Tech, ExplainedGaming, streaming and media
Product

Ad verification

Was the ad seen, by a human, somewhere safe?

BEFORE ITAFTER ITAd verificationDoubleVerifyMonetizeMoreCOVERSViewabilityBot detectionBrand safetyPre-bid checksAd exchangeoffers eachimpressionSupply-side platform (SSP)serves the adServer-side adinsertion (SSAI)↗ VIDEO STREAMINGreportsstreamad playsblocks riskyad spaceDemand-side platform (DSP)CONCEPTS IT USESAd fraudWalled gardensBEFORE ITAFTER ITAd verificationDoubleVerifyMonetizeMoreCOVERSViewabilityBot detectionBrand safetyPre-bid checksAd exchangeoffers eachimpressionSupply-sideplatform (SSP)serves the adServer-side adinsertion (SSAI)↗ VIDEO STREAMINGreportsstreamad playsblocks riskyad spaceDemand-sideplatform (DSP)CONCEPTS IT USESAd fraudWalled gardensBEFORE ITAd exchange↓ offers each impressionSupply-side platform (SSP)↓ serves the adServer-side ad insertion (SSAI)↗ VIDEO STREAMING↓ reports stream ad playsAd verificationDoubleVerifyMonetizeMoreViewabilityBot detectionBrand safetyPre-bid checksAFTER IT↓ blocks risky ad spaceDemand-side platform (DSP)CONCEPTS IT USESAd fraudWalled gardens
Ad verification receives each impression from the ad exchange, the supply-side platform and stream ad insertion, and reports bot traffic and unsafe pages back to the advertiser's demand-side platform.
opens a page inside this one a page in the same scene a page in another scene a concept page a label only, with no page

Ad verification is software that checks, for an advertiser, that each ad actually appeared on a screen, was seen by a person rather than a bot, and did not appear beside content that harms the brand. It usually includes brand safety checks, and it is sometimes called third-party ad verification. Advertisers buy it from an independent company, not from the company that sold them the ad space.

Why is this needed? In programmatic advertising, software buys each ad space in an auction in under a second. No person at the brand sees where each ad will appear. So three things can go wrong.

First, the ad may not be visible. It may load at the bottom of a long page that the reader never scrolls to. The industry standard counts a display ad as viewable only if at least half of it is on screen for at least one second. For a video ad, the minimum is two seconds. An ad that meets this test is called a viewable impression.

Second, the viewer may not be human. Some websites and apps use bots to create fake visits and fake clicks, and collect the advertiser's money for them. This is ad fraud. Verification software looks for signs of a bot, such as thousands of visits from one data centre, or clicks that arrive at inhuman speed.

Third, the page may be unsuitable. A family car's ad should not appear beside a video of a road accident. This check is called brand safety.

Take, for example, a paint company in Ahmedabad that buys 10 lakh video ad impressions for its monsoon campaign. An impression is one showing of one ad to one person. The verification company's small piece of code runs inside each ad as it plays. At the end of the week, its report shows that 1.2 lakh impressions came from bots. Another 2.5 lakh played in a corner of the screen for less than two seconds. Around 8,000 played beside news of a building collapse. The paint company asks for a refund on the fraudulent impressions. It also adds the worst websites to a block list.

Verification also works before the purchase. The advertiser's demand-side platform (DSP) can ask the verification company about each ad space before it bids. If the space looks risky, the DSP does not bid at all.

Verification has limits. Inside walled gardens, the large platforms that sell ads on their own sites, independent code is often not allowed to run. There, the platform shares only the measurements that it chooses to share. How is verification different from attribution? Verification asks whether the ad was seen. Attribution asks whether it led to a sale.

Tech, Explained is a part of ingrid.fyi. Every page explains one piece of technology in plain words.