Audience personas in media planning: from strategy to segments

Why most personas are decoration

The standard persona exercise produces 'Marketing Mary, 34, likes yoga and podcasts', consumes two days, and then the campaign targets adults 25-44 with interest stacks anyway - because nobody translated Mary into anything a platform could buy. The persona lived in the strategy section; the targeting lived in the platform; the two never met. The waste is not the two days. It is that the audience insight, the part of planning most likely to move performance, never reached the auction.

What a useful persona contains

Three layers, and the third is the one usually missing.

1. Motivation. Why this person would respond: the problem they have, the trigger that puts them in market, the objection that stalls them. This layer shapes creative and messaging.
2. Channel behaviour. Where they can actually be reached, and in what mindset - scrolling, searching, watching. This layer shapes the channel mix.
3. Addressable proxy. The specific targetable segments that approximate them on each platform: the interest and behaviour combinations, the in-market categories, the job-title and seniority filters, the lookalike seeds. This layer is what makes the first two buyable.

A persona with layers one and two is a strategy artefact. Add layer three and it becomes a media planning instrument - the audience column of the plan, priced and reachable.

Building personas from the brief, not from a template

Generic persona templates produce generic personas. The raw material should be the brief and the strategy: the market, the objective, the product's actual buyer versus its aspirational one. Draft from there, then pressure-test each persona against layer three - can it be bought? This drafting-and-matching step is exactly the part AI has made fast: Medusa generates personas from the extracted brief and matches each one to real targetable segments on Meta, Google, TikTok and LinkedIn, which turns the two-day exercise into a review task.

Translating persona traits into platform segments

The translation is per-platform, because each platform speaks a different targeting language. The same trait becomes a different buy on each.

Notice the honest cells: some traits have no good proxy on some platforms. Writing 'deprioritise' into the plan is a finding, not a failure - it feeds straight into the channel allocation, because a channel that cannot address the audience is expensive at any CPM.

Sizing and overlap: the checks that prevent surprises

• Size each mapped segment before committing budget. A persona whose platform proxy is 40,000 people cannot carry a broad-reach line; one that maps to 30 million is not a targeting decision at all.
• Check overlap between personas. Two personas whose segments overlap 70% will bid against each other and double-count reach. Merge them or sharpen the definitions.
• Match the persona to the funnel stage. Prospecting personas and retargeting pools are different objects; putting conversion budget behind a cold prospecting persona misreads what the segment knows about you.

Personas and measurement

If the plan's audience column names the persona, report by it. Performance by persona - not just by channel - is where audience learning actually accrues: which motivations respond, which proxies were too loose, which segment fatigued first. That feedback is what sharpens the next plan, and it only exists if the persona survived from strategy into the campaign structure and back out into reporting.

Keep a library, not a folder of one-offs

Audience knowledge should compound. A persona built, matched, run and measured is an asset - for the next campaign in the category, the next client in the vertical, the annual planning cycle. Keeping a reusable persona library with the platform mappings attached means repeat work starts from evidence rather than a blank page, which for agencies is also how audience quality stops depending on which planner drew the brief. B2B teams will find the same logic applied to professional segments in the LinkedIn integration.