptance and matching between batches and shoppers, the batches are scored with respect to expected resource costs, likelihood of acceptance by the shopper, and/or other quality metrics to iteratively offer the batch to an increasing number of shoppers (prioritizing the scoring factors) until a shopper accepts. The number of shoppers notified of the batch and the frequency that additional shoppers are selected may vary based on characteristics of the batch and likelihood the batch will be accepted by a shopper.

Abstract: The online concierge system generates task units based on orders and assigns batches of task units to pickers. The online concierge system generates task units based on received orders. The online concierge system generates permutations of these task units to generate candidate sets of task batches. The online concierge system scores each of these candidate sets, and selects a set of task batches to assign to pickers based on the scores. Additionally, to determine which task UI to display to the picker, the picker client device uses a UI state machine. The UI state machine is a state machine where each state corresponds to a task UI to display on the picker client device. The state transitions between the UI states of the UI state machine indicate which UI state to transition to from a current UI state based on the next task unit in the received task batch.

Abstract: The online concierge system generates task units based on orders and assigns batches of task units to pickers. The online concierge system generates task units based on received orders. The online concierge system generates permutations of these task units to generate candidate sets of task batches. The online concierge system scores each of these candidate sets, and selects a set of task batches to assign to pickers based on the scores. Additionally, to determine which task UI to display to the picker, the picker client device uses a UI state machine. The UI state machine is a state machine where each state corresponds to a task UI to display on the picker client device. The state transitions between the UI states of the UI state machine indicate which UI state to transition to from a current UI state based on the next task unit in the received task batch.

Abstract: The disclosed computer-implemented method may include matching transportation requests. By collecting and batching match requests over an extended period, a dynamic transportation matching system may identify more efficient matches (e.g., may match transportation requests with greater overlaps). In addition, by dynamically setting and/or extending the upper bound of time that a transportation request may remain batched with other transportation requests, the dynamic transportation matching system may account for contextual information thereby situationally improving matching efficiencies made possible with higher upper bounds while avoiding requestor dissatisfaction, lost conversions, or other inefficiencies that may result from upper bounds that are too high.

Abstract: The disclosed computer-implemented method may include matching transportation requests. By collecting and batching match requests over an extended period, a dynamic transportation matching system may identify more efficient matches (e.g., may match transportation requests with greater overlaps). In addition, by dynamically setting and/or extending the upper bound of time that a transportation request may remain batched with other transportation requests, the dynamic transportation matching system may account for contextual information thereby situationally improving matching efficiencies made possible with higher upper bounds while avoiding requestor dissatisfaction, lost conversions, or other inefficiencies that may result from upper bounds that are too high.

Abstract: Embodiments provide techniques, including systems and methods, for determining matches of requestors and providers based on a dynamic provider eligibility model. For example, a request matching model uses an estimated arrival time for a requestor and estimated travel times for available providers to a pickup location to determine eligible providers for matching to a ride request. The matching model determines those providers that are far enough away from the request location to allow the requestor time to arrive at the pickup location without matching providers that are too far away, causing delay for the requestor and lowering the efficiency of the system by taking provider system resources from other service areas and increasing provider downtime upon matching. Additionally, embodiments provide more efficient matching processing leading to fewer canceled matched requests, fewer requests for a successful match, and fewer system resources necessary to meet requestor demand.

Abstract: Embodiments provide techniques, including systems and methods, for determining matches of requestors and providers based on a dynamic provider eligibility model. For example, a request matching model uses an estimated arrival time for a requestor and estimated travel times for available providers to a pickup location to determine eligible providers for matching to a ride request. The matching model determines those providers that are far enough away from the request location to allow the requestor time to arrive at the pickup location without matching providers that are too far away, causing delay for the requestor and lowering the efficiency of the system by taking provider system resources from other service areas and increasing provider downtime upon matching. Additionally, embodiments provide more efficient matching processing leading to fewer canceled matched requests, fewer requests for a successful match, and fewer system resources necessary to meet requestor demand.

Abstract: The present disclosure relates to systems, non-transitory computer-readable media, and methods for identifying provider devices and corresponding vehicles as candidates to transport time-sensitive (or otherwise prioritized) requestors and providing prioritized transportation options as fast passes for display on such requestors' devices. To provide such a prioritized transportation option, the disclosed systems can identify provider devices either matched or unmatched with requestors as candidates for prioritized transport based on estimated times of arrivals (ETAs) of vehicles for the candidate provider devices at the requestor's pickup location. Based on the ETAs at the pickup location, the disclosed systems can select a closest provider device from among the candidate provider devices to transport a prioritized requestor. After matching the prioritized requestor to the closest provider device, the disclosed systems can further search for providers with sooner ETAs.

Abstract: The disclosed computer-implemented method may calculate individual utility metrics for each combination of potential transportation requestors and cancellations to arrive at a more accurate total expected utility for shared transportation. In one embodiment, the method may reduce computation resource requirements by calculating each cancellation probability independently. In some examples, the method may only calculate utility metrics for some fixed number and/or percentage of the most probable combinations. In some embodiments, the method may account for travel time and/or distance when calculating utility metrics. By making matching decisions for shared transportation that account for the possibility of cancellation, the method may improve the efficiency of the transportation network. Various other methods, systems, and computer-readable media are also disclosed.

Abstract: The disclosed computer-implemented method may include implementing factors and conversion probabilities when matching a transportation requestor to a transportation provider. Matches between transportation requestors and transportation providers that rely solely on an estimation of arrival time may not give requestors or providers the best transportation options. Lacking these optimal transportation options, requestors and providers may move to other platforms. By looking at a various transportation factors and conversion probabilities, the method may provide optimal transportation options to both requestors and providers. Various other methods, systems, and computer-readable media are also disclosed.