The Who-To-Follow System at Twitter: Strategy, Algorithms, and Revenue Impactby Ashish Goel, Pankaj Gupta, John Sirois, Dong Wang, Aneesh Sharma, Siva Gurumurthy

Interfaces

Similar

Quality costs

Authors:
National Council for Quality and The, Reliability
1971

Finding the Twitter users who stood with Wendy

Authors:
Amanda Jean Stevenson
2014

Twitter based TV rating system

Authors:
A. D'Souza, R. Bathla, N. Giri
2013

Nomenclature for Factors of the HLA System, 1987

Authors:
Who Nomenclature Committee Who
1988

Text

This article was downloaded by: [132.206.27.24] On: 09 February 2015, At: 12:31

Publisher: Institute for Operations Research and the Management Sciences (INFORMS)

INFORMS is located in Maryland, USA

Interfaces

Publication details, including instructions for authors and subscription information: http://pubsonline.informs.org

The Who-To-Follow System at Twitter: Strategy,

Algorithms, and Revenue Impact

Ashish Goel, Pankaj Gupta, John Sirois, Dong Wang, Aneesh Sharma, Siva Gurumurthy

To cite this article:

Ashish Goel, Pankaj Gupta, John Sirois, Dong Wang, Aneesh Sharma, Siva Gurumurthy (2015) The Who-To-Follow System at

Twitter: Strategy, Algorithms, and Revenue Impact. Interfaces 45(1):98-107. http://dx.doi.org/10.1287/inte.2014.0784

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial use or systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisher approval, unless otherwise noted. For more information, contact permissions@informs.org.

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitness for a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, or inclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, or support of claims made of that product, publication, or service.

Copyright © 2015, INFORMS

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, management science, and analytics.

For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org

Vol. 45, No. 1, January–February 2015, pp. 98–107

ISSN 0092-2102 (print) — ISSN 1526-551X (online) http://dx.doi.org/10.1287/inte.2014.0784 © 2015 INFORMS

THE FRANZ EDELMAN AWARD

Achievement in Operations Research

The Who-To-Follow System at Twitter:

Strategy, Algorithms, and Revenue Impact

Ashish Goel

Stanford University, Stanford, California 94035, ashishg@stanford.edu

Pankaj Gupta, John Sirois, Dong Wang, Aneesh Sharma, Siva Gurumurthy

Twitter Inc., San Francisco, California 94103 {pankaj@yogins.com, john.sirois@gmail.com, dwoanngg@gmail.com, aneesh@twitter.com, siva@twitter.com}

The who-to-follow system at Twitter is an algorithmic data product that recommends accounts for Twitter users to follow. Building the system involved algorithmic, analytics, operational, and experimental challenges; operations research and analytics techniques played a key role in resolving these challenges. This product has had significant direct impact on Twitter’s growth and the quality of its user engagement, and has also been a major driver of revenue. More than one-eighth of all new connections on the Twitter network are a direct result of this system, and a substantial majority of Twitter’s revenue comes from its promoted products, for which this system was a foundation. To place this contribution into perspective, Twitter is now a publicly traded company with a market capitalization of more than $30 billion, projected annual revenue of close to $1 billion, and more than 240 million active users.

Keywords : computational analysis; optimization; data mining; Internet.

Twitter is a communication and information net-work that is remarkably simple in concept: A user follows other users to subscribe to their tweets—140character messages that can contain embedded media or links, and may be received on a variety of clients, including the twitter.com website, and mobile clients on iOS or Android devices. The vibrancy of the service, whether in informing users of relevant breaking news or connecting them to communities of interest, derives from its users—more than 240 million of them. As of late 2013, users collectively posted more than 500 million tweets per day. Therefore, maintaining and expanding the active user population is a major priority for Twitter.

One important way to sustain and grow Twitter is to help users, both existing and new, discover connections. This is the goal of Twitter’s user-recommendation service, which we call our who-to-follow product.

In the current interface (see Figure 1), the who-tofollow box is prominently featured in the left rail of the Web client as well as in many other contexts across multiple platforms. This box suggests additional Twitter accounts that the user may wish to follow. The recommendations are highly personalized to the user. They also heavily leverage existing connections between users; the recommendations are based not only on users’ interests, but also on the existing connections between users on the Twitter network.

This general concept, called a collaborative filter, is particularly suitable to Twitter, because users often spend considerable effort in deciding which other users to follow. Note that the first of the recommendations in Figure 1 is a promoted account. A promoted account is a form of advertising; if a user follows a promoted account, the account’s owner pays Twitter a small amount, which an online auction determines.

Building the who-to-follow system involved challenges, including determining the best collaborative filtering algorithms, analyzing the large volume of data (each of Twitter’s more than 240 million active users has a set of separate connections), efficiently implementing and fine-tuning the system, improving 98

D ow nl oa de d fro m in fo rm s.o rg b y [1 32 .20 6.2 7.2 4] on 09

Fe br ua ry 20 15 , a t 1 2:3 1 .

Fo r p ers on al us e o nly , a ll r igh ts res erv ed .

Goel et al.: The Who-To-Follow System at Twitter

Interfaces 45(1), pp. 98–107, © 2015 INFORMS 99

Figure 1: (Color online) In Twitter’s who-to-follow product, recommendations appear in the bottom left section of the figure. The first recommendation is a promoted account. Twitter is paid a specified amount, determined by an online real-time auction, for engagements with a promoted product. Note that promoted accounts appear in the same format and place as accounts suggested by the who-to-follow recommendation engine. it based on constant feedback, and doing operational planning.