Tabulation of Data: From Individual Records to a Public Summary

Suppose you have a spreadsheet of service requests and want readers to compare outcomes by branch. Publishing the individual records is not always necessary: a checked summary can answer the question more directly.

Tabulation of data means arranging data systematically in rows and columns, often grouping observations into categories and calculating counts, totals or percentages. The important decisions are what each row represents, what you count and which denominator you use.

Here is a worked example that turns eight records into four publishable summary rows.

Choose the kind of tabulation

• Simple tabulation summarizes one characteristic—for example, the number of requests with each status.
• Cross-tabulation summarizes combinations of characteristics—for example, status within each branch.
• Grouped frequency tabulation counts numerical observations in defined intervals, such as whole-day response times of 0–9 days and 10–19 days. Define intervals so that they do not overlap or leave gaps between them.

A frequency is a count. A percentage expresses that count relative to a stated total. Neither is automatically an average, and a count of records is not necessarily a count of people.

1. Define the input and the question

Illustrative dataset—not real service records: each row below represents one request at the same reporting cutoff. Each request has one branch and one status.

request_id	branch	status
101	North	Closed
102	North	Closed
103	North	Closed
104	North	Open
105	South	Closed
106	South	Open
107	South	Open
108	South	Open

The question is: What share of requests is closed within each branch?

Before counting, check that request IDs are unique, category labels are consistent and the reporting period is defined. If the source instead contains one row per status update, first reduce it to one record per request at the cutoff. Otherwise, you would count updates rather than requests.

Decide how to handle missing branch or status values. Do not silently discard them: either include an explicit “Unknown” category or document the exclusions and their count.

2. Count the categories

A simple tabulation of this sample gives four closed requests and four open requests. Cross-tabulating by branch reveals the difference hidden by that overall split.

In Google Sheets on a computer:

1. Select the source range, including its header row.
2. Choose Insert → Pivot table.
3. Add branch to Rows and status to Columns.
4. Add request_id to Values, and set its summarization to COUNT, not SUM, for these numeric IDs.

Google’s pivot-table guidance documents adding row, column and value fields and requires a header for each source column. COUNT counts numeric values only, so do not use it unchanged for IDs stored as text.

The expected counts are North: three closed, one open; South: one closed, three open. Check these against the eight input records before calculating percentages.

3. State the denominator

For this question, use:

Within-branch percentage = status count ÷ all requests in that branch × 100

Both branches have four requests, so North’s closed share is 3 ÷ 4 × 100 = 75%. Dividing by all eight requests would give 37.5%, but that answers a different question: North’s closed requests as a share of the entire dataset.

For publication, prepare a flat table with one row per branch–status combination:

branch	status	request_count	branch_total	within_branch_pct
North	Closed	3	4	75
North	Open	1	4	25
South	Closed	1	4	25
South	Open	3	4	75

Keep the percentage values numeric and document that 75 means 75%, not 0.75. The repeated branch_total is a denominator, not an additive measure: summing that column would double-count each branch.

4. Check and publish the summary

Before sharing, confirm that:

• request_count sums to eight.
• Each branch’s status counts sum to four.
• Percentages sum to 100 within each branch; real datasets may differ slightly because of rounding.
• No subtotal or grand-total rows are mixed into the publishable data.

Use a clear title, such as Request status by branch—illustrative sample, and explain the row unit, cutoff, denominator and missing-value policy. For real data, add the source, preparation method and update date. W3C explains that table captions identify the content, while summaries explain complex organization; see the guide to writing table captions and summaries for examples.

Upload only information cleared for public release. Aggregation alone does not guarantee privacy: small counts can enable identification and reveal associated attributes, as the UK Government Statistical Service’s data confidentiality guidance explains.

As of October 28, 2026, TablePage supports CSV, TSV, XLSX and XLS uploads and creates public dataset pages with filterable tables. Upload the reviewed summary, then check the published labels, four-row count and numeric values before sharing the link. Readers should be able to filter by branch or status without losing the definition of the percentage they are viewing.