It’s no secret that the digital media ecosystem is unwell. Pervasive misaligned incentives are like a human body with a compromised immune system struggling to fight off a malignant cancer. In this case, the cancer is fraud.
Fraud is a broad term that encompasses a myriad of bad behavior, from “spoof” sites masquerading as legitimate, to publishers using bots to generate false impressions. The reason such fraud continues to occur (and thrive) is because the incentives in the ad tech ecosystem are aligned to support it.
The simple version of the ad tech supply chain consists of brands who serve ads to consumers who are visiting publishers’ sites; however, the reality of the programmatic landscape is complicated by dozens of intermediaries handling ad data throughout the supply chain. What’s more, everyone from the brand’s marketing director to the services responsible for tracking the results of a campaign are evaluated based on the number of impressions, or “views,” generated by the ads. Problematically, impressions can be easily faked and so measuring success on this basis becomes incredibly inaccurate.
Meanwhile, no one is incentivized to do anything to fix the issue. The services used to track the results of the campaign and the brand’s head of marketing all get to report high numbers of impressions to their superiors. Obviously, brand marketers can’t be faulted for the dysfunction permeating the system. It’s too big of an issue for the blame to land on any one player, but that also means it’s going to take turning the present incentives model on its head to cure the cancer of ad fraud once and for all.
If the ecosystem were configured in a way to reward accurate reporting over quantity of impressions, we’d be well on our way. But how do we achieve that?
Combining forces: AI meets blockchain
Artificial intelligence (AI) and blockchain technology have both been hyped as world changing innovations. Each by itself is certainly transformative, but when combined, truly revolutionary results can be achieved. AI is most effective when applied to extremely large data sets. A dataset of a hundred data points could be easily digested and analyzed by a person. But if we expand that out to millions of data points, it’s helpful to allow machine learning-based algorithms (a type of AI) to do the work of ingesting the data, recognizing patterns, and displaying results for human evaluation. The permanent nature of blockchains means that scientists won’t need to make as many assumptions when examining the results of an AI algorithm applied to data recorded on a distributed ledger. And fewer assumptions ultimately lead to more efficient output.
So how might this be applied to fight fraud?
By using AI algorithms to identify the source of fraud, we can leverage blockchain technology to track it, and successfully keep fraud out of the supply chain. Brands would be able to recoup the $19 million lost annually to fraud, publishers would not lose traffic to “spoof” sites, and consumers would benefit from viewing creative from sources that are who they claim to be. We know this type of system is possible, because it is already being applied to finance where blockchain and AI are being leveraged to detect and signal suspicious banking transactions.
If successful, this realignment of incentives will manifest in a public utility accessible to brands, publishers, and users alike. Anyone would be able to check the ledger to see if a certain domain is using bots to create false impressions or if a specific site claiming to publish news is actually disseminating unsubstantiated information for personal or political gain. Once such a utility is created, and the malicious actors are revealed, brands and publishers will mandate the shift to a new, decentralized system. The force of collective action is what will ultimately revolutionize the ad tech ecosystem for good.