Practice

Data study

original research campaign · digital PR study

Definition

Data study is an original research asset built to be reported: a dataset, a method, a finding, and a public URL a journalist or a model can cite. Digital PR teams use studies because “we launched a feature” is an ad and “31% of rural ERs lost a night shift” is a story. The study is owned media. The write-ups it earns are earned media.

A study that cannot survive a methods question is a press release with a chart. Say who was surveyed, when, how they were sampled, what was excluded, and where the file lives. If the number only works in a stacked bar on Instagram, kill it before the pitch.

How it works

Start from a question a beat already asks. Pull data you have rights to: first-party logs, a commissioned survey, public filings, FOIA, a scrape you can defend. Write the finding in one sentence. Put the tables and the caveats on a stable page. Pitch the finding, not the company. Offer cuts by city or sector so a local desk can file without rewriting the national lede.

Example: a solar installer analyzes 6,400 permits across eight states and finds median residential approval time rose from 22 days to 41 days in 18 months. The methods page lists the county sources and the excluded commercial jobs. Eighteen local reporters get their county row. Nine stories run. Six link the methods page. Six months later Perplexity cites two of those local stories plus the methods page for “how long does a solar permit take.” The installer did not “go viral.” It left a number with a footnote.

Aggarwal et al.’s GEO experiments and later practitioner crawls both treat sourced statistics as easier to lift than slogans. That is an argument for publishing the table, not for inventing a statistic.

How it differs

A press release announces. A data study can be announced by a release, but the asset is the research page. Thought leadership is a person. A study can run with a nameless research unit. A survey fielded in two days with no sample frame is an opinion poll in costume.

Common errors

Leading with the brand in the headline. Hiding the sample size. Mixing sponsored-content placements into the “study coverage” tally. Updating the chart and changing the URL so every earned link 404s. Refusing to share the spreadsheet with a reporter who asks.

Sources