# US 2020/0402097 A1 · Generating optimized in-channel and cross-channel promotion recommendations using free shipping qualifier

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2020/0402097 A1
- Document ID: US20200402097A1
- Status: Published application
- Kind: A1
- Application number: 16/915,098
- Filed: 2020-06-29
- Published: 2020-12-24
- Earliest claimed: 2014-04-22
- Assignees: Groupon, Inc.
- Applicants: Groupon, Inc.
- Inventors: Feili HOU; Vyomkesh TRIPATHI; Nipun AGARWAL; Rajesh Girish PAREKH
- Web page: https://patentgrep.com/patent/US20200402097A1

## CPC

- G06Q 30/0255: {based on user history}
- G06F 16/24578: {using ranking}
- G06F 16/955: using information identifiers, e.g. uniform resource locators [URL]
- G06Q 30/02: Marketing; Price estimation or determination; Fundraising
- G06Q 30/0269: {based on user profile or attribute}
- G06Q 30/0277: {Online advertisement}
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

In general, embodiments of the present invention provide systems, methods and computer readable media for recommending contextually relevant promotions to consumers in order to facilitate their discovery of promotions that they are likely to purchase from a promotion and marketing service.

## Claims

### Claim at position 1

1-20. (canceled)

### Claim 21

A computer-implemented method to implement similar promotions ranking, the method comprising:
receiving, by a processor, promotion data describing a set of promotions, wherein the promotion data comprises promotion attributes associated with the promotions, and wherein the set of promotions comprises a particular promotion currently being viewed via a consumer device associated with a consumer;
generating, by the processor, a ranked set of promotions based on the promotion data, wherein generating the ranked set of promotions is based on at least one of adaptive collaborative filtering signals and static similarity signals, wherein generating, by the processor, the ranked set of promotions based on adaptive collaborative filtering signals comprises:
receiving promotion data describing a plurality of promotion pairs (X,Y), wherein the promotion data comprises clickstream data, wherein the particular promotion is a member X promotion of each promotion pair, wherein a member Y promotion of each promotion pair of the plurality of promotion pairs (X, Y) respectively represents one of a group of other live promotions, and wherein each promotion pair (X, Y) of the plurality of promotion pairs was selected based on determining that similarity metrics respectively associated with the promotion pair (X,Y) are respectively above a similarity and popularity threshold;
for each promotion pair (X,Y) of the plurality of promotion pairs (X,Y), calculating, using the associated similarity metrics, a co-view score representing a likelihood that the member Y promotion will be selected by the consumer while the consumer is viewing the member X promotion;
for the member X promotion, generating a ranked list of all member Y promotions in the plurality of promotion pairs, wherein the member Y promotions are ranked in descending order of their associated co-view scores;
generating a set of recommended promotions for the consumer by selecting a subset of the top-ranked promotions from the ranked list of all member Y promotions; and
transmitting the set of recommended promotions to the consumer device for display according to their respective rankings, wherein one or more promotions of the set of recommended promotions are rendered for display together with member X promotion currently being viewed via the consumer device.

### Claim 22 (depends on 21)

The computer-implemented method of claim 21, wherein the similarity metrics comprise:
confidence, P(clickY|clickX), a probability of the consumer clicking Y given the consumer is viewing X, wherein confidence is calculated based on a ratio of a number of consumers who clicked both X and Y to a number of consumers who clicked X and could have clicked Y;
support, P (clickY), a prior probability of consumers clicking Y, wherein support is calculated based on a ratio of a number of consumers who clicked on Y to a number of consumers who could have clicked on Y; and
lift, a ratio of confidence to support.

### Claim 23 (depends on 22)

The computer-implemented method of claim 22, wherein determining that the similarity metrics associated with a promotion pair of the plurality of promotion pairs are respectively above the similarity and popularity threshold comprises:
comparing, by the processor, the confidence metric to a confidence threshold; and
comparing, by the processor, the lift metric to a lift threshold.

### Claim 24 (depends on 22)

The computer-implemented method of claim 22, wherein calculating the co-view score comprises:
calculating, by the processor, a product of confidence and lift.

### Claim 25 (depends on 21)

The computer-implemented method of claim 21, wherein generating the ranked set of promotions based on static similarity signals comprises:
for each of the set of promotions other than the particular promotion currently being viewed by the consumer, calculating, by the processor, at least one static similarity score based on comparing one or more attributes of the promotion to attributes of the particular promotion; and
generating, by the processor, a ranked list of the set of promotions other than the particular promotion, wherein the promotions are ranked in descending order of their associated static similarity scores.

### Claim 26 (depends on 25)

The computer-implemented method of claim 25, wherein generating the set of recommended promotions for the consumer comprises:
selecting, by the processor, a subset of the top-ranked promotions in the ranked list.

### Claim 27 (depends on 25)

The computer-implemented method of claim 25, wherein the attributes of the promotion comprise one or more of category, subcategory, location, promotion persona, or promotion launch date.

### Claim 28 (depends on 21)

The computer-implemented method of claim 21, wherein generating, by the processor, the ranked set of promotions comprises:
determining whether the clickstream data are sparse data; and in an instance in which the clickstream data are sparse data,
generating the ranked set of promotions using adaptive collaborative filtering signals; and using the static similarity signals to backfill the ranked set of promotions using adaptive collaborative filtering signals.

### Claim 29 (depends on 28)

The computer-implemented method of claim 28, further comprising:
in an instance in which the clickstream data are not sparse data,
generating the ranked set of promotions using adaptive collaborative filtering signals.

### Claim 30 (depends on 21)

The computer-implemented method of claim 21, wherein the particular promotion is currently being purchased by the consumer, wherein generating the ranked set of promotions is based on one or more cross-sell signals based on historical co-purchases from a service associated with the particular promotion over a period of time, and wherein the one or more cross-sell signals comprise user-specific recommendations, a views-based conversion signal, a purchase-based signal, a search history signal, and newly launched promotions relevant to the consumer.

### Claim 31 (depends on 30)

The computer-implemented method of claim 30, wherein the particular promotion is associated with a first service, and wherein generating the ranked set of promotions comprises:
calculating, by the processor and using the promotion data, a co-purchase score for each of the promotions associated with a second service based in part on calculating a co-purchase frequency between consumers who purchased promotions associated with the first service and consumers who purchased promotions associated with the second service.

### Claim 32 (depends on 31)

The computer-implemented method of claim 31, wherein calculating the co-purchase frequency comprises:
calculating probability distribution and Jaccard scores.

### Claim 33 (depends on 30)

The computer-implemented method of claim 30, wherein generating the ranked set of promotions comprises:
generating a ranked list of promotions by ranking the promotions in descending order of their respective co-purchase scores.

### Claim 34 (depends on 33)

The computer-implemented method of claim 33, wherein generating the ranked set of promotions further comprises:
selecting a subset of the top-ranked promotions in the ranked list for transmission to the consumer device;
in an instance in which the subset of the top-ranked promotions is smaller than a threshold quantity of promotions,
selecting, using the promotion data, a set of top revenue-earning freshly launched promotions; and
backfilling the selected subset of the top-ranked promotions using the set of top revenue-earning freshly launched promotions.

### Claim 35 (depends on 30)

The computer-implemented method of claim 30, further comprising:
optimizing the set of recommended promotions for the consumer based on a free shipping qualifier threshold amount by performing operations comprising:
selecting a subset of the top-ranked promotions that each meet a price constraint, wherein the price constraint is a sum of a promotion price and a current purchase cart total of the consumer being greater than the free shipping qualifier threshold amount and less than a sum of the free shipping qualifier threshold amount plus an increment; and
in an instance in which the subset of the top-ranked promotions is smaller than a threshold quantity of promotions,
selecting, based in part on the price constraint, additional promotions by selecting one up to a maximum quantity of promotions from each of a set of services respectively associated with at least one of the promotions of the ranked list of promotions.

### Claim 36

A computer program product, stored on a computer readable medium, comprising instructions that when executed on one or more computers cause the one or more computers to:
receive promotion data describing a set of promotions, wherein the promotion data comprises promotion attributes associated with the promotions, and wherein the set of promotions comprises a particular promotion currently being viewed via a consumer device associated with a consumer;
generate a ranked set of promotions based on the promotion data, wherein generating the ranked set of promotions is based on at least one of adaptive collaborative filtering signals and static similarity signals, wherein generating the ranked set of promotions based on adaptive collaborative filtering signals comprises:
receiving promotion data describing a plurality of promotion pairs (X,Y), wherein the promotion data comprises clickstream data, wherein the particular promotion is a member X promotion of each promotion pair, wherein a member Y promotion of each promotion pair of the plurality of promotion pairs (X, Y) respectively represents one of a group of other live promotions, and wherein each promotion pair (X, Y) of the plurality of promotion pairs was selected based on determining that similarity metrics respectively associated with the promotion pair (X,Y) are respectively above a similarity and popularity threshold;
for each promotion pair (X,Y) of the plurality of promotion pairs (X,Y), calculating, using the associated similarity metrics, a co-view score representing a likelihood that the member Y promotion will be selected by the consumer while the consumer is viewing the member X promotion;
for the member X promotion, generating a ranked list of all member Y promotions in the plurality of promotion pairs, wherein the member Y promotions are ranked in descending order of their associated co-view scores;
generating a set of recommended promotions for the consumer by selecting a subset of the top-ranked promotions from the ranked list of all member Y promotions; and
transmitting the set of recommended promotions to the consumer device for display according to their respective rankings, wherein one or more promotions of the set of recommended promotions are rendered for display together with member X promotion currently being viewed via the consumer device.

### Claim 37 (depends on 36)

The computer program product of claim 36, wherein generating the ranked set of promotions based on static similarity signals comprises:
for each of the set of promotions other than the particular promotion currently being viewed via the consumer device, calculating at least one static similarity score based on comparing one or more attributes of the promotion to attributes of the particular promotion; and
generating a ranked list of the set of promotions other than the particular promotion, wherein the promotions are ranked in descending order of their associated static similarity scores.
