Media looks like a content business and is actually a trust business with a monetisation problem attached. That inversion explains most failures to evaluate the sector properly: investors who would never buy a restaurant for its menu buy a publisher for its articles, when the articles are the visible surface of something the balance sheet cannot show, the audience's habit of coming back without being fetched. The examination below runs in the order that surfaces the truth fastest.

Map the revenue, then interrogate its concentration

The examination begins with the revenue table, and its most important column is the one that shows what the business depends on. Three concentration tests:

Platform concentration: what share of traffic or revenue depends on algorithms the company does not control, social feeds, search rankings, app store policies. A change in one line of another company's code can remove years of audience building, which the industry has now lived through several times.

Client concentration: what share of advertising revenue comes from one category, one agency relationship, or a handful of accounts. The loss of a single anchor advertiser is the classic mid-size media failure.

Format concentration: what share rides on one product, one newsletter, one show, one franchise. Depth of one is strength in media only when the franchise is genuinely irreplaceable, which is rarer than its owners believe.

Read the audience data like an auditor

Audience numbers are the most freely inflated figures in the industry, so the evaluation applies the same tests an auditor would. Direct traffic share, the readers who type the address or open the app, measures owned audience and cannot be bought at scale. Return frequency measures habit, and habit is the asset. Subscriber economics, acquisition cost against lifetime value and renewal rate, decide whether a subscription model compounds or churns itself into a treadmill. Where the numbers are self reported, verify against the audited sources that exist for exactly this purpose, and where they cannot be verified, treat them as marketing until proven otherwise.

A useful heuristic from the industry's own history: businesses that publish detailed retention data are usually proud of it, and businesses that publish only growth data usually have something to grow past.

Understand what changed, permanently, in this decade

Two structural shifts rewritten into the risk model:

The referral collapse. For two decades, publishers accumulated audiences from search and social referrals. Answer engines and platform strategy have inverted the flow: systems summarise the content and keep the reader. The consequences compound quietly, display advertising priced off declining traffic, and the strategic premium moving to owned channels: newsletters, apps, live events, licensed data, direct subscriptions.

The AI licensing question. Content that trains and grounds commercial models now has potential licensing value, and the deals being struck establish precedents that will split the industry: those with distinctive, verifiable, high-cost journalism to license, and those whose commodity output no model needs. Ask which side of that line a target sits on, and whether any announced agreements are exclusive, recurring or one-off.

Value the moat, not the metrics

Durable media franchises hold one or more of a short list of defensible positions, and the evaluation names them explicitly. A brand that is the cited authority in a valuable niche, the name that appears whenever the subject reaches a decision maker. Proprietary data or access that competitors cannot replicate, the terminal products, the essential datasets, the sources that talk to no one else. A physical habit with high switching costs, the daily brief embedded in a routine. Or community, the audience that comes for each other rather than only for the content, which is the hardest moat to build and the hardest to dislodge.

Franchises with none of these compete on price, and price competition in media is a race toward free.

The questions that separate franchise from newsletter

Beyond the numbers, the interview questions that reliably separate durable businesses from expensive content habits: What share of readers would notice within a week if you stopped publishing? What did retention look like for each cohort of the last two years? Which single decision by another company would hurt most, and what is the plan for it? What does a subscriber cost to acquire, and how many months of revenue repays it? If AI summarised your best work perfectly, what would readers still come to you for?

The answers, and the comfort of the people giving them, tell an investor more than any deck. Media investing rewards the same discipline every sector does: find the asset that compounds quietly, check that the numbers survive an audit, and pay for durability rather than for the launch. The industry's history punishes the opposite purchase with remarkable consistency.