OPTIMIZING DELIVERY OF CONTENT ITEMS TO USERS OF AN ONLINE SYSTEM TO PROMOTE PHYSICAL STORE VISITS AS CONVERSION EVENTS
An online system promotes physical store visits by presenting users with content items for a physical store location and subsequently logs visits of online system users to the physical store location to track performance of a campaign associated with the presented content item. The online system registers attention events associated with the presented content items and registers attention events as store front visit conversion events if, within a predetermined period of time from a valid attention event, a user has subsequently gone in and visited the physical store front location. The online system then uses this store visit data to build a model that optimizes for store visits as a conversion.
1 . A method comprising:
maintaining, by an online system, a user profile for each of a plurality of users of the online system, the user profile including demographic information for each user;
segmenting the demographic information into a plurality of user buckets, each user bucket including a different range of values for one or more user demographics relative to other user buckets of the plurality of user buckets;
providing content items associated with a campaign for display to users of the online system, each content item in the campaign being associated with a physical store front location and describing a product or service available for purchase at the physical store front location;
receiving attention events for the content items from a content item viewing set of users, each attention event corresponding to a user of the viewing set of users demonstrating an interest in a content item;
storing the attention events at the online system, each attention event being stored to a bucket of the plurality of user buckets corresponding to a user associated with the attention event and the demographic information of the user maintained by the online system in a user profile of the user, the attention event being valid for a predetermined time period and invalid upon expiration of the predetermined time period;
obtaining store front visit events from a visiting set of users of the online system, each store front visit event corresponding to a client device of a user of the visiting set of users being identified at a physical store front location associated with a content item associated with the campaign; and
determining a number of store front visit conversion events for the campaign, each store front visit conversion event corresponding to a user of the viewing set of users performing an attention event and subsequently being identified at the physical store font location associated with the campaign within the predetermined period of time.
2 . The method of claim 1 , further comprising:
identifying user buckets associated with a number store front visit conversion events greater than a threshold number;
training a model to identify a refined set of users that are most likely to visit the physical store font location after viewing the content item based on the identified user buckets associated with the number store front visit conversion events greater than the threshold number; and
providing content items associated with the campaign for display to users of the refined set of users.
3 . The method of claim 1 , further comprising:
identifying user buckets associated with a number store front visit conversion events greater than a threshold number; and
adjusting targeting criteria for the campaign to provide content items to users associated with the identified user buckets.
4 . The method of claim 1 , wherein determining the number of store front conversion events for the campaign includes:
identifying, in each user bucket, attention events with corresponding store front visit events in the plurality of user buckets.
5 . The method of claim 1 , wherein the attention events are at least one of a gaze direction of a user being directed at the content item for more than a threshold period of time, the user selecting the content item to receive additional information associated with the content item, or the user obtaining directions to the physical store front location via the content item.
6 . The method of claim 1 , wherein the one or more user demographics include at least one of age of the user, gender of the user, a city associated with the user.
7 . The method of claim 6 , wherein the different range of values for the one or more user demographics corresponds an age range associated with each user bucket, and wherein each user bucket is associated with a different combination of the age range and gender.
8 . The method of claim 6 , wherein the one or more user demographics include a user location propensity score corresponding to a propensity of the user to visit the physical store front location, wherein determining the user-location propensity score for a user includes:
identifying, based on online behavior of the user, a group of similar users who have opted-in to sharing location history, the location history of the group of similar users identifying one or more locations and a visit frequency associated with each of the one or more locations; and
determining the location propensity score for the user by:
comparing the physical store front location to the one or more locations from the obtained user history;
determining a similarity score between each of the one or more locations to the physical store front location weighted for the visit frequency associated with each location; and
aggregating the similarity scores to determine the location propensity score.
9 . A method comprising:
providing, by an online system, content items associated with a campaign for display to users of the online system, each content item in the campaign being associated with a physical store front location and describing a product or service available for purchase at the physical store front location;
receiving attention events for the content items from a content item viewing set of users, each attention event corresponding to a user of the viewing set of users demonstrating an interest in a content item;
storing the attention events at the online system, each attention event being stored to a user bucket corresponding to demographic information of a user, the attention event being valid for a predetermined time period;
obtaining store front visit events from a visiting set of users of the online system, each store front visit event corresponding to a client device of a user of the visiting set of users being identified at the physical store front location associated with the content item associated with the campaign;
determining a number of store front visit conversion events for the campaign, each store front visit conversion event corresponding to a user of the viewing set of users performing an attention event and subsequently being identified at the physical store font location associated with the campaign within the predetermined period of time;
identifying user buckets associated with a number store front visit conversion events greater than a threshold number; and
adjusting, by the online system, targeting criteria for the campaign to provide content items to users associated with the identified user buckets.
10 . The method of claim 9 , further comprising:
maintaining, by the online system, a user profile for each of a plurality of users of the online system, the user profile including the demographic information for each user; and
segmenting the demographic information into a plurality of user buckets, each user bucket including a different range of values for one or more user demographics relative to other user buckets of the plurality of user buckets.
11 . The method of claim 9 , wherein the attention events are at least one of a gaze direction of a user being directed at the content item for more than a threshold period of time, the user selecting the content item to receive additional information associated with the content item, or the user obtaining directions to the physical store front location via the content item.
12 . The method of claim 9 , wherein the one or more user demographics include at least one of age of the user, gender of the user, a city associated with the user.
13 . A computer program product comprising a non-transitory computer-readable storage medium containing computer program code for:
providing content items associated with a campaign for display to users of an online system, each content item in the campaign being associated with a physical store front location and describing a product or service available for purchase at the physical store front location;
receiving attention events for the content items from a content item viewing set of users, each attention event corresponding to a user of the viewing set of users demonstrating an interest in a content item;
storing the attention events at the online system, each attention event being stored to a user bucket corresponding to demographic information of a user, the attention event being valid for a predetermined time period;
obtaining store front visit events from a visiting set of users of the online system, each store front visit event corresponding to a client device of a user of the visiting set of users being identified at the physical store front location associated with the content item associated with the campaign; and
determining a number of store front visit conversion events for the campaign, each store front visit conversion event corresponding to a user of the viewing set of users performing an attention event and subsequently being identified at the physical store font location associated with the campaign within the predetermined period of time.
14 . The computer program product of claim 13 , further containing computer program code for:
maintaining a user profile for each of a plurality of users of the online system, the user profile including the demographic information for each user; and
segmenting the demographic information into a plurality of user buckets, each user bucket including a different range of values for one or more user demographics relative to other user buckets of the plurality of user buckets.
15 . The computer program product of claim 13 , further containing computer program code for:
identifying user buckets associated with a number store front visit conversion events greater than a threshold number; and
adjusting targeting criteria for the campaign to provide content items to users associated with the identified user buckets.
16 . The computer program product of claim 13 , wherein determining the number of store front conversion events for the campaign includes:
identifying, in each user bucket, attention events with corresponding store front visit events in the plurality of user buckets.
17 . The computer program product of claim 13 , wherein the attention events are at least one of a gaze direction of a user being directed at the content item for more than a threshold period of time, the user selecting the content item to receive additional information associated with the content item, or the user obtaining directions to the physical store front location via the content item.
18 . The computer program product of claim 13 , wherein the one or more user demographics include at least one of age of the user, gender of the user, a city associated with the user.
19 . The computer program product of claim 18 , wherein the different range of values for the one or more user demographics corresponds an age range associated with each user bucket, and wherein each user bucket is associated with a different combination of the age range and gender.
20 . The computer program product of claim 18 , wherein the one or more user demographics include a user location propensity score corresponding to a propensity of the user to visit the physical store front location, wherein determining the user location propensity score for a user includes:
identifying, based on online behavior of the user, a group of similar users who have opted-in to sharing location history, the location history of the group of similar users identifying one or more locations and a visit frequency associated with each of the one or more locations; and
determining the location propensity score for the user by:
comparing the physical store front location to the one or more locations from the obtained user history;
determining a similarity score between each of the one or more locations to the physical store front location weighted for the visit frequency associated with each location; and
aggregating the similarity scores to determine the location propensity score.