Suggestions
Alex Song
Building AI to optimize user acquisition at scale.
Alex Song is the Founder and CEO of Proxima, a data intelligence company he founded in 2022.13 Proxima leverages a proprietary database of over 50 million personas to help digital businesses improve targeting and performance across media platforms like Facebook, Instagram, Snapchat, Pinterest, and TikTok.3
Prior to founding Proxima, Song had a diverse career in finance and entrepreneurship:
-
He worked at Goldman Sachs and Pershing Square Capital Management in investment roles.1
-
In 2014, he founded Innovation Department, a New York venture studio that invests in and builds category-disrupting startups.13
-
He co-founded several other companies, including:
Song's educational background includes:
- A Bachelor's degree from the University of California, Berkeley14
- An MBA from Harvard Business School2
Under Song's leadership, Proxima has achieved significant milestones:
- 400% increase in revenue since launching in 2022
- Expansion of their proprietary network to 12,000 B2C connections
- Growth of their unique shopper personas database to 65 million1
In April 2024, Proxima secured a $12 million Series A funding round led by Mucker Capital.1
Song is known for his expertise in data-driven marketing strategies and his commitment to helping brands optimize their customer acquisition and retention efforts in the face of challenges like rising advertising costs and platform algorithm changes.1
Highlights
Founder communities are incredible for learning. But there’s a hidden trap that’s causing expensive failed experiments.
ICYDK: Tight-knit founder groups are coming together to share strategies, tools, learnings.
And don't get me wrong, they can be powerful and incredibly helpful.
But this blind following leads marketing teams to either jump on winning experiments or avoid certain approaches at all costs. And when they do attempt to replicate others' experiments, they often fail for 2 critical reasons:
- They don't define success correctly
When brands rush to mimic others without understanding WHY it worked for them, they often skip defining what success means for their own business. They're following tactics without strategy.
You're dead on arrival when you can't articulate what winning looks like for YOUR specific situation, audience, and business model.
For example, a fitness supplement brand might copy a skincare brand's influencer strategy because they saw "great engagement," but they never defined whether they were going for awareness, conversions, or retention. Same tactic, completely different success metrics, predictable failure.
- They under-resource experiments
The pressure to quickly check the box on the latest trend means they don't commit adequate capital, time, or effort - leading to half-baked experiments.
A brand might hear "SMS drove 25% of our revenue" and launch with a limited budget, generic templates, and a poorly optimized funnel, then wonder why they're seeing high unsubscribe rates instead of revenue.
When you half-commit to a test, you get false negatives. You write off potentially strong opportunities before they have a real chance to succeed.
-- So instead of succumbing to the fast follower experimentation mentality of rushing to jump on trends without discipline...set clear metrics, commit real resources, and give experiments time to work.
That's how you go from burning money trying to keep up with everyone else to turning experimentation into a real growth driver.
There's a misconception that data-enriched lookalikes are just "better targeting." That it's about finding more of the same audience that broad finds, just with better accuracy and a tighter net. Wrong.
Data-enriched lookalikes are whatever you need them to be.
Want to pack your business with high-LTV customers? Build lookalikes from your top spenders. Running a clearance sale next week? Build lookalikes from your discount buyers who convert on promotions.
With broad targeting, you're throwing spaghetti at the wall. It forces you to use the same targeting for a flash sale that you'd use for a luxury launch because you get whoever Meta decides to give you—discount seekers mixed with big spenders, one-time buyers mixed with loyalists.
It's pure chaos, and minimal helpful indicators of improving LTV, CAC, etc.
One of the biggest advantages data-enriched lookalikes give you is fluid audience targeting. You don't have to only target the same audience. You can target who you want, when you need to.
And with @ProximaAI having access to billions of commerce datapoints—demographics, purchase patterns, line item details, brand preferences—we can dial in exactly who you need. Men, women, specific ages, chemical-free consumers, discount lovers, subscription loyalists, whatever drives your business forward right now.
Playing the spray-and-pray game was left in 2020. The brands crushing it right now are building specific audiences for specific goals - and watching every metric improve as a result.
DM me if you're dealing with this same audience targeting mess. Happy to brainstorm what would work better.
:quality(70)/cloudfront-eu-central-1.images.arcpublishing.com/businessoffashion/Q4ZESBLNPNB67BXNEKXKOBCWK4.jpg)
