There’s a theory floating around the ad industry that keeps me up at night.
It goes something like this: AI will eventually know every consumer so well that we can serve them the exact right message at the exact right moment. The ad will feel less like an interruption and more like a helpful suggestion from a trusted friend. We will eliminate waste. We will eliminate irrelevance. We will eliminate the gap between what the customer wants and what we show them.
Sounds utopian. Sounds logical. Sounds like the endgame of personalization.
It is also a trap.
Here’s the uncomfortable truth that few agencies are willing to say out loud: Perfect personalization kills brand building.
If every ad you serve simply confirms what the consumer already knows they want, you’re not building desire. You’re fulfilling demand. That’s a fulfillment business, not a marketing business. And the moment you become a fulfillment business, you become interchangeable. The customer will leave you the second the algorithm finds a cheaper option.
At Sagum, we’ve spent years building a different philosophy. We limit our client count to ensure focus. We use data like water-we need it to survive, but we don’t drown in it. And we believe the future of AI in advertising isn’t about eliminating tension. It’s about engineering it.
Welcome to the era of the Tension Algorithm.
The Problem With Zero-Risk Advertising
Let me explain what I mean by “tension.”
In the current model, most AI-powered personalization is reductive. It looks backward at user behavior-searches, clicks, purchases-and predicts what the user already wants. Then it serves that. This is convergent thinking. It narrows the funnel until the consumer feels like they’re being chased through a hallway of mirrors.
The result? High conversion rates, low brand equity. The user buys, but they don’t feel anything. They don’t remember who sold them. They remember the price and the convenience.
That’s a race to the bottom.
The brands that win in the long run don’t just fulfill needs. They create longing. They introduce a gap between where the consumer is and where they could be. They make the consumer uncomfortable enough to want to change-but not so uncomfortable that they leave.
This is where AI has been failing. Machines are trained to reduce friction. But friction is often the point.
The Shift: From Convergent to Divergent Personalization
At Sagum, we’re experimenting with a new model. We call it divergent personalization.
Instead of asking, “What does this user already want?” the AI asks, “What is this user ready to want?”
The difference is subtle but profound. It requires the AI to stop optimizing purely for historical data and start looking for latent openness.
Latent openness is the signal that a consumer is ready to be challenged. It’s not a direct intent signal-they didn’t search for your product. But their behavior suggests curiosity. A longer dwell time on a certain topic. A pattern of clicking on articles that challenge their current worldview. A sudden spike in browsing for status-related goods after a life event.
The human brain does this naturally. A good strategist looks at a consumer and thinks, “They’re bored with their current solution. They’re ready for something aspirational. Let me offer them a vision of their future self.”
Now, AI can learn to do this at scale. But it requires a different set of instructions.
How to Train the Tension Algorithm
Here is the tactical framework we’re using with our clients at Sagum. It has three layers:
1. Instruct the AI to fail 10% of the time
This is the hardest sell for most business leaders. They want efficiency. They want every dollar to work. But efficiency without tension is a commodity.
We tell our clients: allow the AI to serve one out of every ten impressions that is deliberately outside the user’s comfort zone. Not offensive. Not irrelevant. Just slightly aspirational. A product tier higher than they usually browse. A message that challenges their status quo. A creative that makes them stop and think, “Wait, do I need that?”
Measure the “surprise metric.” How many users engage with the unexpected message? How many convert later after seeing it? The data will surprise you.
2. Feed the AI brand tension, not just product specs
Most AI tools are fed transactional data: SKUs, prices, past sales. That teaches the machine to optimize for the transaction.
Instead, feed the AI the brand conflict. Why does the customer need to change? What is the tension your brand resolves?
- For a luxury watch brand, the tension is not “you need a watch.” The tension is “your current self is not living up to your future potential.”
- For a B2B SaaS company, the tension is not “your workflow is inefficient.” The tension is “your competitors are passing you by while you cling to outdated methods.”
When the AI understands the tension, it can serve creative that introduces the problem before it offers the solution. That’s how you build desire.
3. Bid on the gap, not the person
Most programmatic advertising buys a person. It says, “This user has these attributes. Show them this ad.”
The Tension Algorithm buys a moment-specifically, the moment between where the user is and where they want to be. This is the “aspirational gap.” The AI learns to identify when a user is most receptive to being pulled forward.
The result is an ad that feels less like a pitch and more like a realization. The user thinks, “I was already thinking about this. This brand just articulated it for me.”
What This Means for Business Leaders
If you’re running a business focused on long-term growth, the Tension Algorithm changes your relationship with AI.
You stop asking, “How do I get the AI to find me more customers just like my current ones?”
You start asking, “How do I get the AI to find me customers who are ready to become someone different?”
You stop optimizing for lowest cost per acquisition.
You start optimizing for engagement velocity-the speed at which a user moves from indifference to aspiration to action.
This is harder to measure. It requires better data infrastructure. It requires a BI dashboard that can track not just conversion, but movement.
It’s worth it.
The Risk of Ignoring This
I’ll leave you with a warning.
The agencies that treat AI as a pure automation tool will commoditize their clients. They will optimize for the cheapest clicks and the safest creative. They’ll win the short-term race and lose the long-term war.
The brands that treat AI as a tension engine will build something different. They’ll build loyalty. They’ll build desire. They’ll build brands that consumers don’t just buy-they chase.
At Sagum, we built our agency for business leaders who think beyond the next quarter. We’re lean. We’re data-first. But we’re also strategic. We know that the best advertising doesn’t just satisfy a need. It creates a gap worth closing.
The future of AI in advertising isn’t about making the customer comfortable.
It’s about making them want.
And that starts with a little bit of tension.
