Make visual data readers can understand and check

To share a spreadsheet clearly, give readers two ways into it: a chart that reveals a pattern and a table that lets them check the values. The useful result is not just a graphic—it is a public page with the data, definitions and sources behind it.

Here, visual data means tabular information presented through charts and readable tables, rather than image or video datasets. Start with one reader question, prepare the supporting columns, then publish both the visual and its underlying values.

1. Define the comparison before choosing a chart

Suppose you want to compare activity across three library branches. “Which branch recorded the most visits?” and “Which branch recorded the most visits per opening hour?” need different measures.

This fictional demonstration dataset contains one row per branch for September 2026:

branch	month	visits	open_hours	visits_per_open_hour
North	2026-09	1200	200	6.0
Central	2026-09	900	100	9.0
South	2026-09	600	150	4.0

Calculate the rate as:

visits_per_open_hour = visits / open_hours

North has the most total visits; Central has the highest rate. Neither chart would be wrong, but they answer different questions. Visits per opening hour describe activity relative to opening time—not service quality, crowding or staff workload.

Write the intended question above your working table. The Office for National Statistics recommends choosing a chart based on both the data type and the relationship you want to communicate, favouring simple, familiar charts focused on that relationship.

2. Prepare values, not just spreadsheet formatting

Create a publication copy with:

• One header row: no merged title cells or grouped headings inside the data range.
• One observation per row: here, a branch in a particular month.
• Numeric measures: store 1200, not the text 1,200 visits; explain units separately.
• Consistent categories and periods: do not let Central and Central Branch become separate groups accidentally.
• Explicit missing-value rules: an unknown visit count is not a confirmed zero.

Keep both the numerator and denominator when publishing a rate. Readers should be able to reproduce 900 / 100 = 9.0, rather than receiving an unexplained score.

If opening hours are zero or missing, leave the rate missing and document why. Do not turn an undefined rate into zero. For more on that distinction, see empty fields, zero and missing values.

Remove personal identifiers and sensitive information before export. Publish only data you are authorised to make public; do not rely on hiding a column in the working spreadsheet.

3. Make the visual answer the stated question

For “Which branch had the most visits per opening hour?”, use a horizontal bar chart of the rate, ordered Central, North, South. Label the value axis “Visits per opening hour” and start it at zero. ONS bar-chart guidance recommends bar charts for comparisons across categories, a zero baseline and a logical category order, typically largest to smallest.

Use a specific title:

Visits per opening hour by branch, September 2026

Add a short interpretation:

In this fictional example, Central recorded 9 visits per opening hour, compared with 6 at North and 4 at South. North recorded the most total visits.

Do not use colour as the only way to identify a branch: use visible text labels too, consistent with W3C guidance on using colour. Keep labels and values available outside hover interactions, such as in the accompanying table.

For a chart image, provide a short text alternative and an accompanying description of important values and relationships. W3C’s guidance for complex images illustrates descriptions that include scales, values, relationships and trends, with a table where useful.

4. Publish the table with enough context to check it

As of October 19, 2026, TablePage advertises CSV, TSV, XLSX and XLS uploads that generate public dataset pages with filterable tables and automatically generated charts; readers can explore without signing up.

Upload the reviewed publication file, then inspect the resulting table and visuals. Use a generated chart only if its measure, grouping and labels match your question. An automatically selected chart is not a substitute for that check.

Alongside the shared page or embed, include:

• A dataset title and coverage period.
• Definitions of visits, opening hours and the calculated rate.
• Source links and the extraction or release date for real data.
• Missing-value rules and important exclusions.
• A clear “fictional demonstration data” label for this example.

For concise table context, use the patterns in accessible data table captions and summaries.

Before sharing, open the public link as a reader. Confirm that all three rows remain, numeric values are intact, and the chart and table agree. The page should make the pattern visible without making its evidence harder to inspect.