Author Archives: Paul McGarvey


Diagrams are graphics that can help explain information or enhance understanding. They are especially helpful for people who find reading difficult or those who are visual thinkers. Diagrams can include:

  • process maps
  • flowcharts
  • illustrations

A clear text description structured with headings and subheadings is the preferred method to use to explain information and communicate processes. We recommend trying this before using a diagram as it means that:

  • the majority of users can understand the content
  • the content is accessible to more users
  • search engines can read the text and display in search results
  • the content can be easily updated
  • users can zoom and alter the size of text for readability without it becoming pixelated

If you want to help your users visualise a complex process that links off in many directions, the use of a diagram can be considered.

If you are thinking about using a diagram, contact at the earliest opportunity and we can consider the options and advise on the best solution.

Process for publishing diagrams

Once you have identified there is user need to use a diagram, email with:

  • the name and date of your release and, if known, the number of the sharepoint tracker
  • the title and the main message of the diagram
  • any links or attachments to prototypes or rough sketches you have in an editable format so that text can be manipulated

We are constantly improving based on research and best practice. Any significant changes to our guidance are available on the Updates page.

Bulletin template

Download our template (Word, 170KB) to help you draft your bulletin in the new structure.

The bulletin template should only be used for bulletins. This is to make sure users have a consistent experience when they come to a bulletin page; the structure should be clearly defined and different to articles.  

We are constantly improving based on research and so there may be some small changes to the guidance; we will keep you updated with any significant changes on the Updates page.

Next section: Bulletin titles and summaries

What to link to

Research tells us that users have two separate needs from related links – to go into more detail, or to find broader but related content. 

Include between three and six links in this section. These links should be to:

  • articles containing deeper analysis of the topic than the bulletin can provide
  • the latest release of other related bulletins, including any related bulletins from the same theme day
  • analysis of data from multiple sources
  • recent ONS publications that also reference this specific topic 
  • relevant articles that are published by other official organisations

Links should help users get directly to relevant content. Do not link to: 

Links to data should sit in the Data section, and links to methodological articles should be in the Measuring the data section. If your release contains more than one bulletin, use the Other pages in this release section to link between them.

The content design team can use analytics to help you choose which links to include – email

Each link should include:

  • the title of what you are linking to
  • the type of content you are linking to (Bulletin, Article, Methodology, Wep page, Report or User guide)
  • the release date
  • up to 30 words describing what the link points to – use the summary of that page if appropriate

Consumer price inflation, updating weights
Article | Released 18 March 2019
An overview of the latest annual update of Consumer Prices Index including owner occupiers’ housing costs (CPIH) weights.

We are constantly improving based on research and best practice. Any significant changes to our guidance are available on the Updates page.

Strengths and limitations

This section should provide information to help users correctly interpret the data, including how the data should or should not be used.

This may include:

  • detail on the data’s accuracy and reliability, for example, if there was a sampling error or gap between collection and publication
  • detail about any uncertainty
  • detail about comparability with other sources or countries’ data
  • detail on any warnings used
  • whether the bulletin contains National Statistics or Experimental Statistics

It should not include any tables or charts. If necessary, you can link to relevant articles so that users can read more about the data’s quality.

See the April 2021 Labour market overview bulletin for an example.

National Statistics

If the bulletin has National Statistics status, use the standard wording and format to detail when it was last assessed and what improvements were made.

Quality and Methodology Information (QMIs)

If your QMI has a section that clearly explains the strengths and limitations of your data, for example the quality summary section, link to it directly from the bulletin.

If the strengths and limitations from your QMI are crucial to interpreting the data and can be summarised clearly in a few short bullet points, include them under subheadings of “Strengths” and “Limitations”.


Our Uncertainty and how we measure it guidance explains the different measures of uncertainty for users. This includes information on standard errorsconfidence intervals and statistical significance. Bulletins should include a standard line that links to this page:

“View more information on how we measure and communicate uncertainty for our surveys.”

See this COVID infection survey release for an example of how to address uncertainty.

We are constantly improving our content guidance based on user research, feedback and best practice. See our Updates page for the most recent improvement.

Next section: Related links

Measuring the data

Finding out more about how the data are gathered and measured is an important task for a smaller group of our audience, namely technical users. Providing short and clear explanations of our methodology helps us establish trust and credibility in our statistics. 

In fewer than 200 words, provide explanations of the data used in your bulletin, covering the following:

  • where we get the data from, for example, the survey or source
  • how we measure the data, for example, sample size and collection method
  • time periods covered, for example, the time periods and geography covered by the data

See the October 2019 Labour market overview bulletin for an example.

If necessary, summarise upcoming changes to the bulletin or methodology. You can also include information on why data revisions may occur; revised data and figures should be included in their own analysis section. If it is not possible to provide enough detail on these topics in this section, link to relevant articles.

Use a clear subheading for each topic to direct users to this information. For example, “Data source”, “Collection method”, “Coverage” and “Upcoming changes”.

For users who need more detail, include links to the Quality and Methodology Information (QMI) report and the user guide under a subheading called “Quality”. 

When linking to the QMI, using the following standard text: 

“More quality and methodology information on strengths, limitations, appropriate uses, and how the data were created is available in the (name of release) QMI.”

This section should include only text; it should provide summary information and so should not need any charts or tables.

Avoid including formulas or lengthy technical explanations in this section. This can be overwhelming for users and the information is available in the QMI and can be easily linked to.


Provide short, understandable definitions for users who may not be familiar with the terms or concepts described on the page.

Briefly explain technical terms in your analysis using plain English. Use the Glossary to give a little more detail about terms and concepts without interrupting your analysis. You do not need to hyperlink each term used in the analysis to the Glossary; the Glossary is in the table of contents and is used consistently across publications.

Include at least three and up to six terms, with a description of up to 50 words for each. List the terms in alphabetical order. 

Choose the most relevant terms that are commonly used in the analysis. Include a clear definition of any complex terms to prevent misunderstanding.

If you have a more detailed list of definitions on another page, link to it at the bottom of this section.

Employment measures the number of people in paid work and differs from the number of jobs because some people have more than one job. The employment rate is the proportion of people aged from 16 to 64 years who are in paid work.

The Guide to labour market statistics contains a glossary of other terms used in this bulletin.

The content design team can help you choose which terms to include and help you to write short, clear definitions – email

Preventing misuse of your data

Use “warnings” to highlight crucial limitations that affect how users interpret the data. They prevent misuse of data, with minimal interruption to the content.

Warnings are designed to stand out from your analysis so that users notice them; using too many, or including too much detail, distracts users. Only use warnings in the analysis section. You should not include any hyperlinks within the warning box.

For an example of how to include a warning box, see the Labour market overview, UK monthly bulletin.

How to write a warning


  • Highlight essential limitations of the data to help users avoid misinterpreting the data.
  • Keep warnings short and clear; when text is hard to understand, people retain less information.
  • Only use warnings when they have a direct effect on how users interpret the content around them.
  • Include only relevant information; use the Measuring the data or Strengths and limitations sections to add detail.

Do not

  • Use warnings too often; it disrupts the reading experience, makes it difficult to understand the content and reduces the effectiveness of each warning.
  • Place warnings next to each other as they overwhelm users and get in the way.
  • Position warnings before your analysis; testing has shown that users find it confusing to read warnings before commentary.
  • Provide definitions in warnings; instead, define terms briefly in your analysis, and use the glossary section to provide more detail.
  • Include hyperlinked text to other sections or articles; instead, any further detail should be included in the Strengths and limitations section, or linked to from there.

Warnings should be short

Warnings have a strict character limit of 280 characters. The shorter the warning, the more effective it will be. Presenting too much information to users makes it less likely that they will understand and retain the warning. Long warnings are also problematic for users on mobile devices or with accessibility needs.

If you need to explain the limitations of the data in more detail, expand on it in the Strengths and limitations section.

The content design team can help you to write short, effective warnings – contact if you want to discuss a warning you are working on.

We are constantly improving based on research and best practice. Any significant changes to our guidance are available on the Updates page.

Next section: Data and methodology


When writing about statistics, focus on what is interesting, noteworthy or important to the majority of users. Consider what is the most important information that supports your message.

If you want to know more about who uses your bulletin, take a look at our user personas or contact the content design team at

Your text should be concise, in plain English and written with the user in mind. Many may not be experts; 56% of our users access the website from home.

Use text and simple charts or tables to give more detail and context. Your written analysis should add to your visuals, not just report what they show. It is not necessary to comment on every trend shown in a chart or table.

Flag concerns about the data using warnings. These should be short and any more detail should go in the Strengths and limitations section.

How much to write

Analytics show that users spend an average of four minutes looking at bulletins. A typical person would read around 900 words during this time. Aim to limit the amount of analysis on the page to be shorter than this.

If you are unsure whether something is going to be useful or interesting, do not include it. Users can still find the data in datasets, and an accompanying article can provide further detail if there is a user need.

If you think your analysis will be more than 900 words, we can help structure your content – email

How to structure your analysis

Use section headings to break your analysis into broad themes. This helps users find the information they are looking for in the table of contents. Section headings should be short, descriptive labels that reflect themes users are interested in.

The analysis in the Employment in the UK bulletin could be structured around the headings “Employment”, “Unemployment” and “Economic inactivity”.

If your bulletin has a single topic of analysis, include the word “analysis” after the topic in the section heading, for example, Employment analysis. This will avoid duplicating the bulletin title in the section heading.

Include updated or revised figures as the final analysis section with the title “[Topic] revisions”.

Within sections use subheadings and chart titles that summarise the main trends to break up your analysis. These do not appear in the table of contents but research shows that subheadings make it easier for users to:

  • get the most important messages at a glance
  • find which part of the page contains what they are looking for
  • get a feel for what the following text or chart is going to tell them

Use a new subheading every time you discuss a new subject or trend. Put the most important point at the start. Subheadings should be a maximum of 75 characters including spaces to prevent the text wrapping over too many lines, particularly on mobile.

Employment rate for women was 72%, the joint-highest on record

You can use a chart title instead of a subheading to break up your text. Chart titles should follow our guidance and highlight an important trend in the figure. Avoid using subheadings and chart titles that say the same thing one after another. 

Main points

Highlight the most important and interesting findings from your bulletin at a glance. Most users read the main points and nothing else.

Rank up to six main points in order of importance for your users.

Each main point should:

  • be a single bullet point
  • contain one message that is expanded on in the bulletin
  • be a single sentence starting with what’s happened, followed by the significance of this; use a semicolon to split up the sentence if necessary

The UK unemployment rate was estimated at 3.8%; it has not been lower since October to December 1974.

Research shows that users want the headline figures quickly, so avoid prefacing your main points with any introduction or warnings. Do not introduce detailed definitions or quality warnings in your main points.

The content design team can help you to write impactful and user-friendly main points – email

If the data do not change month to month and only provide enough detail for Main points rather than full analysis sections, it may be better to write a headline release, a streamlined version of the bulletin structure.

Statistician’s comment

Some bulletins include a statistician’s comment. It is primarily for the media but should highlight something of interest to all users.

The comment puts the main findings in context; it should not just repeat something that the user can read elsewhere in the bulletin. 

The comment should:

  • be short; up to 75 words
  • be placed in double quote marks
  • be followed by the name of the person quoted, their job title and a link to a relevant  ONS Statistician’s Twitter profile (if there is one)

The 75-word limit will make sure the comment does not fill an entire mobile phone screen, or most of a desktop browser screen.

Shorter comments will help users with scrolling and getting to the analysis quicker.

The comment must be approved by the Media Relations Office.

“Marriage rates remain at historical lows despite a small increase in the number of people who got married in 2016. Most couples are preferring to do so with a civil ceremony and for the first time ever, less than a quarter of everyone who married had a religious ceremony. Meanwhile, the age at which people are marrying continues to hit new highs as more and more over 50s get married.”

Kanak Ghosh, Vital Statistics Outputs Branch, Office for National Statistics

Follow Vital Statistics Outputs Branch on Twitter @NickStripe_ONS

How to write comments

Writing statistician comments for media use is different from other kinds of writing on  the ONS website. Read the full Media Relations Office’s guidance (Word, 18.6KB). Also remember these tips:

  • use a conversational tone of voice; imagine you are explaining the point to an intelligent but non-specialist friend in a friendly chat 
  • avoid jargon and technical language; use everyday plain English
  • do not just wrap speech marks around your main points; explain the significance of what the bulletin reveals in words, ideally avoiding numbers altogether  
  • use well-chosen adjectives; for example “a sharp rise” rather than a “15% increase”
  • be careful when using metaphors or figures of speech and avoid clichés (like the plague!)
  • make sure the comment addresses the main angle or issue that the media story is likely to focus on – the Media Relations Office can help with this
  • never make predictions, chase headlines or sensationalise the numbers; public trust in what we do is far more important than a little extra media attention.