Online Audience Building Initiatives

Introduction

This series of case studies showcases both operational excellence and initiatives that have successfully increase audience for newspaper Web sites.

Introduction

About the Author

Click Here for Other Articles in this Series

Audience Building Initiative: ‘DataUniverse’ at the Asbury Park Press

By Rich Gordon

Summary

The Asbury Park Press has established a section of its Web site called “DataUniverse,” which consists of searchable databases of public records, such as public employee salaries, criminal convictions, crime reports and property assessments. In five months since DataUniverse went live in December 2006, it generated more than 25 million page views. The paper’s online traffic is growing rapidly, and its executive editor believes DataUniverse is the most significant driver of this audience growth.

History of this initiative

In the fall of 2006, Executive Editor William “Skip” Hidlay attended a meeting of Gannett Corp. editors at which attendees discussed the company’s new “Local Information Center” initiative. Local data is one of the seven “primary jobs” emphasized in the information center approach. At most papers the focus is on data such as local event calendars.

Hidlay returned from the meeting and met with Paul D’Ambrosio, the paper’s longtime investigations editor. D’Ambrosio, a skilled practitioner of computer-assisted reporting, had led efforts to acquire government databases – such as criminal convictions and property assessments – for use by the paper’s reporters. As early as 2001, he had experimented with publishing real estate transaction data to the Web. Hidlay asked D’Ambrosio if public records databases could be made available to the public on the paper’s Web site. D’Ambrosio suggested creating a site that aggregated public records databases.

“Paul just took the ball and ran with it,” Hidlay said. “He learned the programming and put the data up.”

D’Ambrosio said he was familiar with SQL, the most popular query language for databases, from his work in computer-assisted reporting. For PHP, a friend coached him through the process of using DreamWeaver page-editing software to create the proper coding. Everything else he needed to know, he learned from books.

The DataUniverse site launched in December 2006 with three databases: property sales records, property ownership records and state employee salaries and job titles.

• As of early May, DataUniverse allows people to search these databases:
• Crimes and crime rates for New Jersey counties and towns, searchable by county or town, year, crime rate and more
• State standardized test results and SAT scores for New Jersey schools, searchable by county, school district, school name and more
• Names of New Jersey residents who die, from the Social Security Administration – a popular database for genealogy research
• New Jersey public school teachers and administrators, searchable by name, school district, job title and more

DataUniverse also allows people to search campaign contributions for state elections, budgets and tax rates for taxing bodies for fire protection, criminal convictions in state courts and more. All categories have multiple search fields.

DataUniverse also includes links to other sites that allow people, for instance, to search for disciplinary proceedings against physicians, convicted sex offenders, consumer complaints, state contracts and registered charities.

What the data show

In the five months from December 2006 to April 2007, DataUniverse has generated more than 25 million page views, D’Ambrosio said. That’s an average of more than 5 million page views per month – more than a quarter of the Web site’s total.

The DataUniverse pageview data:

• December 2006: 1.5 million
• January 2007: 2.1 million
• February 2007: 7.3 million
• March 2007: 9.1 million
• April 2007: 5.4 million

“I’ll be honest with you. I did not expect it would generate a huge amount of traffic,” Hidlay said. “What this really shows is that the public really wants access to the raw data. We have been amazed at what a traffic builder and audience builder this is.”

When a new database is published on DataUniverse, D’Ambrosio said, users don’t look at just that one. “They then start exploring all the databases – all the other databases explode with traffic.” Traffic also increases to the rest of the paper’s Web site, D’Ambrosio said.

Michael Maness, vice president for strategic planning for Gannett’s newspaper division, said data distribution connects nicely to the public service mission of the newspaper, and that the Press has shown that it also can generate community conversations.

“The ability for people in New Jersey to go and get access to public information has been completely enabled by the newspaper,” Maness said. “It moves [the newspaper] back to the center of community conversation, which is where we need to be.”

The Press has done many other things to build online usage: increasing the volume of breaking news, enabling and actively managing user comments, and increasing multimedia storytelling. From April 2006 to April 2007, the site’s unique visitors rose 77 percent and page views 146 percent.

“We have aggressive breaking news, vibrant forums and a lot of multimedia,” Hidlay said. “But the data is the key component, and I would never have predicted that.”

How it works

The Press had been publishing some data to the Web for years – for instance, school test score data. “The new concept was to consolidate this into one site and market it as one site. What I wanted to do was create a one-stop shop for Web users to go for public information – how do I look up a doctor, how do I look up crime records?” D’Ambrosio said. “Rather than bounce around the Web or the Web site, they remember DataUniverse and go there.”

Generally, data releases follow the same pattern, D’Ambrosio said. First, a “soft launch” – the data set is published, and D’Ambrosio announces it on the DataUniverse site. Then, on a Sunday, the Press publishes an article saying the data is available. Sometimes, the data release coincides with an investigative report based on the database. For instance, the paper cross-matched state employees and criminal convictions and found 1,800 convicted felons holding jobs in state and local government. If there’s no revelation to be published, the paper just publishes a basic article saying that the data is now available on the Web site.

“The traffic will explode on Sunday and Monday,” D’Ambrosio said. “This shows there is a tremendous amount of connection between the newspaper and the Web site. We’re drawing people into the Web site who don’t often go to the Web site for information.”

Over three weeks in March, the paper released three significant databases – names and salaries for police, public employees other than police and teachers, and Rutgers University employees. Driven by interest in this data, the Press had record Web site traffic that month – 21.3 million page views and 1.2 million unique visitors.

Part of the reason for the intense interest in public employee salaries is that New Jersey has a history of allowing people to hold down multiple taxpayer-financed jobs at the same time. “We found public servants making more than the governor because they had multiple part-time jobs,” D’Ambrosio said. The Press has written articles exposing these abuses.

“People love getting their hands on what journalists do and come up with their own stories,” D’Ambrosio said. “People actually email me with their story ideas: ‘Look at this individual – this person has three or four different jobs.’ ”

“We have really educated our audience,” Hidlay said. “The audience was really hungry for this data and they knew how to use it.”

Promotion and connections to print

The paper has not done any promotion of DataUniverse outside the newspaper. It has run a few house ads. But the main promotion for DataUniverse has been the newspaper articles about the data releases.

“We’re not just putting databases out there,” D’Ambrosio said. “We’re doing stories around them.”

There has been some negative feedback, especially from public employees who don’t think their salaries should be published online and from people who thought publishing the names of dead people and their Social Security numbers would increase the risk of fraud. D’Ambrosio has posted his perspective on these issues on DataUniverse discussion forums.

“We just put up information that anyone can get by filing a public records request,” D’Ambrosio said. As for the Social Security death records, he said, the federal government encourages the dissemination of this information so companies can check it as a means of preventing identity theft.

About the technology

For DataUniverse, D’Ambrosio had to learn more about how to deliver databases using the so-called “LAMP architecture” – the Linux operating system, Apache Web server, mySQL database and PHP scripting language. All are open-source products, so DataUniverse requires no software expenses. He’s done all the programming himself.

“I had used mostly Access and FoxPro [desktop database managers] in my career,” D’Ambrosio said. “The mySQL transition wasn’t very hard. But using PHP for page design and making it all work together on the server was kind of an interesting task. That was the hurdle to jump. There were a lot of sleepless nights trying to get this to work the first time.”

About revenue

Neither Hidlay nor D’Ambrosio would release specific revenue figures for DataUniverse. But every search page and results page has at least one ad banner. Even at a relatively low rate of $10 per 1,000 impressions, DataUniverse would have generated $250,000 in ad revenue from December to April.

What’s next

D’Ambrosio is working on improving the site – increasing the speed of searches, updating databases more frequently. The next phase will incorporate “bigger data sets and a bigger presence,” he said.

Lessons from the Press’s experience

• Recognize the links connecting computer-assisted reporting, database acquisition for reporting purposes and data delivery to online users. Most newspapers have at least one journalist comfortable with computer-assisted reporting techniques, and many routinely acquire useful data sources for newsroom use. The Press’ experience suggests that the public, too, appreciates having access to government data.
• Develop your computer-assisted reporting expertise. DataUniverse was possible in large part because of D’Ambrosio’s vast experience in obtaining, analyzing and finding valuable journalistic uses for government data.
• Consider database acquisition part of the reporting mission of your newspaper. The distribution of relevant local data is one of seven pillars of Gannett’s Local Information Center intiative, and the Press’s experience suggests this strategy can work. Maness said the new data focus is leading many Gannett papers to hire staff, often with library science backgrounds, to focus on data acquisition and delivery.
• Develop LAMP expertise. Linux, Apache, mySQL and PHP are the most popular open-source technologies for database development and delivery. They also come at the right price – free.
• Publish data on government employee salaries. This has been the most popular data on DataUniverse, D’Ambrosio said.
• Connect data releases to articles in print. When articles are published in the Sunday paper about new data releases, “you just sit back and watch the needle go into the red zone,” Hidlay said.
• Give users a variety of search options. One popular feature enables users to type in the name of a municipality and see salaries for all of its employees.

Relevant links

• Asbury Park Press DataUniverse
• Memo from Gannett CEO Craig Dubow describing the Local Information Center initiative
• Presentation about the Local Information Center initiative by Gannett vice president Michael Maness (Inland Press Association)

About the author