The data marketplaces powering programmatic advertising have exploded, with DSPs, SSPs and third-party platforms offering solutions for curating custom audiences.
But, for marketers, combining data segments to reach their target audience can be a confusing and wasteful process.
AudienceMix, a new curation startup, aims to make it more cost effective to mix and match different audience segments using only the data brands need to execute their campaigns. This helps mitigate wasteful overlaps between off-the-shelf audience segments.
The company was founded in April by Tom Mitchel, the former SVP of programmatic sales for measurement and analytics platform InMarket.
AudienceMix already has more than 8,200 syndicated audiences available via data collaboration platform LiveRamp, which is integrated with most major DSPs and SSPs. It sources its data from first-party and third-party data providers, drawing insights from more than three billion devices and a billion daily active users. The company declined to name any data partners at this time.
AudienceMix offers its curated audiences for a 65-cent CPM on average. Mitchel claimed this is less than half the price of most syndicated audience segments available via LiveRamp, which he said typically average between $1 and $3 CPMs.
The startup is already seeing strong demand since it launched in LiveRamp’s data marketplace in July. In August, it earned more revenue than any other data company has in its first month on LiveRamp’s platform, Mitchel claimed. Then, in September, it doubled its August revenue.
Flexible data curation
For Mitchel, that demand for AudienceMix’s curation proves there’s something missing in how most data sales work today.
Mitchel speaks from experience, having spent much of his career in programmatic and data sales. He started as email marketing platform LiveIntent’s first-ever sales rep, then shifted his focus to data sales at mar tech companies MaxPoint Interactive, ownerIQ and InMarket.
According to Mitchel, most DSP and SSP data marketplaces operate on an “and/or” principle. Marketers can pay to apply different data signals to their campaign targeting, he said, but these signals typically get applied as standalone products rather than being combined in an intelligent way.
For example, say a brand wants to reach its target audience of Walmart shoppers. Through a DSP’s or SSP’s data marketplace, the brand might apply targeting segments from multiple location data providers and multiple online and offline purchase data providers.
The problem is, Mitchel said, there could be significant audience overlap between each of these different audience segments. Which could mean marketers are wasting part of their budgets by buying unnecessary segments.
And besides, Mitchel said, the marketer probably doesn’t know and can’t control how the DSP or SSP weighs each audience segment when executing the campaign. He said he’s seen some platforms give priority to whichever is the most expensive segment purchased for the campaign, whereas others prioritize the least expensive segment.
AudienceMix lets marketers control which segments have priority. Mitchel described this strategy as letting marketers “mix and match their own cocktail with these raw ingredients.”
Using the solution, a marketer could apply two different location data signals, but use one signal deterministically while using the other probabilistically. Marketers also decide how much priority each probabilistic signal has in the platform’s audience modeling.
For example, a cereal brand might be targeting a deterministic audience of grocery store shoppers who provided an email address for a rewards program. They can then apply probabilistic modeling to target shoppers within this cohort who are likely to buy cereal based on past purchase data, and then layer on additional probabilistic signals to further narrow down this audience. The advertiser can set how much weight each probabilistic signal should receive, while AudienceMix handles the actual modeling work.
Currently, marketers communicate directly with AudienceMix to design custom curated audiences. But Mitchel said he’s working on a self-serve user interface.
AudienceMix also syndicates some common targeting methodologies so they’re available for other brands to activate. These appear as off-the-shelf products in data marketplaces.
Data networking
In addition to offering flexible audience curation for buyers, AudienceMix helps niche data sellers connect to more brands.
Take, for instance, a data provider that has very granular data sourced from five million mobile devices, Mitchel said. That segment could be valuable for the right brand. But at such a small audience scale, the data provider would have trouble getting a seat in most DSP or SSP data marketplaces.
It can also be cost-prohibitive for smaller data providers to sell their data through programmatic marketplaces, Mitchel added. Getting integrated into LiveRamp usually involves signing a six-figure contract, he said. And it would probably take a million-dollar effort for the data provider to spin up its own data sales team.
But combining these niche data providers into its network lets AudienceMix keep its own costs down while offering marketers and data vendors greater scale, Mitchel said.
And because it’s hard to purchase these niche data segments elsewhere, AudienceMix can give buyers access to a new audience pool that they probably couldn’t reach using off-the-shelf solutions, he said.
Complementing the marketplace
While it’s already seeing strong adoption via LiveRamp, AudienceMix is still very much in the startup phase.
It’s largely a one-man operation run by Mitchel. But he works with about six independent contractors sourced through Amazon Web Services’ partner program. AWS also provides the cloud infrastructure to support the solution.
For data privacy compliance, AudienceMix works with the law firm Frankfurt Kurnit Klein & Selz. It is also a registered data broker in the four states that require such registration: California, Vermont, Oregon and Texas. The company does not handle any personally identifiable information, Mitchel said, “and we don’t even touch the [data] categories that are deemed sensitive.”
AudienceMix is almost entirely self-funded and bootstrapped by Mitchel. And his data partners are paid on a revenue share model based on the size of their demand contribution to the platform.
However, the company recently received its first-ever outside investment: $100,000 in funding for AI expansion granted through Nvidia’s Inception accelerator program for tech startups.
Going forward, AudienceMix is exploring integrations with other data collaboration platforms like LiveRamp. And Mitchel is building a sales team.
But, rather than gunning for large incumbent data providers and curation platforms, Mitchel is keeping his goals for AudienceMix realistic. He’s not sure if the company needs to be a $100 million business, and he said he would be satisfied just having “a piece of the market.”
“I don’t need to compete with others,” he said. “I want to complement and find that extra incrementality for people looking for efficiency.”
