Skip to content
TablePage.ai Open the app

A Reference Sheet Makes Your Dataset Usable

Build a dataset reference sheet with column definitions, units, missing-value rules and provenance, then keep it accessible alongside your public table.

Share X in f
Wei Hu

Before sharing a spreadsheet, write down what a reader needs to interpret it: what each row represents, what the columns mean, which units apply, and why some cells are empty. Put those explanations in a reference sheet that travels with the data.

Here, the reference sheet is a compact data dictionary, not a formula cheat sheet. USGS describes data dictionaries as documentation that communicates data structure and meaning, including field definitions, data types, reference codes and missing-data rules. USGS guidance also recommends updating the dictionary when the structure changes.

The result should let someone use your published table without asking you to decode it.

Start with one row per field

Create a working tab named Reference. Use one row for each column in the dataset, copying the column names exactly. A renamed header should trigger a matching change in the reference sheet.

Here is an illustrative reference sheet for fictional monthly service-request data. It is a layout example, not an official dataset.

Field Meaning Type or format Unit Allowed values and missing rule
district_code District responsible for the closed requests Text None Must match the district lookup table; no blanks
month Calendar month in which requests were closed Text, YYYY-MM Month No blanks
requests_closed Number of requests closed in that district and month Integer Requests Zero or greater; blank means unavailable, not zero
median_close_days Median elapsed time from opening to closure among closed requests with valid timestamps Decimal Days Zero or greater; blank means no valid timing observations
release_status Whether the row is provisional or final Text None provisional or final; no blanks

This structure separates a definition from a storage type. “Decimal” tells a reader how a value is represented; “days from opening to closure” tells them what it measures.

If the dataset uses category codes, add a separate lookup table with one row per code and its label. Do not leave a rule such as “valid district code” unexplained.

Add the context that column definitions cannot carry

Keep a short dataset note alongside the field table. For the example above, it would explain:

  • Row meaning: One district in one calendar month. The district–month pair must be unique.
  • Coverage: Closed requests only. Open requests are excluded, so this is not a count of all requests submitted.
  • Calculation: The median uses records with valid opening and closing timestamps. State how elapsed days are calculated and how invalid timestamps are handled.
  • Provenance: Identify the original source, link to it, and describe aggregation or other transformations.
  • Release: Record the dataset version, publication date, coverage period and any revision policy.
  • Reuse: State the applicable license or usage restrictions rather than assuming public availability permits unrestricted reuse.

These are recommended documentation fields, not a substitute for the original methodology. If a source definition is unclear, retain that uncertainty instead of silently choosing an interpretation.

Check the sheet against the file you will publish

Review the final export, not just the working workbook:

  1. Every exported column has a reference entry, and every entry names a column that exists.
  2. Units and scales are explicit—for example, whether a percentage is stored as 25 or 0.25.
  3. Zero, blank, unknown and suppressed values have distinct, documented meanings where applicable.
  4. Codes match their lookup tables, and the declared row key has no duplicates.
  5. Definitions still match the calculations and filters used to produce the release.

Keep meaning in text, not only in cell colors or formatting. Microsoft notes that text-format exports remove formatting and that CSV saves only the active sheet. A Reference tab therefore will not automatically accompany an Excel CSV export. Export it separately or publish its contents as accompanying documentation. Microsoft’s export guidance

Keep the reference accessible beside the public table

Publish only data and documentation cleared for public release. Check the reference sheet too: internal notes can contain sensitive information even when the data table does not.

As of October 4, 2026, TablePage lists support for CSV, TSV, XLSX and XLS uploads and public dataset pages with filterable tables. For a straightforward publishing workflow, prepare a clean data file and a separate reference file; do not assume a workbook’s documentation tab will appear beside the published table.

Wherever you share the dataset link—an article, project page or release announcement—also link to its reference documentation. Give both the same release identifier so readers can tell which definitions apply. For the surrounding explanation, use a table caption and summary to introduce the dataset, leaving detailed field rules in the reference sheet.