Video Business Models

How Video Ad Auctions and Targeting Work

A clear look at how ads are chosen, priced and matched to viewers alongside online video, including auctions, bids, targeting signals and measurement.

On this page
  1. The problem the auction solves
  2. The main players
  3. What happens in one ad request
  4. How the winner is chosen and priced
  5. Direct deals alongside auctions
  6. What targeting means
  7. Privacy and its effect on targeting
  8. Measurement and quality checks
  9. What this means for different people

The problem the auction solves

Every time a video is about to play with an ad, someone has to decide which ad to show and what it is worth. The space might be a pre-roll before a video, a break in the middle or a slot in a live-style channel. There are many advertisers who might want it, and their interest differs from one viewer and one moment to the next. Fixing a single price for all of them would leave money on the table or leave slots unsold.

An auction solves this by letting demand set the price at the moment the slot appears. The process is automated and typically takes a small fraction of a second, so the viewer notices nothing except that an ad loads.

The main players

Several kinds of companies are usually involved, and the terms are worth knowing:

  • The publisher or platform owns the place where the ad will appear and wants to earn as much as it reasonably can.
  • The advertiser, or an agency acting for them, wants to reach certain people at an acceptable cost.
  • A supply-side system acts for the publisher, offering the slot to buyers.
  • A demand-side system acts for advertisers, deciding whether to bid and how much.
  • An ad exchange connects the two and runs the auction.

Large platforms often run much of this themselves within their own systems. Others use outside tools. The overall logic is similar either way.

What happens in one ad request

A simplified sequence looks like this:

  1. A viewer opens a video and the player asks for an ad.
  2. The request includes information about the opportunity: the video’s topic or category, the device, the approximate location and the ad format and length allowed. What else is included depends on the platform, privacy settings and regional rules.
  3. Buyers receive the request and decide whether the opportunity matches a campaign.
  4. Interested buyers submit bids, usually a price they will pay for the impression.
  5. The auction picks a winner according to its rules.
  6. The winning ad is delivered to the player and shown.
  7. The system records whether the ad was shown, whether it was watched and sometimes whether the viewer clicked or acted afterward.

How the winner is chosen and priced

The highest bid does not always win in a simple way, and the price paid is not always the bid. Auction designs differ. Two broad types are often described:

  • In a first-price auction, the winner pays what they bid.
  • In a second-price auction, the winner pays just above the next-highest bid.

Which type a given system uses, and what other rules apply, varies. Many include a floor price, a minimum the publisher will accept. Some also weigh factors beyond the bid, such as the expected quality of the ad or how likely the viewer is to watch it, so that a slightly lower bid can win if the ad is expected to perform better. These details are set by each platform and change over time.

Prices are commonly quoted per thousand impressions. For video, there are also models tied to viewing, where a buyer pays only if the ad is watched for a minimum time. Different pricing bases suit different campaign goals, such as awareness or direct response.

Direct deals alongside auctions

Not every ad is sold in an open auction. Some are bought through direct agreements: an advertiser and a publisher agree on a price and volume in advance. Others sit between the two, with a private marketplace where invited buyers compete for a defined set of inventory. Premium or live content, such as big sports broadcasting, is often sold at least partly in advance, because advertisers want certainty that their ad appears there.

What targeting means

Targeting is the process of matching an ad to the people most likely to be relevant to the advertiser. Common approaches include:

  • Contextual targeting, which matches ads to the subject of the video, such as showing a kitchen product beside a cooking video. It does not require information about the individual viewer.
  • Audience-based targeting, which uses information about the viewer, such as inferred interests, past behavior or broad demographics.
  • Location and device targeting, which uses where the viewer is or what they are watching on, such as connected TVs or phones.
  • Time-based targeting, which concentrates ads in certain hours or days.

Because the viewer is more likely to be valuable to some advertisers than others, targeting changes the bids coming in and therefore the price. This is one reason the same video can earn different amounts per view in different places.

Privacy and its effect on targeting

Targeting depends on data, and the rules about data have been tightening in many regions. Laws and platform policies differ by country, and viewers can often change settings that limit tracking. This has led to a stronger interest in contextual methods and in information that a platform holds about its own users. The details are complicated and this is not legal advice; anyone running ads or collecting data should check the rules that apply to them.

Measurement and quality checks

Advertisers want proof that their money was well spent. Measurement usually looks at whether the ad was delivered, whether it was viewable on screen, how much of it was watched and what happened afterward. Some of these are tracked with standardized definitions, and others vary by company, so numbers from different sources do not always match. Marketers who study marketing metrics tend to pay attention to exactly what is being counted.

There are also checks against fraud and unsuitable placement. Advertisers may use lists of topics or channels where they do not want to appear, which can affect which videos attract higher bids. For creators, this can matter, as described in the discussion of creator monetization.

What this means for different people

For viewers: The ad you see is the result of matching and bidding, not a random choice. Settings may let you reduce personalization.

For creators: Revenue per view is shaped by who watches, where, and how much advertisers want to reach them, plus platform rules. Understanding this explains why income can change even when views stay the same. Reviewing video analytics for audience details can help.

For advertisers: Cost and results depend on targeting choices, the type of auction and the format of the ad.

The mechanics are well established, but specific rules, formats and privacy conditions keep changing, so current documentation from the relevant platform is the reliable source for details.

The complete guide to video business models

How Streaming Services Make Money
Read the guide

Keep exploring

Guides from other topics

All topics →