By Sean Clarke, Founder of PacificIQ and EcomIQ

Our paid media team and I have been looking at Meta's TRIBE v2 this week.

If you haven't seen it, TRIBE v2 is an AI model of the human brain. It predicts how people respond to sight, sound, and language. Meta released it on March 26 wrapped as a scientific experiment, for the good of humanity. But really, it's about helping advertisers understand how people will react to ads and content.

A lot of the industry is excited. I've seen agencies, gurus, and consultants getting really into it.

But when we sat down to actually deploy this in a real workflow, we ran into a few problems.

Here's what we found.

It doesn't tell you if your ad is good or bad

This is the first thing that hit us. TRIBE v2 doesn't say “this ad is working” or “this ad is broken.” It just gives you a brain map.

So you need someone on your team who can take that brain map, translate it back into marketing language, and then make changes to the creative. It's not very linear in terms of next steps.

That translation layer is the whole product, and it doesn't exist yet. The frameworks people are building right now are one-person Medium posts and prototype demos. Not real tools you can plug into a workflow.

It's an average model. Not your customer

The other thing we identified is that it's an average person's brain. Right now that average is mostly Western, well-educated, well-off, what researchers call WEIRD samples.

If you service a more niched customer segment, the predictions aren't going to be accurate. A tradie in Geelong, a small business owner in Manila, a 22-year-old skincare buyer in Singapore, the model has no idea what their brain does. It only knows what the average lab volunteer's brain does.

For most Shopify brands, that's a problem.

It can't tell if your customer is tired, stressed, or distracted

Same person, same ad, completely different response depending on their state. Are they on the couch with their morning coffee, or scrolling on their phone at 11pm half-watching TV? The model doesn't know.

And that's the reality of how Meta ads run. They run 24/7, to people who are distracted, tired, multitasking on their phones. The model is blind to all of it.

So there's a generalist nature to TRIBE v2 that I don't think is super helpful today. I'm sure it'll come, but it's not there yet.

And you can't legally use it commercially anyway

Meta released it under a research-only license. Commercial use requires negotiating directly with Meta.

Smaller brands are messing with it, seeing how it goes. Bigger brands can pay for it. But even at agency level, the question we're sitting with is, do we want to pay for this tech? Is it going to provide any value?

As of today, we're not sure. I think it's still too early.

So what should operators actually do?

Right, so we're not deploying this in client work. What's the move for a Shopify brand doing $250K to $1M?

Four practical things. None of them take more than an hour.

1. Spend 10 minutes on the demo.

Go to aidemos.atmeta.com/tribev2. Click around. See what it actually does and doesn't do.

This is purely so the next time someone in your DMs or on a sales call drops “TRIBE v2” as a buzzword, you know what's real and what's hype. Cheapest insurance you'll buy this year.

2. Audit your actual creative testing.

Honest question for brands at this level. Are you making creative decisions on real data, or on vibes?

For most operators doing $250K to $1M, it's vibes. You launch an ad, it does okay, you keep running it. You don't run structured tests. You don't track which hooks work. You don't have a creative scorecard.

That's the real gap. Worry about that before you worry about brain AI.

3. Apply the principles, without the model.

The things TRIBE v2 actually measures, attention, surprise, semantic clarity, are things good creative directors have nailed for decades.

Hook in the first 1.5 seconds. Pattern interrupt at second 4. One clear message per ad. These aren't new ideas. The model just confirms what works.

You don't need a research model to do this. You need to ship more creative variants and be honest about what's actually performing.

4. Watch for it inside Ads Manager.

The moment to engage with this technology is when it shows up inside Meta Ads Manager as a “predicted creative score” or something similar. That's when small brands need to pay attention.

Meta giving away TRIBE v2 for free isn't generosity. It's strategy. Every researcher and agency that builds on this model makes Meta's ad platform more valuable. Same playbook they ran with PyTorch and Llama. Open the technology, capture the platform value.

I reckon predictive creative scoring lands inside Ads Manager in the next 18 to 24 months. That's where this matters for operators.

Where we landed

It's very cool tech. Will it help move the needle on return on ad spend, conversion rate, or your creative team's output?

As of today, probably not.

Until then, crawl, walk, run. Don't sprint into the hype.

Want operator-led reads on what's actually worth your time?

Subscribe to our mailing list. We send the takes that cut through the noise, before they're published anywhere else. Real numbers, real brands, no fluff, just what's working right now for operators in the trenches.