Why a programmatic advertising target audience needs a written profile

What to take away

The IAB's 2026 Outlook StudySeptember Update gives planners forward-looking digital ad forecasts, so audience definitions can be tied to expected market conditions rather than last year's numbers.
A target audience is a written profilewho they are, what they do, where they can be reached and what you will exclude.
Every segment needs a named data source, a lawful basis and a retention period before it goes live.
Test segments in a small campaign first, then scale what meets your cost-per-acquisition threshold.
Review the profile quarterly, because platform targeting options and privacy rules both change.

Start with the commercial question

Before opening any ad platform, write down what the campaign must achieve. A retailer may need first-time buyers within 20 miles of a store. A software firm may need decision-makers at companies of a certain size. The audience definition follows from that, not the other way round.

Audience definition by campaign goal

Retailer

Goal: First-time buyers
Geography: Within 20 miles
Company size: Not specified
Audience follows: Store proximity

Software firm

Goal: Decision-makers
Geography: Not specified
Company size: Certain size
Audience follows: Job role

Put the answer in one sentence a colleague could repeat. For example: adults in England who bought gardening equipment online in the past six months and have not purchased from US. That sentence sets the inclusion rules and the exclusions at the same time.

Gather data you already hold

Your own systems are usually the richest source. Look at purchase history, enquiry forms, newsletter engagement and on-site search terms. These tell you what real customers do rather than what a survey says they might do.

First-party data sources to mine

Purchase history
Enquiry forms
Newsletter engagement
On-site search terms
Own survey for attitudinal data
Third-party provider documentation

Where you need attitudinal data, run your own survey. Our guide to programmatic advertising survey tools in England covers the sampling and weighting decisions that keep results usable.

For third-party data, read the provider's documentation on how each segment is built. If the provider cannot explain the source, do not buy it.

Map the audience onto platform controls

Each platform exposes different controls. LinkedIn's ad targeting options let you combine job function, seniority, company size and industry, which suits business-to-business campaigns where the buyer's role matters more than their postcode.

Consumer platforms lean on interests, behaviours and location. Write down which control matches each part of your profile, and note any part you cannot address. Gaps are useful: they tell you where the plan is optimistic.

Keep the number of stacked filters small. Every extra condition shrinks reach and raises the chance the segment is too thin to learn from.

Check the privacy and transfer rules

Programmatic buying moves data between platforms, often across borders. The Information Commissioner's Office explains the safeguards required for international transfers under UK GDPR, which matters whenever a demand-side platform or data partner sits outside the UK.

Record the lawful basis for each segment. Consent is common for marketing cookies; legitimate interests may cover some analytics. Document the retention period and who can access the data.

If a segment depends on sensitive information, such as health or political views, treat it as out of scope unless you have explicit consent and a clear need.

Build a segmentation test plan

Run two or three candidate segments against the same creative and budget. Hold the bid strategy constant so the audience is the only variable that changes.

Set a success measure before launch. Cost per acquisition, cost per qualified lead or return on ad spend all work, provided the finance team agrees the definition. Review after enough conversions to be meaningful, not after a fixed number of days.

For a worked method of structuring these tests, see our audience research guide, which sets out how segmentation, measurement and reporting fit together.

Write the profile down

A one-page profile keeps buyers, planners and legal reviewers aligned. It should name the segment, the source data, the platform controls, the exclusions, the lawful basis and the review date.

Version it. When someone changes a filter in the platform, the profile should change too. Otherwise the document drifts away from what is actually running.

Checklist before launch

One-sentence audience definition agreed with the client or stakeholder.
Each segment traced to a named data source.
Lawful basis recorded for every data type used.
Cross-border transfers checked against ICO guidance.
Platform controls mapped, with unmapped gaps noted.
Exclusions listed, including existing customers where relevant.
Success measure and review date set.
Test budget approved separately from the main campaign.
Retention period and access list documented.
Profile saved in the campaign folder and dated.

Common questions

How many segments should a first campaign use?

Two or three. More than that makes it hard to tell which segment performed and stretches the test budget too thinly.

Can we reuse a segment built for another client?

Only if the lawful basis covers the new use and the data was collected for a compatible purpose. Check before copying anything across accounts.

How often should the profile be reviewed?

Quarterly is a sensible default, plus an immediate review after any platform policy change or regulator guidance update.

What if the audience is too small to spend the budget?

Loosen the least important filter first, usually recency or a secondary interest, and record the change in the profile.