Why pricing models matter
A pricing model is more than a unit of account. It decides what the advertiser is paying for, who carries the risk that the media does not perform, and what the seller is motivated to optimise. When a buyer pays per impression, the seller is rewarded for delivering impressions, and the buyer carries the risk that those impressions do not lead anywhere. When the buyer pays per click, the seller takes on some of that risk and in return controls which impressions are shown to whom. When the buyer pays per action, the seller takes on more risk again, prices accordingly and gains a reason to find actions wherever they are cheapest, including actions that would have happened anyway.
None of these is better in general. Each fits some objectives and not others. A brand objective written as reach against an audience is naturally bought on impressions, because reach is what the advertiser wants. A performance objective written as a cost per lead might be bought on clicks or actions, provided the advertiser can verify the actions and judge whether they are incremental. The mistake is to choose a model for its apparent safety rather than for its fit with the objective. The media buying hub places pricing alongside the routes and the negotiation that complete a buy.
The models
The main models appear across digital and offline media in different forms. The table sets out what each charges for and where the risk sits.
| Model | The buyer pays for | Common in | Risk mostly carried by | Suits objectives that are |
|---|---|---|---|---|
| Cost per thousand impressions | Every thousand ads served or estimated exposures | Display, online video, connected television, audio, social, digital out-of-home | Buyer | Reach and frequency based |
| Cost per click | Each click on the ad | Paid search, some social and display | Shared | Traffic and consideration |
| Cost per action or acquisition | Each defined conversion | Affiliate, some performance deals | Seller (priced in) | Defined actions that can be verified |
| Cost per view or completed view | Each view meeting a defined threshold | Online video | Shared | Video consumption |
| Cost per rating point | Each point of audience delivered against a trading audience | Television, some radio | Buyer, with delivery guarantees | Broad reach |
| Per spot or insertion | Each airing or each print insertion | Radio, print, some television | Buyer | Frequency and presence |
| Per site or panel per period | A location for a fixed time | Out-of-home | Buyer | Local reach and presence |
| Fixed fee or sponsorship | An agreed package for a period | Sponsorships, partnerships, branded content | Buyer | Association and access |
| Revenue share | A share of sales generated | Affiliate, some retail media | Seller | Direct sales |
CPMCost per 1,000 impressions. is the default currency of impressions and the basis most plans use for comparison; the glossary entry for cost per thousand sets out the formula. CPCAverage cost of each click. and CPACost of each conversion. move the unit closer to an outcome. CPVCost of each video view. applies the same idea to video, with the definition of a view varying by platform and contract. The core metrics guide covers how these rates relate to each other in reporting. The cost per thousand, cost per click and cost per acquisition calculators convert between spend, volume and rate for any line. Retail media mixes several of these models in one proposal, and the guide to retail media sets out how it is bought and what to ask before accepting a retailer's package.
Guaranteed versus non-guaranteed
The second axis of pricing is whether the seller guarantees delivery. In a guaranteed buy, the seller commits to deliver a specified volume (impressions, ratings, spots) in specified placements over a specified period at an agreed price, and the contract sets out a remedy if it falls short. In a non-guaranteed buy, the advertiser bids or agrees a price but the seller commits to nothing; delivery depends on what inventory becomes available and whether the bid wins.
Guaranteed buys suit lines where specific placements or a fixed volume matter: a launch that needs a certain weight in a certain week, a sponsorship, a premium context. They usually cost more per impression because the seller carries the risk of holding inventory back. Non-guaranteed buys suit lines where flexibility and efficiency matter more than certainty: always-on display, performance lines, testing.
Between the two sit the programmatic deal types, which trade some certainty for some flexibility: fixed prices without guaranteed volume, or guaranteed volume bought through software. The programmatic deal types guide covers how those deals are structured and set up; for planning purposes, the question is only which lines need a guarantee and what the plan is prepared to pay for it.
Offline pricing: ratings, spots, sites and pages
Offline media have their own conventions, and they are worth understanding before converting them.
Television is traditionally traded on rating points: a rating point is a share of the trading audience watching at a given time, and the buyer agrees a cost per rating point for a campaign. Total rating points across the campaign roughly equal reach multiplied by average frequency, which is why television plans speak directly in reach and frequency terms. Because a rating pointA share of the trading audience watching at a given moment; television is bought by the point, and a schedule's total roughly equals reach times frequency. is a share of an audience rather than a count of people, a quote in points has to be converted into impacts before it can be set beside one in impressions. Prices vary by daypart, programme, season and how far ahead the booking is made, and broadcaster sales houses publish their own trading conventions. Connected television increasingly sells on impressions instead, which makes it easier to compare with online video but harder to combine with linear ratings.
Radio is often sold by spot, by package (a set of spots across dayparts) or on audience-based pricing where measurement allows. Out-of-homeAdvertising in places people pass on the way somewhere else, such as roadsides, stations, shelters and screens, bought by site, panel or screen. is sold by site or panel for a period, with audience estimates from the industry measurement system, and digital screens increasingly by share of time or by impression. Print is sold by page or fraction of a page per insertion, with rate cardsA seller's published price list for advertising space, airtime or creator work, treated by buyers as an opening position, not the price paid. that vary by position and frequency, and an audited circulation or readership estimate behind it. The guide to buying print and press advertising explains how to check which of the two stands behind a quote and convert it to cost per thousand target readers. Sponsorships are almost always a fixed fee for a package of rights and assets. Sponsored articles on publishers' websites are usually sold the same way, and the guide to sponsored articles and advertorials covers what to ask for in the package and how to judge it. For out-of-home, the guide to buying out-of-home advertising turns the site price and its audience estimate into a campaign, from choosing sites to checking proof of posting.
The point of knowing these conventions is to ask the right questions: which audience and measurement source the price is based on, whether it is guaranteed, and what is included.
Comparing buys on a common basis
A plan containing a cost per rating point, a cost per click, a site price and a sponsorship fee cannot be judged by eye. The buyer converts every line to a common basis before comparing, and the most useful basis is cost per thousand of the target audience.
The conversion has three steps. First, estimate the impressions or exposures each line delivers against the target audience, not the total audience: a broad television buy delivers a large total audience of which only part is the target. Second, divide the line's full cost (including fees, see below) by those target impressions and multiply by a thousand. Third, adjust for quality where it matters: an impression that is unlikely to be seen, or seen for a fraction of a second, is not equivalent to a full-screen video view with sound. Viewability and attention adjustments belong to the programmatic measurement guides; the planner applies them where the evidence supports it rather than inventing weights.
Outcome measures such as cost per acquisition and return on ad spend (ROASRevenue generated per unit of ad spend.) are essential for judging performance lines after they run, but they are a poor basis for comparing media before the campaign. They depend on creative, offer, landing page, timing and the work of every other channel, and they are not available in comparable form for broad-reach media whose effect is delayed.
A furniture retailer compares four quotes
Damaskwren Interiors, an invented furniture retailer, has four quotes for its target audience of homeowners aged 35 to 64. Radio: £24,000 for a package estimated to deliver 3.2 million target impressions, which is £7.50 per thousand. Out-of-home: £30,000 for panels with 2.5 million estimated target exposures, £12.00 per thousand. Online video: £18,000 at a quoted £20 per thousand total, but only 60% of the audience targeted is in the target group, making £33.33 per thousand target impressions. A magazine sponsorship: £15,000 fixed, reaching an estimated 400,000 target readers twice, £18.75 per thousand. On a target basis the video line looks least efficient, so the buyer asks for tighter targeting before signing.
For video, the CPM versus CPV comparison shows how to translate a cost per view into the same unit as an impression-based price.
Fees, commissions and what the price includes
A quoted price is rarely the whole cost. Before comparing, the buyer needs to know what sits inside and outside it.
Agency commission and fees. Some markets quote media gross of an agency commission and some net of it. Agencies may be paid by commission, by fixed fee, by time or with a performance element. Two quotes on different bases cannot be compared until they are converted.
Platform and technology fees. Programmatic buys carry demand-side platform fees, data costs, ad serving and verification charges, and sometimes fees on the sell side as well. The IAB has published work on supply chain transparency that sets out where these costs arise. A low media price with high technology fees can cost more than a higher media price with few fees.
Production and adaptation. Television, out-of-home and print each need production and adaptation that the media price does not cover, and some sponsorships require activation spending to be worth the fee.
Volume deals and rebates. Agencies and large advertisers may receive volume-based discounts or rebatesA volume-based discount or payment from media owners to an agency or large advertiser, which the agency agreement should assign and report. from media owners. The contract should say who benefits and how it is reported.
Taxes and surcharges. Media taxes and premium position surcharges vary by market and medium.
The working rule is to compare buys on full cost to the advertiser, on the same basis, against the same target audience. Anything less flatters some lines and penalises others. Before a direct buy is signed, the insertion order sign-off checklist confirms the order states the buying model and price unit that were actually negotiated.
For video billed per view, the impression calculator walk-through for video view campaigns shows how to turn a cost per view into impressions and paid views.
For programmatic buys, the impression calculator guide to programmatic deals builds an all-in viewable CPM so open auction and private deals can be compared on the same basis.
Common mistakes
- Choosing a model for apparent safety. Paying per action feels safe but hands control of targeting and placement to the seller, who prices the risk in.
- Comparing native rates directly. A cost per click and a cost per thousand measure different things. Convert first.
- Using total audience rather than target audience. A cheap rate against a broad audience can be an expensive rate against the people the plan is about.
- Ignoring fees. Platform, data, serving and verification fees can change the ranking of quotes.
- Mixing gross and net. Ask which basis every quote uses.
- Judging broad media on outcome rates. Return on ad spend is not a fair comparison for media whose job is reach.
- Paying for guarantees the line does not need. Guaranteed inventory costs more; reserve it for lines where certainty matters.
