He spent nine years reading markets for Goldman and Bill Ackman. Now he is handing small brands the data science that used to belong only to the giants.
Every founder story has an origin moment. For Alex Song, it was a privacy setting. In 2021, Apple rolled out a change to iOS that let people opt out of the tracking that had quietly powered a decade of digital advertising. For the giants, it was a nuisance. For the thousands of small consumer brands that lived and died by their ability to find the right customer on Facebook, it was closer to an earthquake. Acquisition costs jumped. Targeting went blurry. The playbook stopped working.
Song had spent his career learning to read moments like this. Where most people saw a broken system, he saw a company waiting to be built.
That company is Proxima, and the idea behind it is almost stubbornly simple: the biggest advertisers in the world win because they have data science teams and mountains of first-party signal. The corner-store version of an ecommerce brand does not. Song wanted to close that gap - to give a five-person Shopify business the kind of audience intelligence that used to require a war chest.
Before there was a startup, there was a trading floor. Song graduated from the University of California, Berkeley with a degree in business administration and highest honors, then went straight into the deep end of finance. He spent nearly a decade as an investor - first at Goldman Sachs, rotating through investment banking and private equity, and later at Pershing Square, the hedge fund run by Bill Ackman.
It is the kind of resume that usually leads to a quiet, well-compensated life managing other people's money. Song took a different lesson from it. He talks about those years less in terms of deals and more in terms of temperament. Wall Street, he has said, taught him grit and resilience - the ability to sit with uncertainty, to be wrong in public, to keep going when the numbers are ugly. Those, he argues, are the traits that actually matter for building something from nothing.
Nine years reading markets teaches you one thing: grit beats genius.On the lessons of Goldman Sachs and Pershing Square
The investor's instinct never really left him. It just changed direction. Instead of picking companies to bet on, he started building them.
In 2014, Song founded Innovation Department, a New York venture studio built to invest in and launch category-disrupting startups, mostly in ecommerce and direct-to-consumer. A studio is a strange, useful vantage point. You are not just funding companies from the outside; you are inside the machinery, watching how a brand actually gets built, marketed, and scaled - or how it stalls.
Out of that period came several companies he helped bring to life, including a health and wellness venture and a customer engagement business. But the more valuable output was the pattern-recognition. Studio after studio, brand after brand, Song kept seeing the same bottleneck. The founders who could afford serious data science pulled away. The ones who could not spent blindly, guessing at who their customer was and paying more each quarter to reach them.
There is a version of Song who stays a studio founder forever - diversified, hedged, a little bit invested in everything. That is not what happened. In 2022, he did the thing investors are trained to be suspicious of: he concentrated. He took everything the studio had taught him and poured it into a single company.
Proxima's premise is that the most valuable question in marketing is also the most basic one - who actually buys? Not who clicks, not who browses, but who reaches for their wallet. To answer it, Proxima built a data set drawn from billions of annual social-commerce transactions across more than 70 million verified shoppers. Feed that signal into machine-learning models and you can predict which audiences are worth chasing, and prove which spend is working.
With data science, it doesn't have to be either/or. Advertisers can achieve both.Alex Song, on privacy versus performance
That last point matters to Song more than the mechanics. The reigning assumption in his industry is that you can protect people's privacy or you can target them well, but not both. He rejects the trade-off. Proxima's data is anonymized and aggregated; the promise is that a brand can find high-value customers without knowing, or needing to know, anything personal about any one of them. Privacy and performance, in his telling, are not enemies. They are just poorly introduced.
Ideas are cheap in adtech. What separated Proxima was that brands using it saw results they could put on a slide. One health and wellness company, Obvi, scaled its ad spend by roughly 130% while cutting the cost of acquiring a new customer and improving its return on ad spend - the rare combination of spending more and paying less per outcome.
The company shipped a self-service platform in the spring of 2023 so smaller brands could plug in without a sales call, then added an AI business health check that reads a brand's performance and tells it where it is leaking money. By its own account, Proxima's revenue has grown roughly 400% since 2022.
Investors noticed. In April 2024, Proxima closed a $12 million Series A led by Mucker Capital. Among the backers was Aglae Ventures, the family office connected to the founder of LVMH - an interesting vote of confidence from the empire that arguably understands consumer desire better than anyone.
Spend enough time listening to Song and a philosophy emerges that goes beyond targeting math. He believes the wall between business marketing and consumer marketing is coming down. The buttoned-up, jargon-heavy way companies have historically sold to other companies feels dated to him. People are people, whether they are buying sneakers on a Sunday or software on a Tuesday, and they have been trained by a decade of slick consumer brands to expect the same polish everywhere.
B2B marketing must look and feel more like B2C.Alex Song
It is a tidy summary of how he thinks, and it explains why Proxima itself does not feel like a typical data vendor. He treats plumbing - the deeply unglamorous work of transaction signal and audience modeling - as something that should be handed to a founder as cleanly and confidently as a consumer app.
Ask Song where this goes and the ambition is not subtle. He wants Proxima to become the essential data intelligence platform for consumer marketers - the layer that sits underneath every brand's decisions about who to reach and how to measure it. That is a big claim in a crowded field. But it fits the shape of his whole career, which has been a slow move from the outside of companies to the inside, and now to underneath them.
The through-line is a kind of restlessness. Song describes himself, in three words on his own profile, as a builder and world traveler - someone who keeps moving, geographically and professionally. He could have stayed at the hedge fund. He could have stayed a comfortable studio investor spreading his bets. Each time, he chose to build the harder, more concentrated thing instead.
There is a certain logic to a former investor deciding that the best asset to own is the one you make yourself. Song spent years betting on other people's companies. Proxima is the one he is betting on with everything, and the wager is refreshingly democratic: that the small brand, armed with the right data, can go toe to toe with the giant. If he is right, a lot of five-person teams are about to punch far above their weight.