Seven steps to define a connected TV advertising target audience

What to take away

Most teams start with broad age and gender splits, then wonder why connected TV budgets underperform.
A useful connected TV advertising target audience is built from viewing behaviour, not demographics alone.
You need four inputsfirst-party data, platform reporting, published UK viewing data and a budget you can defend.
Seven steps below take you from raw data to a tested, documented audience definition.
Write the definition down before briefing anyone, so planners and buyers work from the same list.

Start with behaviour, not demographics

The common mistake is copying a linear TV audience and calling it a connected TV audience. Connected TV viewing is on demand, so reach builds differently and frequency behaves differently. A 35 to 44 split tells you almost nothing about who will watch a 30 second streaming ad to completion.

Instead, ask what people watch, when they watch and on which device. Barb publishes weekly top 50 shows covering UK TV programme popularity, which gives you a sense of the scale behind individual titles. Treat that as a starting point rather than an answer.

Turning that viewing data into segments a platform can actually buy takes more work. The connected TV advertising audience research guide sets out a full research design if you want to go further than a single screen.

Gather the four inputs

Step 1. Pull first-party data. Export customer lists, sign-up records and purchase history. Segment by recency and value, not just age. Before using personal data for targeting, check that you have an appropriate lawful basis and meet your transparency obligations under UK GDPR.

Four inputs to gather

Pull first-party data by recency and value
Check DSP completion rates and dayparts
Read published UK viewing data such as Barb
Set budget and realistic cost per thousand

Step 2. Check platform reporting. Your demand-side platform, such as DV360, The Trade Desk or Amazon DSP, can show results such as completion rates, device types and dayparts from campaigns run through it. Reporting and available inventory depend on the buying route, so check what the platform can actually measure. Even a small test gives you something to work from.

Step 3. Read published UK viewing data. Use BARB for programme viewing, and cross-check relevant audience context with Ofcom's Media Nations reports or IPA TouchPoints. Comscore and Nielsen are further audience-measurement sources; check each source's coverage and methodology before using it to size a CTV segment.

Step 4. Set the budget. Record the approved campaign ceiling, delivery period and how the budget will be split between segments. Ask each shortlisted seller for a written forecast of reach and frequency at that ceiling, then compare plans using the same assumptions; a general TV-planning guide is not a substitute for an inventory-specific proposal.

Define the audience in writing

Step 5. Write one sentence per segment. For example, adults aged 25 to 44 in English cities who stream drama series at least twice a week and have bought online in the past 90 days. Treat this as a testable planning hypothesis, then state which data source supports each attribute and how you will check whether it can be activated. Vague definitions cause wasted spend.

Step 6. Map each segment to actual inventory and confirm the available targeting with its seller. UK options include ITVX, Channel 4 streaming, My5, Sky AdSmart, Netflix, Amazon Prime Video and YouTube; targeting controls vary by service, format and buying route. Ask whether the proposed inventory can reach the segment by geography, viewing context or device rather than assuming every service offers every control. For programmatic buying, check eligible inventory in DV360, The Trade Desk or Amazon DSP. FreeWheel and Magnite are ad-tech providers in the video supply chain, not interchangeable audience definitions or substitutes for a DSP.

Step 7. Set a measurement plan before launch. Decide which metric matters: completed views, incremental reach or website visits.

Check the segments against people, not just data

Data alone will not tell you whether a segment is real. Ask colleagues in sales, customer service and product what they hear from customers, then compare those observations with the viewing and campaign data. That internal knowledge often explains anomalies in the numbers.

Use a short survey to test assumptions about viewing habits and attitudes. Google Forms or SurveyMonkey can collect responses; use the connected TV advertising survey tools in England page to plan the questionnaire and sampling, and do not treat a convenience sample as representative.

For audience targeting that uses personal data, check the Information Commissioner's Office (ICO) guidance and your obligations under UK GDPR. Confirm the lawful basis, transparency information and any data-sharing arrangements before activating a customer list or audience segment.

Set reach and frequency expectations

Connected TV rarely delivers the same reach as linear TV at the same spend. Plan for narrower, deeper audiences and accept that frequency will climb faster. Split your budget across two or three segments rather than one broad definition.

Review results after four weeks. If one segment delivers a higher completion rate and better downstream conversion, move budget towards it. Keep the losing segment documented so you do not repeat the test next quarter.

Common questions

How many segments should a first connected TV campaign use?

Two or three is manageable. More than that makes it hard to tell which segment performed and why.

Do I need first-party data to start?

No, but it improves targeting quickly. Platform reporting and published viewing data can carry an early test.

How often should the audience definition be reviewed?

Quarterly is sensible. Review sooner if completion rates, frequency or delivery shift sharply.

Can I reuse a linear TV audience definition?

Only as a starting point. Streaming behaviour, device mix and frequency differ enough to justify a fresh definition.