Cross-channel online advertising attribution

• US 10,169,778 B1
• Filed: 03/26/2015
• Issued: 01/01/2019
• Est. Priority Date: 03/26/2015
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Collin, Guthrie Chipman, Craycroft, Timothy John, Despins, Paula Marie, Edwards, III, Richard Everett, Khoshnevisan, Cyrus, Nilles, John Michael, Perevodchikov, Evgeniy V., Ramamoorthy, Jaikannan, Ramji, Shiven
• Primary Examiner(s): Van Bramer, John
• Application Number:
  US14/669,945
• Time in Patent Office: 1,377 Days
• Field of Search: 705 141, 705 1445
• CPC Class Codes: G06Q 30/0246 Traffic

Abstract

The systems and methods described herein enable measurement of the exposure of a user to advertising campaigns across different channels (e.g., websites accessed by web browsers of laptops, desktops, or mobile computers, electronic readers or tablets, applications on mobile devices, and the like) and credit the exposure of the user to the advertisement to a specific advertisement in the campaign along with subsequent shopping events generated by the user, such as visiting product detail pages, visiting the read all reviews page for a product, adding a product to a wish list, adding a product to a shopping cart, subscribing to a product through an online subscription service, and/or purchasing a product.

Claims

1. A non-transitory computer-readable medium storing computer-executable instructions which, when executed by a processor, cause the processor to perform operations comprising:

receiving, by a server comprising one or more computer processors, event data comprising a first set of data from a first user device and a second set of data from a second user device, the first user device and the second user device in communication with the server over a network;

transmitting, by the server to an identity resolution service (IRS) server, the event data;

transmitting, by the server to a consolidated event store (CES) server, the event data;

receiving, by the server from the IRS server, identity mappings generated using the event data, wherein the identity mappings comprise a mapping of (i) anonymized identifiers associated with a user to (ii) identifiers associated with user devices;

determining a canonical identifier that is associated with the identity mappings, wherein the canonical identifier comprises at least a portion of the anonymized identifiers;

linking, based at least in part on the canonical identifier and the associated identity mappings, in a database, one or more shopping events in the event data with a conversion event, the canonical identifier identifying respective users of both the first user device and the second user device;

receiving, by the server from the CES server, advertisement traffic events identified from the event data;

determining that a first counter is not equal to a first threshold length of time;

determining that a second counter is not equal to a second threshold length of time that is less than the first threshold length of time;

determining that a third counter is equal to a third threshold length of time that is less than the second threshold length of time; and

generating, by the server using one or more attribution models after the third counter equals the third threshold length of time, ad attribution data indicative of an attribution of the conversion event to an advertisement of an advertisement campaign.

2. The non-transitory computer-readable medium of claim 1, wherein the operations further comprise:

transmitting, by the server, the ad attribution data to a datastore; and

retrieving, by the server, the ad attribution data to generate a path to purchase report.

3. The non-transitory computer-readable medium of claim 1, wherein generating the ad attribution data further comprises:

generating, by the server, the ad attribution data on an hourly, daily, and weekly basis.

4. The non-transitory computer-readable medium of claim 1, wherein the ad attribution data is generated using the event data from an attribution window.

5. A method comprising:

receiving, by a server comprising one or more computer processors, event data from a first user device and a second user device in communication with the server over a network;

transmitting, by the server, the event data to an identity resolution service (IRS) server;

transmitting, by the server, the event data to a consolidated event store (CES) server;

receiving, by the server from the IRS server, identity mappings generated using the event data, wherein the identity mappings comprise a mapping of anonymized identifiers associated with a user to identifiers associated with user devices;

determining a canonical identifier, wherein the canonical identifier is associated with the identity mappings, wherein the canonical identifier comprises at least a portion of the anonymized identifiers;

linking, based at least in part on the canonical identifier and the associated identity mappings, in a database, one or more shopping events in the event data with a conversion event, the canonical identifier identifying one or more users of both the first user device and the second user device;

receiving, by the server from the CES server, advertisement traffic events identified from the event data;

determining that a first counter is not equal to a first threshold length of time;

determining that a second counter is not equal to a second threshold length of time that is less than the first threshold length of time;

determining that a third counter is equal to a third threshold length of time that is less than the second threshold length of time; and

generating, by the server, ad attribution data indicative of an attribution of the conversion event to an advertisement of an advertisement campaign.

6. The method of claim 5, further comprising:

transmitting, by the server, the ad attribution data to a datastore;

retrieving, by the server, the ad attribution data; and

modifying, by the server, the one or more attribution models using the ad attribution data.

7. The method of claim 5, further comprising:

executing, by the server, the one or more attribution models, wherein the one or more attribution models use the identity mappings and the advertisement traffic events to generate the ad attribution data.

8. The method of claim 5, wherein the ad attribution data is generated using the event data from an attribution window, wherein the attribution window is a period of time prior to the conversion event used to identify the advertisement that may receive credit for the conversion event.

9. The method of claim 5, further comprising:

setting, by the server, the first threshold to be less than the second threshold; and

setting, by the server, the second threshold to be less than the third threshold.

10. The method of claim 5, further comprising:

identifying, by the server, a first conversion event from the identity mappings and advertisement traffic events;

identifying, by the server, a first candidate advertisement from the identity mappings and the advertisement traffic events;

determining, by the server, the first candidate advertisement is a secondary advertisement;

determining, by the server, a first view discount rate of the first candidate advertisement;

generating, by the server, a first attribution probability using a generated random number and the first view discount rate;

comparing, by the server, the first attribution probability to an attribution threshold; and

generating, by the server, the ad attribution data attributing the first candidate advertisement with the first conversion event in response to determining the first attribution probability exceeds the attribution threshold.

11. The method of claim 10, further comprising:

identifying, by the server, a second conversion event from the identity mappings and advertisement traffic events;

identifying, by the server, a second candidate advertisement from the identity mappings and the advertisement traffic events;

determining, by the server, the second candidate advertisement is a secondary advertisement;

determining, by the server, a second view discount rate of the second candidate advertisement;

generating, by the server, a second attribution probability using a second generated random number and the second view discount rate;

determining, by the server, the second attribution probability does not exceed the attribution threshold;

identifying, by the server, a third candidate advertisement from the identity mappings and the advertisement traffic events;

determining, by the server, the third candidate advertisement is the secondary advertisement;

determining, by the server, a third view discount rate of the third candidate advertisement;

generating, by the server, a third attribution probability using a third generated random number and the third view discount rate;

determining, by the server, the second attribution probability does not exceed the attribution threshold; and

generating, by the server, the ad attribution data attributing the third candidate advertisement with the second conversion event in response to determining the third attribution probability exceeds the attribution threshold.

12. The method of claim 10, wherein the first view discount rate is specified in an associated advertisement campaign.

13. The method of claim 10, wherein the secondary advertisement is a below-the-fold advertisement.

14. The method of claim 10, wherein the first view discount rate is a value between 0 and 1 and the generated random number is a value between 0 and 1.

15. The method of claim 5, wherein generating ad attribution data further comprises:

generating, by the server, a first ad attribution data using the one or more attribution models and a product identifier product engagement scope;

generating, by the server, a second ad attribution data using the one or more attribution models and a brand product engagement scope; and

generating, by the server, a third ad attribution data using the one or more attribution models and an order product sales product engagement scope.

16. The method of claim 15, further comprising:

generating, by the server, one or more reports using the first ad attribution data, the second ad attribution data, or the third ad attribution data.

17. A computer system comprising:

at least one memory that stores computer-executable instructions;

at least one processor configured to access the at least one memory and execute the computer-executable instructions to;

receive event data generated and received from a first device and a second device, the first device and the second device in communication with a computer device over a network;

transmit the event data to an identity resolution service (IRS) server;

transmit the event data to a consolidated event store (CES) server;

receive, the IRS server, identity mappings generated using the event data, wherein the identity mappings comprise a mapping of anonymized identifiers associated with a user to identifiers associated with user devices;

determine a canonical identifier, wherein the canonical identifier is associated with the identity mappings, wherein the canonical identifier comprises at least apportion of the anonymized identifiers;

link, based at least in part on the canonical identifier and the associated identity mappings, in a database, one or more shopping events in the event data with a conversion event, the canonical identifier identifying one or more users of both the first device and the second device;

receive, from the CES server, advertisement traffic events identified from the event data;

determine that a first counter is not equal to a first threshold length of time;

determine that a second counter is not equal to a second threshold length of time that is less than the first threshold length of time;

determine that a third counter is equal to a third threshold length of time that is less than the second threshold length of time; and

generate, using one or more attribution models, ad attribution data indicative of an attribution of the conversion event to an advertisement of an advertisement campaign.

18. The system of claim 17, wherein the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

transmit the ad attribution data to a datastore.

19. The system of claim 18, wherein the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

retrieve the ad attribution data from the datastore; and

modify the one or more attribution models using the ad attribution data.

20. The system of claim 18, wherein to generate the ad attribution data, the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

generate a first ad attribution data using the one or more attribution models and a product identifier product engagement scope;

generate a second ad attribution data using the one or more attribution models and a brand product engagement scope; and

generate a third ad attribution data using the one or more attribution models and an order product sales product engagement scope.

21. The system of claim 17, wherein the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

execute the one or more attribution models, wherein the one or more attribution models use the identity mappings and the advertisement traffic events to generate the ad attribution data.

22. The system of claim 17, wherein the ad attribution data is generated using the event data from an attribution window and wherein the attribution window is a period of time prior to the conversion event used to identify the advertisement that may receive credit for the conversion event.

23. The system of claim 17, wherein the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

identify a first conversion event from the identity mappings and advertisement traffic events;

identify a first candidate advertisement from the identity mappings and the advertisement traffic events;

determine the first candidate advertisement is a secondary advertisement;

determine a first view discount rate of the first candidate advertisement;

generate a first attribution probability using a generated random number and the first view discount rate;

compare the first attribution probability to an attribution threshold; and

generate the ad attribution data attributing the first candidate advertisement with the first conversion event in response to determining the first attribution probability exceeds the attribution threshold.

24. The system of claim 23, wherein the at least one processor is configured to access the at least one memory and to further execute the computer-executable instructions to:

identify a second conversion event from the identity mappings and advertisement traffic events;

identify a second candidate advertisement from the identity mappings and the advertisement traffic events;

determine the second candidate advertisement is a secondary advertisement;

determine a second view discount rate of the second candidate advertisement;

generate a second attribution probability using a second generated random number and the second view discount rate;

determine the second attribution probability does not exceed the attribution threshold;

identify a third candidate advertisement from the identity mappings and the advertisement traffic events;

determine the third candidate advertisement is the secondary advertisement;

determine a third view discount rate of the third candidate advertisement;

generate a third attribution probability using a third generated random number and the third view discount rate;

determine the second attribution probability does not exceed the attribution threshold; and

generate the ad attribution data attributing the third candidate advertisement with the second conversion event in response to determining the third attribution probability exceeds the attribution threshold.

25. The system of claim 23, wherein the first view discount rate is specified in an associated advertisement campaign.

26. The system of claim 23, wherein the first view discount rate is a value between 0 and 1 and the generated random number is a value between 0 and 1.