Cookies on

Cookies are small files stored on your device when you visit a website. We use some essential cookies to make this website work.

We would like to set additional cookies to remember your settings and understand how you use the site. This helps us to improve our services.

You have accepted all additional cookies. You have rejected all additional cookies. You can change your cookie preferences at any time.

Skip to main content

Content types on the ONS website Datasets


A dataset is a page that contains a downloadable spreadsheet in XLS or CSV format, or a customisable dataset (opens in a new tab) 

The page also includes any essential information for users about the data and how to use them. It cannot contain any analysis, charts or presentation tables.

Types of dataset pages

There are two main types of dataset pages: single file datasets and multiple file datasets.

See the following examples:

We also produce some customisable dataset pages for high-profile topics. Here, the user can filter and customise the data download for their needs. View an example in Annual GDP for England, Wales and the English regions (opens in a new tab) 


Once you have chosen the page type and the dataset page has been created, it cannot be changed.

Single file datasets

A single file dataset is where one file is shown on the page, for example, the monthly Output in the construction industry dataset (opens in a new tab) 

This file is replaced each time a new version of the dataset is published. The file is labelled as the “current edition of this dataset” and this heading cannot be changed. Superseded versions of the file are available on the “previous versions” page.

Important information:

Use a single file dataset if you only have one file and your users only need the most up-to-date data.

Multiple file datasets

Multiple file dataset pages are where more than one file are shown on the page, for example, the Weekly deaths dataset (opens in a new tab) 

This means you can present the most up-to-date data for multiple time periods and geographies on a single page, without users having to scroll through all the previous versions. Each file has a unique label showing the time period or geography it covers. Each file also has its own previous versions page with superseded versions of the data.

Important information:

Use a multiple file dataset if you want to present the latest data for different time periods or geographies, or want to publish provisional, final and revised data.

What to include on a dataset page

There are a small number of editable fields on a dataset page:

TitleA title is the most impactful way to help users find data. Use our guidance to ensure that titles are short, clear and in plain language
ContactA team name and team email address for users to contact with any queries about the data.
About this datasetA summary of what the dataset includes. Try to include words and phrases that users may search for.
Your download optionAn automatically generated list of links to the latest release and previous versions of the data.
Important notes and usage informationAn optional section which appears beneath the downloads list. This content is not very prominent, and an unreliable way to give users information about the data. Use sparingly.
RelatedThere are a number of different boxes that can be used to link to:
  • Related data
  • Related QMIs
  • Publications that use this data

Dataset titles and summaries


Dataset titles should describe what the dataset contains. They should:

  • be frontloaded (with the most important information first)
  • include the geographical region the data apply to
  • be short – aim for 60 characters including spaces
  • not include any acronyms – write any phrases out in full for the title and include any acronyms in the keywords
  • not include the word “dataset”
  • time series datasets should include “time series” at the end of the title
  • not include the time period (this should only be used in the spreadsheet title and edition name)

If you have one dataset, use the same title as the publication (but without the data period) for consistency.

Deaths registered weekly in England and Wales, provisional

The Public sector employment, UK: June 2020 (opens in a new tab)  bulletin is published with the Public sector employment (opens in a new tab)  dataset.

If you have multiple datasets, make sure the titles are clear and concise so that users can easily distinguish between them.


The “About this dataset” section of a dataset page provides users with a summary of what they can expect to find in the Excel downloads.

This summary should be clear, concise and frontloaded, using plain language.

It should contain the geographical coverage, if it is not already in the title, and the frequency of the data for example, annual, quarterly, monthly. It should not include the time period for the data as this will change with each release and the page may contain multiple dataset downloads with different time periods.

The summary should not exceed 160 characters (including spaces) and end in a full stop.

Annual burglary data from the Crime Survey for England and Wales (CSEW): when incidents happened, information about offenders, the victim’s perception of the incident, and what items were stolen.

Official statistics in development

If you are publishing a data only release that is labelled as official statistics in development (opens in a new tab) , include an additional sentence at the end of your summary that says: “These are official statistics in development.”

Daily UK flight numbers and rolling seven-day average, including flights to, from, and within the UK. These are official statistics in development.

You can also include a short sentence summarising how the data are being developed in the Important notes and usage information section (opens in a new tab) 

Important notes and usage information

Important usage information for your data can be added to the main dataset page. This should be limited to essential points your user needs to be aware of when interpreting your data. 

This section can include:

  • a summary of important main points if the data will not be released alongside a bulletin or article; these should be no more than six bullet points and a single sentence for each 
  • essential warnings or limitations for using the data to avoid misinterpretation
  • a short sentence summarising how the data collection method or quality are being developed or improved

Formatting a dataset file

Finding out what format your users need your data in is the best approach to providing a data download. Depending on the size of your data, datasets can be provided as XLS or CSV. If you are looking to use a compressed version of a large amount of data, a CSV file is best for this – but bear in mind that CSV files are designed to be read by machines and do not allow any formatting.

All dataset files must be accessible so that they can be used by everyone. Use the built-in accessibility checker (opens in a new tab)  in Excel to identify and fix any issues. However, do not rely on this check alone.

Important information:

Read more detailed guidance from the Government Analysis Function on releasing statistics in spreadsheets (opens in a new tab) 

File names

Datasets should have a unique, lower case file name using dashes instead of spaces or underscores between words. File names should be short and concise. They should make it clear to the user what is included in the data download so that it makes sense out of context.

All dataset file names must be shorter than 218 characters or the file may not open for some users.

Creating the dataset

Datasets should be “marked up” as tables and the dataset title should be added in cell A1. The dataset title should follow house style, be front-loaded with the topic and include the geography, frequency and data period.

When creating your table make sure you:

  • use a sans serif (tail-less) font, for example, Arial
  • use at least size 10 font for body copy and at least size 12 for titles
  • select the “automatic” colour
  • avoid using italic text anywhere or bold text for the data
  • avoid the use of cell borders on cells with lots of text in
  • avoid highlighting text with a background fill
  • check that all text is horizontal

Contact details

If you include contact details within your dataset file, use a team name and team email address. This is because we present our content as a collective organisation: the ONS.


Give each tab a unique name that clearly describes the information found in the worksheet and avoid using non-descript titles like “Sheet 1” as screen readers will read out tab names. Aim for fewer than 31 characters for each tab name and remove any blank tabs for easier navigation.

Ensure the dataset file opens on the first tab to avoid confusion. Include contact details and the date of the next update on the first tab. If the file contains several worksheets include a table of contents on the first tab.

Separate tables out onto individual worksheets for easy navigation.


Use clear column headings to provide context to the data and help users to navigate and understand the contents.

Users with dyslexia can find it difficult to read a lot of crowded text. It can be helpful to make the most of white space and extend the height or width of columns so that text sits comfortably and clearly within the cells.


  • Use a simple table structure with clear headings and subheadings as most accessibility software will navigate an Excel file by reading the table header rows.
  • Use clear and consistent titles across all tables for straightforward data usability.
  • Do not split or merge cells as this affects accessibility.
  • Do not hide or use blank rows or columns, instead adjust the column width and row height to create space. 
  • Give your tables one tagged header row. When marking up select “my table has headers” and make sure to write in your header.
  • Wrap text within cells to make sure all text is visible and clearly spaced out.
  • Avoid including images and charts.
  • Do not use colour in a table to convey meaning that is not shown in another way, as this can make the text hard to read and distinguish from the background.
  • Delete any blank worksheets.
  • Avoid putting content below the table and having more than one table on a worksheet.

Screen readers keep track of their location in a table by counting table cells. Try not to:

  • include blank cells, columns or rows
  • nest tables within other tables
  • merge or split cells

These things can be misleading to a screen reader. This type of formatting can cause it to lose count and cannot provide helpful information to the user.

Using colour

Do not use colour to convey meaning that is not shown another way, such as with a symbol. If there is a situation where you cannot avoid using colour, make sure the contrast meets WCAG AA standard (opens in a new tab)  and that the information is explained clearly elsewhere in the dataset.

Use black or dark grey for all text as the higher the level of contrast between the text and background, the more people can see and use the content. This is especially helpful for users with low vision.

Cells with no data

When cells with no data are left empty it can make it difficult for users of assistive technology to work out where the table starts and ends.

Blank cells should not cause accessibility issues if:

  • there is only one consistent reason a cell in a table may be left blank
  • the table is marked-up correctly
  • there is a note above the table, in a cell in column A, explaining that some cells are left blank and why

However, if there are several reasons a cell in a table may be left blank, you will need to describe why a cell has no data. Do not use symbols like full stops (..) or dashes (-). Instead, you should use the Government Analysis Function's guidance on how to use symbols in tables. (opens in a new tab) 

Put these in square brackets. If cells need to be left empty for more than one reason explain the reasons for this on the cover sheet.

Do not use “NA” to describe cells with no data. This shorthand is ambiguous, some may read it as “Not Applicable”, others may read it as “Not Available”.


Short footnotes

You should include as much information as possible where it is needed. For example, when data is provisional or revised put the whole word in square brackets instead of using ‘[p]’ or ‘[r]’. If you need to use letters to avoid visual clutter make sure you include a key to explain them in column A.

Do not use superscript text or symbols. For more advice, see the Government Analysis Function's guidance for more advice on how to use symbols in tables (opens in a new tab) 

Detailed footnotes

Signpost to detailed footnotes using the word “note” and a number in square brackets, for example: “[note 1]”. If a cell refers to more than one footnote you should list each in a separate pair of brackets, for example: “[note 1][note 2][note 3]”.

Always try to put note markers in table titles, column headings or row labels. If you need to mention notes for specific data points, you should add a notes column to the table, on the right. You should describe which cell or cells the note applies to, for example: “[note 1] This note applies to B10, C10 and D10”.

Put an explanation on the cover sheet highlighting where notes can be found and how note markers are presented, for example: “Some tables refer to notes. When notes are mentioned the note marker is presented in square brackets. The note text can be found in the notes table.”

We recommend creating a worksheet called “Notes” that contains a table listing all of the detailed notes for the spreadsheet.

Footnotes for a specific cell

If a note marker would not be useful in a title, column or row heading, you can create a “notes” column on the right side of the table. Here, you should describe which cell or cells the note applies to, for example: “[note 1] This note applies to B10, C10 and D10”.

Codes and symbols

  • Use nationally recognised classifications, such as geography codes, and keep users updated with any changes to these where possible.
  • If you are using symbols for communicating uncertainty, include a guide to explain to users what each of these mean.  
  • Add codes and symbols in a separate column to the data.
  • Avoid using indentation to show geographic hierarchy. 

Alt text

Datasets should not include images, graphics, charts or embedded objects. Where this cannot be avoided, include alt text with all visuals to help users navigating the file with accessibility software understand the importance.

Users accessing your data with a screen reader are able to scan a list of hyperlinks in order to find what they are looking for. To make this as easy and useful as possible, make sure to use meaningful hyperlink text that provides accurate information about the link destination.

Read our links guidance

What is a data-only release?

A data-only release is a release that contains either one dataset or multiple datasets. It does not include any accompanying commentary or analysis with the data.

Use a data only release when there is:

  • little change in the data month on month
  • limited user engagement
  • no additional commentary or analysis is needed alongside the dataset

If some additional main points and quality information are needed alongside your data, consider using a headline release

Moving to a data-only release from a bulletin

If you publish a bulletin or headline release and are thinking about making it data only, you should consider:

  • whether users rely on the supporting content in a bulletin or headline release, and how they will be affected if that ends
  • the implications of users using the data without easy access to caveats, strengths and limitations
  • whether you might need to regularly include information or analysis that helps users understand the data

Let users know about changes to a bulletin

A notice is a short alert that lets users know a dataset will no longer have an accompanying bulletin. It sits in a blue banner at the top of the page. It is important to use a notice to let users know about changes.

Use the following templates to create a notice.

One dataset release

This bulletin has been discontinued and does not contain the latest data. We will continue to publish the latest data for [Topic/bulletin name] in the [Dataset name and link] dataset.

This bulletin has been discontinued and does not contain the latest data. We will continue to publish the latest quarterly data for Consumer trends in the Consumer trends time series dataset (opens in a new tab) 

Multiple dataset release

This bulletin has been discontinued and does not contain the latest data. We will continue to publish the latest data for [Topic/Bulletin name] in the following datasets:

  • Dataset name and link
  • Dataset name and link
  • Dataset name and link