Having invested years developing and optimizing ad targeting ML models [1][2] at one of the traditional web2 ad giants, I can strongly emphasize the pivotal role that vast amounts of data play in the success of ad campaign performance - point in case, why the likes of Google, Meta rule the advertising world. 

Have you come across the phrase “Data is the new gold”?; it is true! The amount of first party data amassed by these Big Tech companies put them at a pedestal that remains out of reach for its competitors. My curiosity to find solutions to break the monopoly in the advertising game led me to explore decentralization.

Unlike the centralized nature of web2 platforms - using blockchain data (being public and beyond the control of any single organization) levels the playing field for everyone in the ad game. In web2, smaller players in the advertising arena are compelled to rely on cookies and third-party data, which are often inaccurate and privacy invasive. The recent imposition of restrictions and bans on third-party cookies by major browsers like Chrome and Safari, and with Apple's recent initiatives to prevent ad tracking further complicates the ability of independent ad networks to effectively target their campaigns. This solidifies the position of the ad monarchs without fostering healthy competition.

In the realm of blockchain, every transaction provides valuable insights into user behavior. For instance, when wallet_xyz engages with a gaming dapp or purchases a music-based nft, we can deduce the user's interests and present relevant ads.

‍

At Persona, our commitment to data privacy is paramount. After meticulously indexing 25,000 smart contracts from 5,000 dapps and 1,000 nft collections – amounting to approximately a billion rows of transactions on Ethereum and Polygon – we have constructed user personas for over 200 million wallets. Leveraging AI language models, we inferred dapp and nft categories by feeding in their descriptions to the model. Through a comprehensive understanding of wallet interactions with the dapps and nfts, we created wallet personas based on user interests.

In stark contrast to the web2 landscape, our ad targeting models focus solely on user interests, avoiding the use of privacy-invasive personal data such as age, gender. These wallet personas enable us to develop highly refined, accurate, and privacy-friendly audience segmentation for ad targeting. Furthermore, we expanded this approach to construct look-alike segments based on similar wallet behavior to compensate for limited data in some instances.

Undoubtedly, this approach presents its own set of challenges. The immense volume of data necessitates robust infrastructure and considerable computing power for processing. Achieving a critical mass of active wallets per day is imperative to ensure a satisfactory ad fill rate – meaning that for every ad requested by a wallet, the wallet must be pre-indexed, and a relevant ad must be available to be matched with the wallet persona.

At Persona, we harness the potent combination of artificial intelligence and on-chain data to revolutionize advertising. Capitalizing on the key attributes of blockchain – transparency, security, and immutability – advertisers can create more personalized and effective campaigns while mitigating the risks of fraud and ensuring data integrity. This paradigm shift holds the potential to empower companies in reaching their target audience in a more meaningful and impactful way.

If you find this blog interesting feel free to DM me on LinkedIn! Let’s have a conversation. 

[1] Machine Learning Techniques to  predict offsite user interactions based on onsite machine learned models

[2] Optimizing a resource usage per conversion for offsite content

‍