How a public relations firm can use ONS data

What to take away

How to cite ONS releases correctly in a press release

Start with the release itself, not with a headline built on it. Every ONS publication has a title, a reference period and a set of tables. If your press release cannot name all three, the claim is not ready.

Useful starting points include Labour market overview, UK, Consumer price inflation, UK, GDP monthly estimate, UK and Retail sales, Great Britain. For survey-led stories, look for the Opinions and Lifestyle Survey and check its sample, coverage and measures before using a result. Find the exact release and table on the ONS site, then cite their titles and the period they cover.

ONS methodology notes explain how a figure was produced, what it covers and what it excludes, as set out in the ONS methodology guidance. Read that before you write the sentence that uses the number. The Code of Practice for Statistics, published by the UK Statistics Authority and overseen by the Office for Statistics Regulation, is a useful quality benchmark for official statistics, not an advertising code for PR teams. The UK Statistics Authority was established by the Statistics and Registration Service Act 2007. Applying these checks helps make a defensible claim and avoid a retraction.

Use the ONS as the source in the body text, for example "according to the Office for National Statistics". Link to the exact release page. Do not link to a newspaper summary, because the newspaper may have rounded, spliced or dropped a caveat that your client now owns.

Quote the unit precisely. A figure can be a count, a rate, an index or a percentage change. "Output rose" and "output rose by 0.4 per cent" are different claims. The second needs the period, the base and whether the change is month on month or year on year.

Be careful with derived figures. If you calculate a percentage from two ONS numbers, say so in the release and show the arithmetic in a note. Readers who cannot reproduce your sum will assume the worst.

For agency-sector claims, the England PR agency market guide provides industry context, but it cannot make an ONS dataset measure a specific industry. For claims about PR businesses, identify the relevant industry classification, SIC 70.21, “Public relations and communication activities”. The Annual Business Survey may help with relevant business measures where it publishes suitable data, but check that its coverage and measure support the claim; it does not automatically represent the PR agency market.

Use ONS definitions as written. The ONS about-us page sets out its remit and how it describes its own outputs, and those descriptions travel with the data. Renaming a category to suit a campaign is the fastest route to a correction.

Keep the release date visible. A statistic from 2023 quoted in a 2026 press release needs the date attached, or the reader will assume it is current. Date every figure in the body, not only in a footnote.

Where a release has a statistical commentary, read it. The commentary usually states the uncertainty the tables cannot show, and it is written by the people who know the data best.

If the client insists on a stronger line than the data supports, weaken the line, not the caveat. A weaker true claim survives scrutiny; a stronger false one does not.

Checking sample sizes, confidence intervals and revisions

A percentage from a survey is not a fact about a population unless the sample supports it. Before quoting any survey-based ONS figure, find the achieved sample size and the group it represents.

Small subgroups are the usual failure point. A UK-wide survey may have enough respondents for a national estimate and too few for a reliable Wales or Northern Ireland breakdown. If the ONS suppresses or flags a subgroup, treat that as a warning.

Confidence intervals tell you how much a figure could move. If the interval is wide, report the range or describe the direction rather than a precise number. Quoting the midpoint alone overstates certainty.

A difference between two periods or two groups is only a finding if it is larger than the uncertainty around each estimate. Many "record highs" in press releases are inside the margin of error.

Revisions are normal, not scandalous. The ONS revises earlier estimates as more data arrives, and the ONS revisions policy explains how and when that happens. A figure that was correct in March may be different in September.

Check whether your number has already been revised before you publish. If it has, use the current value and note the revision if the change is material to the story.

Build a two-line evidence note for every statistic your campaign will use. Line one: source, release, period, sample size. Line two: the caveat a hostile reader would raise. This habit sits at the centre of any PR measurement system, keeping evidence levels apart instead of blending survey, administrative and modelled data.

Where the ONS publishes experimental or modelled statistics, say so. These outputs are useful for direction and unsuitable for precise claims, and the ONS labels them for that reason.

Never round a figure in a way that changes its meaning. Rounding 0.46 per cent to 0.5 per cent is fine; rounding it to "half a per cent" in a headline invites a complaint.

Avoiding ASA and IPSO pitfalls in statistics-led PR

The ASA's misleading advertising rules apply to marketing communications within the CAP Code's remit; the BCAP Code governs broadcast advertising. A statistics-led press release promoting a product or service can fall within the relevant code. CAP Code rules 3.1 and 3.7 cover misleading advertising and substantiation; the corresponding BCAP Code rules are 3.1 and 3.7. The test is whether the overall impression is misleading, not whether each word is literally true.

Use the CAP Code for non-broadcast marketing communications and the BCAP Code for broadcast advertising when those codes apply. Check whether a press release falls within the relevant code's remit; a press release is not automatically an advertisement. Under the applicable code, hold evidence to substantiate objective claims before publication, and keep it for the channels where they appear.

IPSO statistics pitfalls are a separate risk. The Editors' Code requires care over statistics and distinguishes comment from fact. A press release that presents a forecast as a finding gives newsrooms a problem they will pass back to you.

Comparisons need a stated basis. "Twice as many" needs to say twice as many as what, in which period, for which population. Without that, the claim is unverifiable and the ASA can treat it as misleading.

Superlatives need evidence of a full comparison. "The fastest-growing region" requires a dataset that covers every region on the same basis. If your source covers England only, say England.

Keep a substantiation file for each campaign claim. It should hold the release link, the table, the calculation and the date checked. If a complaint arrives, that file is your defence and your correction plan.

Correct fast when you are wrong. A published correction is cheaper than an upheld ruling, and both the ASA and IPSO look more kindly on a prompt, visible fix.

Before release, run the advertising checks that apply to the material you are putting out, including any paid support behind it. The rules do not change because the channel is earned rather than bought.

Using ONS release calendar to time a campaign

The ONS release calendar lists publication dates in advance, which makes it a planning tool rather than a news feed. Search the ONS site for the release calendar and the exact release title you plan to use. Put relevant releases in the campaign plan at the start, not the week before launch.

Work backwards from each release date. Drafting, legal review, client sign-off and media briefing all take days, and a release that lands at 9.30am leaves no room for a same-morning scramble.

Build a simple timing table for the campaign.

| ONS release used for a planned claim | Prepare and sign off the press release | Two to six weeks, depending on the release. | | Release with a regional breakdown | Check geography and prepare a regional angle | Two to six weeks, depending on the release. |

Avoid the pile-up problem. When several releases fall in the same week, journalists are saturated and your story competes with the ONS's own numbers. Pick a quieter window or accept a smaller share of attention.

Do not brief a figure before publication. The Pre-release Access to Official Statistics Order 2008 regulates access before publication: the usual maximum is 24 hours, with specified cases allowing up to five days. Check which arrangement applies to the statistic and do not treat permitted access as permission to publish early.

If a release is postponed, have a fallback claim ready that does not depend on it. Calendar changes are routine, and campaigns that rely on a single date are fragile.

Align regional angles with the release that actually covers the region. Nomis is a route to local labour-market statistics, while ONS Explore local statistics provides local area information; check the geography and measure in the particular data. A Scotland or North West England story needs a dataset with a Scotland or North West England sample, not a UK total with the region's name attached.

A statistics citation checklist for PR teams

Run this before any statistics-led press release leaves the building.

Worked example. A client wants to say its region is the UK's fastest-growing for new businesses. The ONS release Business demography, UK is a relevant starting point: check its business births by region table for the precise geography, period and ranking before making the claim.

If the current table supports only a narrower statement, use that wording, name the period, link the exact release and identify the geography it covers. Keep the superlative out unless the data supports a complete like-for-like ranking.

That version is defensible because its wording follows the table's coverage. The original would not be, because it implied UK coverage and a ranking the data did not provide.

Keep the checklist with the campaign files. When the next release revises the figure, you will know exactly which sentences need updating and which do not.

For fee or volume claims built on agency surveys, the PR agency benchmark research guide explains why population, unit and method have to be stated before any comparison is made.

Common errors when using ONS data in PR

Confusing correlation with cause is the most common. ONS data can show that two things moved together. It rarely shows that one caused the other, and a press release that implies causation invites challenge.

Letting the headline outrun the table is next. The headline is the claim most likely to be quoted and most likely to be complained about. Write it last, from the evidence note, not from the ambition.

Define your metrics before the campaign, not after the results arrive. A shared dictionary of PR metrics stops the same word meaning three things across a release, a blog and a client deck.

Common questions