A practical public data request guide for online newsrooms

A press release may not tell a local newsroom how public facilities are used or where a program's spending went. But asking for everything an agency holds can produce a long exchange and files that still do not answer the reporting question. Requesting public data is an editorial process: connect a question to identifiable records, track what arrives, and decide what the evidence supports.

This guide covers that workflow rather than legal entitlements. A formal public-records request and an ordinary request for reporting assistance are not interchangeable. Check the relevant agency's current official guidance for the applicable process, deadlines, fees, and review options. Obtain qualified advice when a decision requires legal interpretation.

Separate the reporting question from the record you need

“Is this facility being managed well?” is a reporting question, not the name of a record. Opening hours, closure logs, attendance summaries, and maintenance records may each illuminate a different part of it. List the claim you want to examine alongside the existing records that could support or challenge it.

Distinguish a request for records from a request for a new analysis. Asking why satisfaction fell belongs in an interview; requesting the existing monthly survey results belongs in the records workflow. Keep interpretive questions in a follow-up list rather than hiding them inside a vague demand for data.

Begin with the smallest useful scope. Before submitting, describe the comparison you would make if the requested material arrived. If that is difficult, narrow or clarify the question before creating more work for everyone.

Check existing publications and identify the record holder

Briefly inspect the agency's website, published reports, and available data listings. If the material is already available, save its location, download date, and reference period. Public availability does not establish that a file is current or complete.

The office receiving inquiries may not be the office that maintains the records. When ownership is unclear, ask about the record's name and responsible department. A phone conversation can help, but keep the final scope and submission route in writing so another reporter can understand what happened.

For comparisons across agencies, check how each organization defines the requested activity. Identical wording can retrieve incompatible records when local classification systems differ. Resolve that uncertainty before treating the responses as a shared dataset.

Specify dates, units, and boundaries

Replace “recent usage figures” with a defined start date, end date, facility scope, and reporting interval. Clarify whether the measure counts visits or distinct visitors, includes cancellations, and represents provisional or finalized figures. These distinctions affect the story, not just the spreadsheet.

A facility that opened halfway through the period may not be comparable with one operating throughout it. Consider requesting identifiers and operating periods alongside totals. Avoid asking for names, contact details, or individual activity histories when aggregate records would answer the question.

If the scope changes during discussions, record the old and new versions and the date of agreement. Mark what the narrower request no longer covers. Otherwise, a successfully completed request can quietly become an unsupported claim that the entire original question has been answered.

Describe the structure as well as the preferred format

For calculations, explain that a machine-readable format such as CSV or a spreadsheet would be useful. Ask what formats actually exist and can be supplied. An unavailable preferred format does not mean the underlying records do not exist.

Request available field definitions, measurement units, code explanations, and missing-value conventions. A zero, an empty cell, a withheld value, and a field that was never collected may mean different things. If the response includes several files, establish the period and scope of each one.

When records arrive as scans, compare any extracted text or tables with the originals. Keep converted working files separately. Check totals and dates after conversion rather than assuming that optical character recognition preserved the meaning of the document.

Track the next action, not just the submission

A shared request register can begin with a request ID, agency, scope summary, submission time, acknowledgment, assigned reporter, next check date, and link to the original request. Keep an agency's stated response date separate from the newsroom's internal follow-up date.

Useful statuses include draft, submitted, acknowledged, scope discussion, response received, clarification pending, and closed. Replace “in progress” with an actionable note such as “confirm the holding department before resubmitting the revised scope.” Someone covering an absence should be able to act without reconstructing the entire email thread.

Do not invent a deadline for a formal process. Record the basis in the acknowledgment and applicable official guidance. Preserve later changes as additions to the history rather than overwriting the original dates without explanation.

Distinguish incomplete responses from other outcomes

Receiving an attachment does not mean the request is complete. Map each requested item to supplied, partly supplied, not supplied, or awaiting explanation. Note how missing periods or the provision of totals instead of detailed records limits the planned reporting.

A response saying that records are not held differs from a decision not to disclose a field or a request for clarification. Do not collapse all three into “the agency refused to release the data.” Read and accurately describe the explanation actually provided, then seek clarification where needed.

An editor can weigh the public interest, available reporting time, and alternative sources before pursuing further steps. Any formal review or appeal must be assessed under the applicable process. An internal tracking sheet is not a substitute for checking eligibility, requirements, or deadlines.

Validate definitions before drawing conclusions

The first task after receipt is to understand the fields, not rank the organizations. A spending figure can change meaning with its accounting period or included categories. Compare totals with detail rows, look for duplicate entries, and check whether classifications changed during the period.

Treat an unusual value as a question rather than immediate proof of an error or wrongdoing. Send the relevant record location, what seems inconsistent, and a specific clarification question. If a corrected file arrives, retain the receipt date and explanation and identify which version supports the published analysis.

Records may establish a difference without establishing its cause. Use interviews and on-the-ground reporting to test possible explanations. Do not let a clean spreadsheet create more confidence than the underlying evidence warrants.

Explain the evidence and its limits to readers

Identify the providing agency, covered period, reference date, and important definitions in language readers can understand. Describe missing material and its effect on the comparison. If the definitions cannot be reconciled, avoid forcing the figures into a single league table.

When publication must proceed before all responses arrive, separate confirmed information from material still outstanding. Silence is not itself evidence of the suspected problem. Keep an accurate record of when questions were sent and what they covered, and decide what new evidence would trigger an update or follow-up story.

Review privacy, permitted use, and the appropriate publication scope before sharing source files with readers. Receiving a document during reporting does not automatically establish that its entire contents should be uploaded publicly. A carefully explained summary may be more appropriate than an unreviewed attachment.

Pilot the workflow with one focused request

For a hypothetical facility-use story, a draft could specify the facilities, period, monthly visit counts, unit definitions, and missing-value conventions. An editor should test whether those items answer the reporting question before submission. This is a planning example, not an official form suitable for every agency or jurisdiction.

After closing the request, review why follow-up was necessary. Perhaps the period was ambiguous, the wrong office was contacted, or field definitions were checked too late. Record the outcome even when the records could not be obtained, so the next reporter does not repeat the same request without learning from it.

The goal is not to collect the largest archive. It is to request relevant records, understand what was supplied, and be candid about what remains unknown. Linking the original request, next action, response scope, and published source version gives a small newsroom a practical chain from question to evidence.