How ONS and BARB data can inform UK programmatic audience planning

What to take away

Why official UK data beats platform-reported audiences

Platform-reported audiences are useful for delivery, not for planning. They describe who a system thinks it reached after the fact. They rarely tell you how many people in Leeds, Glasgow or Cardiff could plausibly be reached in the first place.

Official UK data gives you the denominator. ONS population and census data tell you how many people and households exist in a geography, with age, employment and housing characteristics attached. BARB viewing figures tell you how many people watch specific channels and services, including on connected TV sets.

That combination matters because UK media planning operates under rules that platform dashboards do not model. UK GDPR and PECR constrain how identifiers can be used. The ICO enforces those rules. The CMA has examined digital advertising markets, and the Digital Markets, Competition and Consumers Act 2024 gives it stronger powers over firms with strategic market status.

A planner who starts with official data can show a client why a segment is small in North East England and large in Greater London before a single impression is bought. That is a different conversation from one about platform reach estimates that change when settings change.

Platforms also report on their own inventory, not the market. A broadcaster's streaming service, a commercial network and a video-sharing platform each count differently. Official data does not remove those differences, but it gives a common reference point.

For a fuller treatment of how panels and first-party data fit together, see our guide to audience research. The point here is narrower: official data sizes the market, platform data reports the buy.

ONS population and census data for segment sizing

The ONS census is the backbone of UK audience definition. It provides demographic and household data down to small geographies, which is what you need to turn a segment description into a number.

The headline release is titled Population and household estimates, England and Wales: Census 2021. It put the two nations at 59,597,300 people on census day in March 2021.

Start with the variables that actually change media behaviour: age bands, household composition, tenure, car ownership, employment status and qualifications. These are available for England, Wales, Scotland and Northern Ireland, though the devolved statistical agencies publish some outputs separately.

Census data is not a targeting list. You cannot lawfully push census records into a programmatic platform as identifiers. What you can do is use aggregate counts to size a segment and to weight a postcode or region in your plan.

ONS home internet and social media usage statistics add the behavioural layer. They tell you what proportion of adults use the internet, how that varies by age and region, and which activities are common. That is directly relevant when you decide whether a segment is reachable online at all.

Device ownership matters just as much. ONS internet access data on device ownership informs audience reach because a segment that skews to older adults on a fixed connection behaves differently from one on multiple mobile devices. You can read the detail in the ONS internet access release.

The individual-level figures come from the ONS bulletin Internet users, UK: 2020. That bulletin put recent internet use at 92% of UK adults, with higher rates in younger age bands. Later Great Britain bulletins put the rate a little higher.

The census itself is at the ONS census hub, and the broader internet and social media series sits alongside it. Use both when you build a written profile, because a target audience that is not sized is only a description.

A segment sizing method

Step five is where UK GDPR and PECR bite. If you cannot lawfully address a group programmatically, its addressable size is lower than its population size, sometimes much lower.

BARB viewing figures for video and connected TV planning

BARB is the UK's television audience measurement body. Its viewing figures cover broadcast channels and, increasingly, streaming and connected TV viewing on television sets. For video planning, it is the closest thing to a common currency.

BARB's Establishment Survey describes the panel and the equipment found in UK homes. The weekly viewing figures appear in BARB's viewing summary. The panel covers around 5,100 homes, which represents roughly 12,000 people, and that sample sits behind every published figure.

Connected TV planning in the UK depends on BARB because the set is where the viewing happens. A household may stream through a smart television, a games console or a set-top box, and BARB's panel captures much of that activity alongside linear viewing.

Use BARB data to answer three questions. How many people watch the content you are considering? What is the age and gender profile of that audience? How does viewing vary by region and by time of day?

Those answers change channel selection. A campaign aimed at 25 to 34 year olds in Manchester may find better video reach through streaming inventory than through a linear spot, and BARB data can show the difference without relying on a platform's own estimate.

BARB does not cover every platform. Some digital video is outside its measurement, and reporting lags mean the latest weeks may be modelled or provisional. Treat BARB as the best available estimate for television-set viewing, not as a full picture of online video.

It also does not replace platform reporting for addressable buys. The two are complementary: BARB tells you the size of the viewing opportunity, the platform tells you what it delivered. Our note on measurement without vanity metrics explains how to reconcile them.

How to use BARB in a plan

Pull the audience estimates for the programmes and dayparts that matter. Convert them into a regional picture using BARB's regional breaks. Compare that picture with the ONS population base for the same regions.

Regional breaks come from BARB's planning service AdvantEdge, or through a BARB data licence held by your agency or media owner. BARB regions follow broadcast areas, so they do not line up exactly with ONS regions or ITL geographies. Reconcile them with a postcode or local authority lookup and record the mapping as an assumption in the plan.

Where BARB shows strong viewing but ONS shows a small population, the opportunity is concentrated. Where ONS shows a large population but BARB shows weak viewing, you may need different channels or different creative.

Combining ONS and BARB inputs into programmatic audience segments

Combining the two sources produces segments that are sized, reachable and defensible.

The output is a range with a stated method. That is more useful to a client than a platform estimate, because it can be checked and updated as new ONS releases and BARB data arrive.

This approach also supports buyer personas that are grounded in UK data rather than assumption. A persona with a population count behind it survives scrutiny in a media review.

Keep the layers separate in your documentation. If the client challenges the size, you can show which input moved. If BARB revises its figures, you can update one layer without rebuilding the whole plan.

ONS home internet and social media usage statistics are the bridge between population and platform. They tell you how many people in a group are online, which is the precondition for any programmatic buy. The main ONS internet and social media usage collection is the place to start.

External research can fill gaps that official data leaves. Ipsos publishes audience measurement research relevant to media planning, and its UK research hub is a reasonable secondary source for attitudes and media habits that ONS and BARB do not cover.

BBC R&D publishes work on media technology and audience behaviour, including research into how viewing is changing. Its research pages are useful when you need to explain why connected TV measurement is evolving.

The four-layer model behind a sized segment

Limits of both datasets and how to state uncertainty

ONS data is sound but not immediate. Census outputs are published years apart, and between censuses the ONS relies on surveys and administrative data. Population estimates for small areas carry uncertainty.

ONS internet access releases often cover Great Britain rather than the whole United Kingdom. That leaves Northern Ireland out of some headline figures, so a UK-wide plan needs a separate source or a stated assumption.

BARB is a panel. Panels are precise about the behaviour they capture and limited in how far they can be broken down. Regional and small-audience estimates carry wider confidence intervals than national ones.

BARB covers television-set viewing well, but not all video. YouTube on a phone, social video in a feed and some streaming apps fall outside its core measurement. Do not present BARB figures as total video reach.

There is also a definitional gap. ONS counts people and households. BARB counts viewers of content. A person can be in both datasets, but they are not joined at the individual level, and you should never imply that they are.

State uncertainty explicitly. Give a range, name the source of each input, and say what would change the estimate. A limitations statement is not an admission of weakness; it is what makes the plan auditable.

Regulatory context belongs in that statement too. The ASA polices advertising claims, including some targeting and data practices. The IAB UK and the IPA publish guidance that shapes how UK agencies document audience work. Ofcom regulates broadcasters and has an interest in measurement standards.

A worked UK audience planning example

A UK retailer wants to reach adults aged 30 to 49 in the West Midlands and Yorkshire and the Humber with a connected TV campaign for a homeware range. The budget is modest, so the plan must be tight.

Step one is the ONS base. Census 2021 puts the West Midlands at about 5,950,800 people and Yorkshire and the Humber at about 5,480,800, a combined 11.4 million. Adults aged 30 to 49 make up roughly a quarter of that total, so the planner starts with a base of about 2.9 million people.

Step two is online and device reach. ONS internet use runs at 92% for UK adults overall and higher in the 30 to 49 band, so the planner carries an online base of about 2.9 million adults. Most of that group reaches video through a smart television or an equivalent device.

Step three is viewing. BARB's viewing summary gives the weekly reach for each strand in the two regions. Take a working assumption of 30% weekly reach among 30 to 49s for the chosen connected TV inventory, which gives a weekly reachable pool of about 870,000 adults.

Step four is addressability. The retailer's consent and identity setup means only a share of the reachable group can be targeted programmatically. On a working assumption of 60%, the addressable pool is about 520,000 adults a week, and the planner records that share as an assumption.

Step five is the range. A reach assumption of 25% gives a weekly pool of about 725,000 and an addressable figure of about 435,000. The planner presents the low and high estimates alongside the platform's own forecast, not instead of it.

Budget follows from the addressable figure. Multiply planned impressions by the CPM the platform quotes for the chosen inventory, then add production and measurement costs. The CPM depends on targeting, daypart and time of year, so the planner carries it as a separate assumption rather than a fixed price.

Step six is the check. After the campaign, the planner compares delivered reach with the modelled range. If delivery sits well outside it, the inputs or the addressability assumption need revisiting.

Checks to run before the plan goes to the client

Common questions