Traffic quality

Not all traffic is equal. Traffic quality asks whether visitors are real people, whether ads shown to them could be seen, whether they paid attention, and whether they ever do anything useful afterwards. It is broader than invalid traffic: a source can be entirely human and still be poor quality, such as accidental clicks from an interstitial ad or visitors lured by incentivized traffic.

Common ingredients include IVT rate (GIVT and SIVT), viewability and time in view, attention metrics, bounce and scroll behaviour, brand safety of the environment, ad density and refresh rates (see made-for-advertising sites), supply-path transparency via ads.txt and sellers.json, and outcomes such as conversions, retention and incrementality.

Different companies package these into scores. Publishers use traffic quality to vet traffic vendors and prove value to buyers; advertisers use it to prune sites, apps and sellers. A sensible approach is to judge quality per source and to look for consistency between metrics: high clicks with zero conversions, or high impressions with low viewability, usually signal a problem.

Think of it like this

Traffic quality is like judging a party not by how many people came through the door, but by how many actually knew the host, stayed and enjoyed it.

An example

Two ad networks each deliver 100,000 visits to an Argentine retailer. Network A: 1% IVT, 70% viewability, 1.8% conversion. Network B: 9% IVT, 35% viewability, 0.1% conversion. Same volume, very different quality.