How UK planners read BARB data for connected TV

What to take away

What BARB actually measures across linear and streaming viewing

BARB, the Broadcasters' Audience Research Board, is the UK's official source for television audience measurement.

What is Barb and how do we measure total viewing? sets out the principle: Barb reports total viewing, not viewing of one device or one app.

That matters for connected TV advertising because the same household can watch the same broadcaster through an aerial, a smart TV app, a games console or a streaming stick. BARB's job is to count the viewing wherever it happens, then attribute it to a channel or service.

In practice this means the panel captures viewing on the television set itself, whether the signal arrives by broadcast or by internet. Viewing on laptops, tablets and phones is captured separately and reported as non-TV-set viewing, which is a different line in the data and a different planning conversation.

The practical consequences for a planner are straightforward. A broadcaster's total viewing figure includes its streaming app, which is why a BVOD campaign can look underweighted if you only read platform dashboards. It also means you cannot simply add platform impressions to BARB reach, because the same viewer may appear in both.

Barb's Total Identified Viewing summary gives the headline volume picture for UK viewing. Read it as the denominator for any share or reach argument you make, and as a sanity check when a platform claims a large share of a small audience.

If you are new to the category, it helps to research the audience before you argue about currencies, because the definitions you choose shape what you can later prove.

Linear and streaming are not separate universes

Linear channels and their streaming equivalents are reported within one measurement system. A drama shown on a broadcast channel and then streamed on the broadcaster's app contributes to the same service total, split by platform.

This is the single most useful fact for UK planners. It means a connected TV buy can be planned against the same audience definitions as a linear buy, with the split shown rather than assumed.

It also means the phrase "CTV audience" is a delivery mechanism, not a demographic. The audience is still defined by age, sex, class and region, and BARB reports it that way.

What BARB does not measure

Barb does not measure whether an advert was seen, whether it was viewable, or whether it drove a sale. Those are the responsibility of the buyer and the platform.

Barb also does not cover subscription video on demand services that do not participate in the panel, and it does not report the walled-garden campaign data that platforms hold internally. If your plan depends on those numbers, Barb is the wrong tool and you should say so early.

The BARB metrics UK planners receive for connected TV advertising

The metrics arrive through the Barb data licence, usually inside a planning tool such as AdvantEdge, TechEdge or a licensed third-party interface. What you get is a standard set, and knowing which one answers which question saves hours.

The table is deliberately short. Most planning arguments are won or lost on reach, share and minutes, with demographics used to prove the audience was the one you bought.

Reach and frequency work together. A plan that buys heavy frequency in a narrow set of programmes will show strong average minutes and weak incremental reach. Barb data lets you see that before you spend, provided you look at the reach curve rather than the average.

Share is the metric most often misused in pitches. A high share of a small total is not the same as a large audience, so always pair a share claim with the total identified viewing figure behind it.

Demographic splits are where regional planning lives. Viewing patterns differ across the UK's nations and English regions, and Barb reports at that level so you can weight a schedule accordingly.

Barb metrics are not campaign metrics. They describe viewing of services and programmes, and you translate that into an advertising plan using spot or impression factors supplied with the data. If that translation step is skipped, the plan will not reconcile with the platform report.

When the numbers start to disagree across sources, it is usually a definitions problem rather than a data problem, and it is worth reading up on common measurement mistakes before rebuilding the plan.

Reporting panels, sample design and confidence intervals explained

BARB's numbers come from a reporting panel: a recruited sample of UK homes whose viewing is metered continuously. The panel is the instrument, and like any instrument it has a margin of error that planners should quote rather than hide.

As a working guide, a national monthly reach figure typically carries a confidence interval of 1 to 2 percentage points. A single region measured over a single week typically carries 5 to 8 percentage points. Use these magnitudes to judge whether a difference is worth acting on.

Sample design matters more than sample size in isolation. The panel is built to be representative of UK households by region, household size and other characteristics, and it is refreshed on a rolling basis so that it does not drift away from the population it represents.

Each home's viewing is captured by meter, matched to programme and service information, and then weighted up to the population. The weighting is what turns a few thousand homes into a national figure, and it is also where most of the uncertainty enters.

Confidence intervals express that uncertainty. A reach figure is an estimate with a range around it, and the range widens as you slice the data more finely: by region, by small demographic cell, by individual programme, by week.

This is the practical rule to carry into meetings. National monthly figures are stable enough to trade on. A single programme in a single region in a single week is not, and treating it as precise is how plans get overfitted to noise.

Small channels and niche streaming services sit at the thin end of the panel. Their numbers move around, and a week-on-week change may be sample variation rather than a real shift in behaviour. Say that out loud when someone presents a dramatic rise.

Confidence intervals also affect how you compare services. Two channels whose ranges overlap are not clearly different, however neat the ranking looks in a spreadsheet.

For a fuller treatment of how to design the measurement around these limits, the argument that measurement needs a plan before launch is worth reading alongside the Barb documentation.

Reading a confidence interval in a planning meeting

Ask three questions. What is the estimate? What is the range around it? How does that range compare with the decision threshold?

If the range straddles the threshold, the data cannot settle the argument. Buy the plan on judgement, and set up measurement that will settle it next time.

Panel hygiene and why it is published

Barb publishes its methodology because the currency only works if buyers and sellers trust the instrument. Panel recruitment, meter technology and quality controls are documented so that a planner can defend a number under challenge.

That transparency is a competitive advantage for UK planners. In markets without a joint currency, every platform supplies its own numbers and reconciliation becomes a negotiation. Here, there is a reference point.

Joint industry currencies and how BARB fits alongside other data

A joint industry currency is a measurement system owned and governed by the buyers and sellers who trade on it. Barb explains the model in its note on Joint Industry Currencies, and the principle is the same across UK media.

RAJAR does it for radio, JICMAIL for mail, and Barb for television. Each exists because a single party owning the numbers creates an obvious conflict, and because a shared number makes trading faster and cheaper.

The IPA, ISBA, Thinkbox and Radiocentre all sit in this world as trade bodies and marketing bodies. They do not produce the currency, but they shape how it is used and explained to advertisers.

For a connected TV plan, the important point is that Barb is the television currency, not the campaign currency. Platform-reported impressions, your own ad server data and first-party conversion data all sit outside it.

Those sources are not competitors to Barb. They answer different questions, and the skill is knowing which question each one can settle. Barb tells you who was watching and for how long. Your ad server tells you what was served. Your CRM tells you what happened next.

In practice, a UK planner reconciles three layers: the Barb viewing layer, the platform delivery layer, and the advertiser outcome layer. Getting them to agree is a planning discipline, not a data purchase.

Turning BARB data into a connected TV media plan

This is the section that earns its keep. The following steps take you from Barb metrics to a schedule you can defend.

Step seven is the one most often skipped, and it is the one that decides whether the post-campaign meeting is calm. If the reporting frame is set after the campaign, everyone argues about definitions instead of results.

A worked example

Suppose a brief asks for reach among 25 to 54 adults in the West Midlands and Yorkshire and the Humber, with a modest budget and a two-week window.

For illustration, suppose Barb reports a four-week reach of 38 per cent for the target audience in those two regions combined, with a share of viewing of 14 per cent for the services you are considering. 5 percentage points at the 95 per cent level.

These are illustrative figures, not actual Barb data; pull the real numbers from your Barb licence for the specific period.

You pull Barb reach and share for the relevant services in those two regions, build a reach curve, and find that a narrow buy in high-rating programming delivers strong frequency but flattens the reach curve early. Widening to lighter-rated services adds incremental reach at lower average minutes.

You then set a frequency cap that preserves the incremental reach, state a confidence range around the forecast, and write down the conversion assumptions. The plan now has a defensible shape, not just a spreadsheet.

What to check before you send it

Where BARB numbers stop and your own measurement starts

Barb stops at viewing. It will not tell you whether your advert was in the break, whether it was watched, or whether anyone remembered it.

That boundary is not a flaw, it is a division of labour. The currency measures the audience; the buyer measures the campaign.

So your own measurement has to cover delivery verification, viewability where relevant, brand lift or attention if the budget supports it, and any conversion outcome the advertiser cares about. None of that comes from Barb.

Be explicit about the join. If your Barb-derived reach forecast says one thing and the platform dashboard says another, the difference is usually definitional: different audience, different device coverage, different de-duplication rules.

When you take key metrics to a board, lead with the Barb-based audience story and keep platform metrics as supporting delivery evidence. Boards understand reach and share; they distrust a dashboard screenshot with no denominator.

There is also a governance point. If you are combining Barb data with first-party data, the combination is subject to UK GDPR and PECR, and the ICO expects you to be able to explain the lawful basis and the retention rules. That is a planning constraint, not an afterthought.

Publishing BARB figures: the guidelines buyers should respect

The BARB publishing guidelines set out how viewing data may be cited publicly, including by journalists and by agencies publishing case studies.

The core discipline is attribution. Name Barb as the source, state the period covered, and give the audience definition. Do not present a share as a reach or a programme figure as a campaign result.

The guidelines permit headline figures for a service or campaign over a stated period. Reproducing a full week's ratings for all channels, or a long run of weeks, requires a licence. The exact limits depend on your licence terms and the purpose of the publication.

A selective figure quoted without its period and definition misleads even when the number itself is correct. For agencies, this matters commercially. A case study that cites Barb properly survives scrutiny. One that does not invites a correction and damages the number everyone else relies on.

If you need data for planning rather than publication, the route is a licence rather than a press citation, and How can I get Barb data? explains the options available to agencies and advertisers.

Respecting the guidelines is also a self-interested act. The currency is only useful while buyers and sellers accept it, and every sloppy citation erodes that acceptance a little.

Common questions