Ad fraud cost the advertising industry $7.2 billion in 2016 and takes $1 for every $3 dollars spent on digital ads1. Here at Adara we have a fraud rate of less than 1%2, far below the industry average. We strive everyday to reduce this further and further, but how do we do it?
We use pre-bid and post-bid strategies, along with our secret ingredient, our data science team.
Pre-Bid
It all starts before we ever deliver an impression. Let’s imagine you’re buying some new shoes online, to make things easier you filter out what you don’t want. For example, that might be size, color or style. We do just the same thing for the sites we serve ads on. We work with Integral AD Science (IAS) to filter out the sites and placements that don’t meet our high standards. However, instead of color, style and size of shoe, we filter for brand safety, media quality and viewability.
Brand Safety
The brand safety segments protect brands from being displayed alongside potentially offensive or damaging content. Here are a few examples of the things we can filter out:
- Adult sites
- Alcohol
- Illegal downloads
- Drugs
- Hate speech
- Offensive language
- Violence
- Unrateable content
Media Quality
The media quality segment allows us to filter out sites with suspicious activity from bots and adware. A fraudulent site may use bots to increase the number of impressions they can serve, therefore increasing their revenue. Adware can be used to automatically download pop-ups, banners or viruses from fraudulent websites and then attack the user’s device.
Viewability
Viewability is a measure, usually presented as a percentage, that tracks the impressions that can be seen by the user. Excluding impressions that might be off screen, below the fold or hidden by another ad.
Post-Bid Integrations
Post-bid integrations, as you may have guessed, occur after the ad has been served. At this stage we use tools and processes to ensure the ad was served inline with the filters we set out in pre-bid, and that it was correctly viewed by a real user. We use both IAS and DoubleVerify to confirm that what the exchanges are telling us, actually happened. Each ad we serve either has an IAS or DoubleVerify pixel that allows us to track the following:
- Transparency: whether the site that the bid was placed on is actually where the ad ended up appearing
- Suspicious activity: impressions on the ads are being generated by actual humans, not through bots or adware
- Viewability: the ad was viewable in the browser screen, not below the fold or stacked behind another ad
Additional Steps
Aside from pre-bid and post bid strategies we also employ a couple more steps to achieve a 1% fraud rate.
- Malware Scanning – Advertising requires many different parties and an ad can be fraudulently altered at any step of the process. Adara’s partnership with The Media Trust provides constant monitoring to ensure that Adara’s ad tags have not been altered for any malware or malvertising.
- Ads.txt – an initiative by the Interactive Advertising Bureau (IAB) aims to help publishers prevent unauthorized selling of their inventory. A publisher, let’s say The New York Times, adds a small text file to it’s top level domain (i.e. nytimes.com/ads.txt) that lists all the companies authorized sell that inventory. If the company you’re buying off isn’t on that list, they are fraudulently selling the inventory. Similarly, and less manually, ad exchanges can integrate Ads.txt files to confirm the inventory is being sold legitimately.
What is the Adara difference?
Now, you may be wondering how we manage an industry beating 1% fraud rate if we only use tools that everyone has access to?
The Adara difference lies in our data science team, and they are awesome. Adara’s Data Science team uses data from IAS and DoubleVerify to curate and update our own internal blacklists of websites. We have built up data and knowledge on inventory across the world wide web, allowing us to prioritize bids only on sites that present the very lowest levels of fraud. This feedback loop of iteration, both manual and automated, coupled with our data ecosystem of real people doing real transactions on very reputable sites sets us apart from the industry, which has an average fraud rate of between 5% and 30%.
Why is this important?
Setting aside the $6.5 billion the industry could save, huge enterprises trust Adara with their advertising. Ensuring brands appear in a safe environment, and appearing correctly, is crucial to all involved.
Trust is key to everything we do at Adara, we work hard to achieve such a low fraud rate and are continually striving to reduce it further.
1 Invesp. https://www.invespcro.com/blog/online-ad-fraud-statistics/
2 based on IAS data