US 11,379,874 B1Granted

Recommending unique product inventory

AssigneeAirbnb, Inc.

InventorsShike Mei, Teng Wang, Shawn Chen, Ganesh Venkataraman

Life of this patent

Estimating the term…

Loading patent…

Abstract

One of more unique products can be selected for advertisement over a digital marketing channel. The selection is based on the calculation of a base impression budget, calculated per product per day, which calculation considers information related to a product, including supply and demand at the market level, characteristics of the property, and popularity of the listing. A real-time, or current, impression budget is calculated to determine whether a particular product should be recommended to a user. Every time the product is advertised to a user, a user intent value is calculated, indicating the user's likelihood of purchasing the product within a given period of time, along with the user intent of every user to which the product has been advertised. The user intent calculation may consider information specific to the user, such as the user's activity history and profile. These user intent values are subtracted from the base impression budget to obtain a real-time impression budget. If the real-time budget is greater than zero, the product will be advertised to a user. By these means, a bound is set on the number of times a product may be advertised before it is assumed to be sold.

Claims17

Dependent claims

A method for real-time transmission of a web-based advertisement for a single-quantity product from an advertiser to a user via a digital advertising channel, the method comprising:

collecting a set of user activity data relating to user activity with an advertiser from a database, the user activity data includes a search history of the user with the advertiser and a purchase history of the user with the advertiser, and wherein user activity data is stored in the database in association with information related to (a) one or more products of the advertiser and/or (b) one or more advertisement communications of the advertiser;
creating a training set comprising the set of user activity data;
training a machine learning model using the training set to generate a user intent value;
obtaining, from a memory, a target advertisement for a single quantity product from the advertiser for transmission to the user via the digital advertising channel;
storing, in a memory, a market-level characteristic relating to the product and a product-specific characteristic of the product;
calculating, by a processor, a base budget for the product based on the market-level characteristic and the product-specific characteristic;
transmitting, by the processor, the target advertisement for the product to the user via the digital advertising channel;
calculating, by the processor, using the machine learning model, the user intent value from a viewing of the target advertisement for the product by the user, wherein the calculated user intent value reflects the user's predicted likelihood to purchase the product based on the target advertisement;
determining, by the processor, whether the product has been sold;
calculating, by the processor, a budget allocation for the product, wherein calculating the budget allocation includes:
subtracting the calculated user intent value from the calculated base budget for the product to generate the budget allocation;
or
setting the budget allocation to zero upon the determination that the product has been sold;
in a case that the budget allocation is equal to zero, determining not to transmit the target advertisement to a second user via the digital advertising channel;
and
in a case that the budget allocation is a value greater than zero:
(i) transmitting, by the processor, the target advertisement to the second user via the digital advertising channel;
and
(ii) recalculating, by the processor, the budget allocation for the product based on a user intent value calculated from a viewing of the advertisement by the second user so as to decrease the value of the budget allocation, wherein the calculated user intent value for the second user is different from the calculated user intent value for the user.

16 more elements

A system comprising:

a memory, wherein the memory is configured to store a product characteristic relating to a single-quantity product, a user characteristic relating to the activity of a user on a digital platform associated with the single-quantity product, and one or more machine learning models to generate an intent value using information stored in the memory;
at least one processor coupled to the memory, the at least one processor being configured to execute instructions stored in the memory to:
(a) calculate, based on the product characteristic, a budget value for the single-quantity product, the budget value being associated with a number of times the product may be advertised to any of a plurality of users via any of a plurality of digital advertising channels;
(b) determine whether the product has been advertised to a user of the plurality of users;
(c) calculate, in a case that the product has been advertised to a user of the plurality of users, using the one or more machine learning models, a first intent value indicating a predicted likelihood of the user to purchase the product, based on the user characteristic for that user;
(d) subtract, from the calculated budget value, the calculated first intent value, so as to generate an updated budget value;
(e) in a case that the updated budget value is equal to zero, determining not to deliver an advertisement to any of the plurality of users via any of the plurality of digital advertising channels;
(f) in a case that the updated budget value is a non-zero value, deliver a web-based advertisement for the product to any of the plurality of users via a digital advertising channel and repeating steps (b)-(f) until the updated budget value is equal to zero, wherein each calculated intent value is unique to a corresponding user of the plurality of users.

5 more elements

A method performed by a computer system connected, via a network, to a remote server, the method comprising:

(a) obtaining, from the remote server, a first set of user metrics relating to the delivery of advertising content to a user via any of a plurality of digital advertising channels;
(b) obtaining, from a memory of the computer system, a second set of user metrics relating to user activity on the computer system via any of the plurality of digital advertising channels;
(c) creating a training set comprising the first set of user metrics and the second set of user metrics;
(d) training a machine learning model to generate a user intent value using the training set;
(e) obtaining, from a memory of the computer system, information relating to a single-quantity property listing;
(f) calculating a base budget allocation B for advertisement of the property listing via any of the plurality of digital advertising channels, based on the information relating to the property listing;
(g) delivering a web-based advertisement to a plurality of users U, the delivery being respectively performed for each of the plurality of users U via any digital advertising channel of the plurality of digital advertising channels;
(h) calculating, using the machine learning model, for each user u of the plurality of users U, a user intent value Vu indicating the user's predicted likelihood of booking the property listing, wherein each user intent value is unique to the user;
(i) calculating a revised budget allocation R for advertisement of the property listing via any of the plurality of digital advertising channels, the calculating being performed in accordance with the following formula:
R
=
P
*
(
B
-
∑
u
⁢
∈
⁢
U
⁢
(
V
u
)
)
where R is a revised budget allocation for advertisement of the property listing, B is the calculated base budget allocation for the product, U is the plurality of users to whom advertisements for the property listing have been delivered, Vu is the calculated user intent value of a respective user u of the plurality of users U, and P is a boolean value that denotes if the property listing has been booked;
(j) determining, in a case that R is equal to zero, not to transmit the advertisement for the property listing to any user of the plurality of users U;
and
(k) delivering the advertisement for the property listing to a user in the case that R is a non-zero value.

29 more elements

Dependencies are inferred from the claim text.