Moloco claims new machine-learning video streaming DSP competes with walled gardens

It has identified a flux of interest and spend, citing the Interactive Advertising Bureau (IAB)’s 2023 Compass claiming ad-supported video viewing adspend will double from £1.17bn in 2021 to £2.31bn in 2026.

It aims to help media companies expand out of the traditional linear TV business by proving the tools to monetize content in streaming environments. It claims that streaming video provides a huge opportunity but has a high barrier to entry in building an ad product – something even Netflix is finding out.

It is on the hunt for a broad range of streaming/OTT media, digital content owners, social media platforms, and utility apps to partner with. Publishers, particularly those with specific niches of interest to advertisers can pull their first-party data into the DSP. Moloco’s Ahn says this “democratizes access to potential customers for brands of all sizes and advertising budgets”.

But in particular, it is aimed at publishers that won’t have the in-house capabilities to get the most out of their data in the streaming video environment.

At a time when the industry is buzzing around the possible applications of AI tech, Moloco’s been pointing to its machine learning capabilities. It says it will apply the tech to optimize campaigns by matching ad relevance to first-party data, which will likely include inferred user interests. “Machine learning [drives the] neural network of campaigns and outcomes for any advertiser goal. Traditional solutions only provide outcomes and predictions in a linear and waterfall manner.”

This, Ahn says, will be the product's unique selling point. “The problem with most ad solutions today is that they treat every user the same. They serve ads based on price (CPM), priority, and static audience groups rather than relevance, which is crucial to driving conversions and advertiser outcomes... The tech that makes performance possible has only been available to walled gardens.”

He claims that Google led the way here, by calculating the “expected value of ad-user combinations by evaluating every ad against every user. Then, Meta (Facebook at the time) perfected that model with outcome-based buying (CPI)”.

The adtech firm is now claiming it has a workable model that can do that for publishers. “The models automatically identify users most likely to convert by parsing real-time user behavior and expand audience pools by uncovering hidden signals.”

It’s early days for the new product. The business claims it is working with “several AVOD/OTT design partners” which it will announce later in the year. It currently facilitates more than $1bn of ad expenditures annually for its customers.