Ad Network

Ad Network is the broker layer between advertisers (people who want to buy ads) and publishers (people who own the screens those ads appear on). Google Display Network, Meta Audience Network, Taboola, Outbrain - all ad networks, all doing the same job: pooling inventory across thousands of sites and apps so you can buy across them in one place.

Without them, advertising on the open web would mean negotiating with every publisher individually. The network exists because that doesn’t scale.

How an ad network actually makes money

Three layers of margin:

The advertiser pays a CPM (cost per thousand impressions), a CPC (cost per click), or a CPA (cost per action) depending on the campaign type.

The publisher gets a share of that - usually 60-70% on the bigger networks. AdSense, for example, pays publishers 68% of revenue on display.

The network keeps the difference, plus charges for any premium features - better targeting data, fraud filtering, brand safety controls.

The model is volume-led. A network with a million publishers and ten million advertisers makes pennies on each transaction at scale that makes those pennies into a real business.

The three flavours of ad network

Vertical networks specialise in one category - finance, gaming, parenting. Higher CPMs, narrower reach, useful when you want concentration.

Horizontal networks pool everything they can find. Google Display Network reaches roughly 90% of internet users. Cheap CPMs, wide reach, but uneven audience quality.

Premium networks handpick publishers - only inventory from established sites with real editorial standards. Higher CPMs, better brand safety, smaller reach. Ideal when your brand can’t risk appearing next to scraped content.

What’s changed

The line between “ad network” and “demand-side platform” (DSP) has blurred almost completely. Modern networks bid programmatically and run real-time auctions. The label is mostly a marketing choice now.

Bigger shift: third-party cookies dying. Networks built on cross-site behavioural targeting are losing their main lever. The ones investing in contextual targeting and first-party data partnerships are the ones still relevant in 2027.

An example

A B2B SaaS sells project management software for marketing agencies. They’ve maxed out branded search and want to scale awareness without burning through budget on broad keyword bidding.

They run a six-week test split across three networks: Google Display Network for reach (CPM $4.20, 12M impressions), LinkedIn Audience Network for B2B targeting (CPM $14.50, 280k impressions), and a vertical network focused on marketing-industry publishers (CPM $9.80, 1.1M impressions).

Result by branded search lift after the campaign: GDN +18%, LinkedIn +9%, vertical network +27%. The vertical network’s narrower but more relevant audience moved branded search the most per dollar - the broader cheap reach made noise but didn’t shift behaviour for an audience that was already saturated with display.

Where ad networks fall down

Fraud is the structural problem. Bot traffic, click farms, ads served on inventory the network technically didn’t authorise - every network has this issue, the bigger ones at industrial scale. Verification tools (DoubleVerify, IAS, Moat) exist precisely because you can’t trust the network to grade its own homework.

The other issue: brand safety. Programmatic placement means your ad can land next to anything the algorithm decides is “contextually relevant.” Categorical exclusion lists are mandatory hygiene now, not optional.

Related terms

• AdSense - Google’s publisher-side product, the other half of the GDN equation
• Google Ads - the buy-side platform that plugs into the largest ad network
• Banner Ad - the format ad networks were originally built to distribute
• Affiliate Network - the performance-only cousin built on the same broker model
• Retargeting - the targeting tactic ad networks made cheap and easy