We help you understand audience attention. Follow me: @amontalenti - - PowerPoint PPT Presentation

Website: parse.ly Blog: blog.parse.ly Email: andrew@parsely.com

We help you understand audience attention.

Follow me: @amontalenti Our research: @parsely Our podcast: @attnpod

How? Parse.ly Analytics.

Web content visits represent attention at global scale.

+ hundreds of other companies who run thousands of high-traffic sites. + the long tail.

Sites with content and audience Platforms

Parse.ly measures content and audience …

Page views Visitors Engaged time Social shares Audience loyalty Devices Video Titles Authors Sections Tags Referrers Campaigns Publish dates Channels

+

Much more

… to tell the story behind the story.

Our dashboard can answer this question: What’s gaining attention on your sites and apps?

Provide a real-time and historical window into what’s happening with your content when it comes to audience attention.

• 30,000 monthly active users across

350+ media companies.

• Measures the attention of over 2 million

page views per minute at peak time.

• Sub-second data latency with 99.99%

internal SLO.

We make data accessible and essential.

Parse.ly Analytics: 
 What’s running under the hood?

Powered by mage:

• 100+ Elasticsearch nodes storing
• ver 20 terabytes of production live

query data.

• 3,600+ real-time processing CPU

cores using Storm.

• Kafka and Cassandra for rock-solid

distributed streaming data.

• Elastic scalability for hourly and

nightly jobs using Spark.

Parse.ly Analytics: 
 What does the team release publicly?

We love open source!

• streamparse is our publicly-maintained

and popular project for running production parallel computation systems with Python 2.x and 3.x, using Apache Storm.

• PyKafka is the community’s fastest and

most production-tested Python driver for Apache Kafka.

+ PyKafka + parsely_raw_data + time-engaged + others

Why now? Parse.ly Currents.

Aggregate attention data already guides the industry.

And answers questions it could never answer on its own.

Our network data can answer this question: What do people care about?

Front row seat to the web interests

• f over 1 billion people per month

and 150 million people per day. Categories include: news, entertainment, finance, politics, sports, opinion, culture, and more. Apply modern machine learning and natural language processing techniques.

Parse.ly Currents: 
 What is our petabyte-scale analysis stack?

Petabytes of event data and terabytes of web crawl data.

• BigQuery used with day-partitioned tables to do

fast aggregation over petabyte-scale event data without running a cluster.

• PyData stack used for statistics and machine

learning over time series data.

• Natural language processing on text data using

Python, leveraging a web-based ontology (knowledge graph), domain-specific keyword/entity lists, word vectors, document classifiers, unsupervised clustering, and more.

1 billion unique visitors per month 20 billion page views per month 5 billion clicks from search, social, & others 900k posts published and analyzed each day 2 million topics, categories, and keywords

Does discovery vary by topic?

87.1% Facebook 61.4% 60.8% 59.5% 58.9% 53.5% 52.7% 41.3% 36.3% 35.5% 21.3% 19.2% 14.1% 11.9% 3.7% 84.4% 39.0% 14.1% 30.4% 50.4% 18.0% 60.8% 22.3% 42.2% 20.7% 43.0% 28.9% 29.7% 22.6% 24.6% 22.2% 24.4% 19.8% 21.3% 15.9% 24.6% 10.1% 29.1% 12.3% 26.2% 6.2% 6.7%

Google

Job Postings Business & Finance Sports Technology State & Local Politics World Economy National Security Local Crime & Incidents Criminal Justice Education & Research U.S. Presidential Politics Entertainment Local Events Lifestyle 2.7k posts 39k posts 210k posts 67k posts 17k posts 26k posts 49k posts 98k posts 55k posts 36k posts 110k posts 190k posts 96k posts 110k posts

Topics are derived from posts in the Parse.ly network of sites from 2016 using a topic modeling algorithm called LDA (Latent Dirlichet Allocation). 
 For more information: parsely.com/authority

Number of posts for each topic

110k

posts U.S. Pres. Politics

43% 47% 10%

Desktop Mobile Tablet

Device tra ic breakdown Number of posts for each topic

26k

posts World Economy

46% 45% 9%

Desktop Mobile Tablet

Device tra ic breakdown

CLINTON

PRESIDENT CAMPAIGN DONALD

PRESIDENTIAL OBAMA ELECTION PARTY HILLARY

STATE POLITICAL DEMOCRATIC WHITE CANDIDATE VOTE SANDERS HOUSE

C O M M O N W O R D S I N P O S T S

TRUMP

U.S. Presidential Politics

REPUBLICAN

CHINA

OIL

EU PERCENT

CHINESE ENERGY SINCE PER EUROPEAN TRADE

C O M M O N W O R D S I N P O S T S

STOCKS BREXIT PRICES DEAL BANK CENT NFL AP UK

World Economy

ACCORDING MARKETS TRADING BILLION BRITAIN MARKET STOCK WORLD GLOBAL POWER

Google Search Facebook Other

43.0% 36.3% 20.7%

External referral sources

4.6% 4.0% 2.4% 1.4% 1.1% 0.9% 0.9% 0.8% 0.7% news.google.com twitter.com yahoo! drudgereport.com flipboard.com bing linkedin.com reddit.com tra ic.outbrain.com Facebook Google Search Other

59.5% 24.6% 15.9%

External referral sources

4.3% 4.1% 1.9% 1.1% 0.9% 0.7% news.google.com twitter.com drudgereport.com yahoo! bing reddit.com

Can Internet attention predict public opinion?

Can Internet attention predict a film’s revenue?

600k 500k 400k 300k 200k 100k 10k 20k 30k 40k 50k 60k 70k Cumulative Box Ofice Gross Revenue

Print Ad Cost in US $

600k 500k 400k 300k 200k 100k Cumulative Box Ofice Gross Revenue

Negative Cost in US $

50k 100k 150k 200k 250k 200k 600k 500k 400k 300k 200k 100k 400k 600k 800k 1M Cumulative Box Ofice Gross Revenue

Unique Views

0.955

Pearson Correlation Coeficient when excluding PG rated movies

Movies rated PG Movies not rated PG

0.474

Pearson Correlation Coeficient when excluding PG rated movies

0.829

Pearson Correlation Coeficient when excluding PG rated movies

Revenue Compared to Unique Views

for Related Web Posts 3 Days Prior to Release

Revenue Compared to Print Ad Cost in US $ Revenue Compared to Production Cost in US $

Total unique views for posts related to a movie three days prior to its release has the highest correlation with revenue compared to production cost and advertising budget.

200k 600k 500k 400k 300k 200k 100k 400k 600k 800k 1M

Cumulative Box Ofice Gross Revenue Unique Views

0.955

Pearson Correlation Coeficient when excluding PG rated movies Movies rated PG Movies not rated PG

Revenue Compared to Unique Views

for Related Web Posts 3 Days Prior to Release

We are a partner you can trust. 400+ paying clients. 3000+ big sites. 1B+ network visitors.

We’re small and nimble, yet we

• perate with scale and integrity. We

are 70+ people.

• A client services, support, and
• ps team of 40 people, with a

head office in NYC.

• A fully distributed product team
• f engineers, data scientists,

and designers. 30 people across US, Canada, and Europe.

• $12M+ USD in financing raised

from 2011 to 2017.

Three asks for the audience today.

Sign up free, give us feedback!
 
 http://parse.ly/currents

Follow me on Twitter! @amontalenti

Website: parse.ly Blog: blog.parse.ly Email: andrew@parsely.com

Let’s continue the conversation about internet attention.

Follow me: @amontalenti Our research: @parsely Our podcast: @attnpod