Tabular Results: Publish Numbers Readers Can Check

To publish tabular results, give readers more than a column of final scores or percentages. Include what each row represents, the inputs behind calculated values, and enough context to compare like with like.

Tabular results are the outputs of an analysis arranged in rows and columns. A useful public results page lets readers identify a group, inspect its result, and understand how that result was calculated—without opening your working spreadsheet.

Start with one clearly defined result per row

Suppose you want to publish the percentage of service requests completed within a target time at each branch. Define a row as one branch in one reporting month, then include both the within-target count and the eligible count.

This fictional example uses invented counts, not actual branch performance:

branch	period	within_target	eligible_requests	within_target_pct	status
North	2026-09	80	100	80.0	reported
South	2026-09	150	200	75.0	reported
West	2026-09				not_reported

The expected result is a table where readers can compare reported branches and distinguish a missing report from zero performance.

Use one header row and keep titles, methodology paragraphs, and decorative spacing outside the data rectangle. If your input is a presentation-style worksheet with merged headings and subtotals, first reshape the Excel report into tabular data.

Keep the calculation visible

For this example:

within_target_pct = 100 × within_target / eligible_requests

Store 80.0 in the percentage column and document that its scale is 0–100. Do not leave readers guessing whether 0.8 means 0.8% or 80%.

Keeping the counts exposes an important distinction:

• Overall request-level rate: 100 × (80 + 150) / (100 + 200) = 76.7%, rounded to one decimal place.
• Unweighted mean of branch rates: (80 + 75) / 2 = 77.5%.

These answer different questions. Use the first to describe the share of eligible requests completed within target across the reporting branches. Use the second only when you intend to give each reporting branch equal weight. Neither includes West, whose counts are unavailable.

Keep any overall total separate from branch records so readers do not accidentally include it again when summing or filtering.

Define blanks, units, and coverage

A header does not carry the whole methodology. CSV itself has no built-in mechanism for declaring column types or uniqueness constraints; the W3C CSV on the Web primer explains how accompanying metadata can document and validate a table.

For this example, publish a short note alongside the dataset:

Illustrative service-request results, September 2026. Each row represents one branch. Eligible requests are requests whose target deadline falls within the reporting month; the numerator counts those completed by that deadline. Percentages use a 0–100 scale and are rounded to one decimal place. Blank numeric cells with not_reported mean no counts were supplied, not zero requests or zero performance. All values are fictional.

For real results, also identify the source, exclusions, extraction date, release version, and reuse terms. If the denominator is zero, leave the rate undefined and distinguish that status from an unreported denominator.

Check the publication file, then the public page

Before uploading:

• Check that each branch–period combination occurs only once.
• Confirm reported counts are nonnegative integers and within_target never exceeds eligible_requests.
• Recalculate rates where the denominator is positive and check the rounding.
• Confirm that blank cells retain their documented meaning.
• Remove sensitive information and assess whether the remaining data could identify individuals, including through links to other public information. The UK Information Commissioner’s anonymisation guidance explains why removing direct identifiers alone is insufficient.

Keep your working workbook separately. When exporting from Excel to CSV, only the current worksheet is saved, and some spreadsheet features are unsupported. Excel also interprets CSV columns using its default format settings, which can affect dates and leading-zero identifiers. Microsoft’s import and export guidance describes these limitations. Inspect the exported file and re-import it with deliberate column types before publication.

As of October 2026, TablePage accepts CSV, TSV, XLSX, and XLS uploads and generates a public dataset page with a shareable link and filterable table. Upload only the reviewed publication file—not a workbook containing private supporting sheets.

Open the resulting link as a reader. Compare headers, counts, percentages, and missing values with the approved file; test a branch filter; and make sure the methodology is available on the accompanying webpage or in a linked document. Share the page only when its numbers and their meaning agree.