The Brand Network Effect: AI changes how brands scale, not why people choose them.

Opinion – Week in Focus

When brand signals reinforce across memory, social media and machine systems, you stop buying visibility and start accumulating it. This changes how we think about brand strategy, argues Havas’ head of strategy and effectiveness.

AI has given marketing extraordinary new powers. It has also introduced a new tension.

Being seen has never been easier. Being wanted has never been harder.

We’ve become fluent in designing for systems that rank, recommend and predict. But growth still depends on human preference. And somewhere in the race to optimise, we’ve lost sight of that.

I keep coming back to this: visibility and desire are not the same thing. Confusing them is costly.

For decades, the role of media was straightforward: create repeated exposure, build salience, and show up at the right moment. Today, visibility is increasingly shaped by systems that filter and surface brands before conscious choice begins. Search engines rank. Platforms recommend. AI systems summarise. Behavioural data now shapes visibility as much as paid exposure ever did.

The fundamentals haven’t changed. Brands still grow by becoming easier to notice and easier to choose. But in optimising for systems, we’ve started to undervalue something that still does all the heavy lifting — desire. A distinctly human force. And one that no algorithm creates.

The question is no longer whether brand building matters. It is how brand building works in a world shaped by social and machine systems.

Brands now operate across three systems 

Something I’m increasingly noticing is how brands today operate across three interconnected systems: memory, social and machine. Growth increasingly depends on how coherently those systems reinforce one another.

Thinking about brands as networks isn’t new. But the more important question today is how those networks compound. That’s what I call the ‘Brand Network Effect’.

Borrowed from platform economics — where products become more valuable as more people use them — the Brand Network Effect describes something similar for brands: when your signals reinforce across memory, social and machine systems, you stop buying visibility and start accumulating it.

Memory remains foundational

Much of the best thinking in marketing has rightly centred on memory.

Brands grow when they become easier to notice and easier to retrieve at the moment of choice. That ease comes from repeated exposure to stable, distinctive signals – visual cues, sonic assets, language, characters and product shapes consistently linked to the brand over time.

Most advertising is consumed in low attention conditions. People absorb patterns rather than messages. When those patterns remain stable, brand recognition gets faster and recall becomes more automatic. Familiarity reduces perceived risk and shortens decision time.

Creative work people actually enjoy strengthens this. Intrusive, purely functional work makes it harder. But the principle holds: brands grow by building memory structures.

Memory prepares the brand for the moment of choice. It just doesn’t operate in isolation anymore.

The social network multiplies signals

Brand signals do not stay in one place. 

They get shared, referenced, debated and reframed in the public domain. That’s crucial because it strengthens familiarity and legitimacy in a way that paid exposure alone can’t.  

This isn’t about chasing virality. Most viral moments don’t accumulate into anything lasting. The real question is simpler: are your brand signals coherent enough to travel? Flexible enough to be referenced, shared and repeated without falling apart?

When brand identity is stable, signals propagate. When it fragments, that propagation weakens. Social amplifies consistency.  

The machine network stabilises visibility

The third layer is the most structurally significant.

Search engines rank. Retail platforms prioritise. Recommendation systems filter. AI interfaces summarise and suggest. Increasingly, brands are surfaced (or excluded) by systems trained on behavioural data.

It is important to focus on that last part. 

Machine systems do not create desire. They learn from what real people find desirable. 

Recent research shows that brands with the highest levels of third-party mentions and cultural conversation are around four times more likely to be cited by AI than those without. The job of the machines is to reflect what people already think.

Which means if you optimise for machines without first building genuine preference, you might end up highly visible but not particularly wanted. 

The machine system is powerful, but it is downstream of human desire. 

The reinforcing loop

Strong memory increases recognition and interaction. That interaction drives social sharing and behavioural signals. Those signals train machine systems, which surface the brand more. More visibility reinforces memory. That’s the loop. 

Netflix is the clearest example. Ta-Dum. 

Distinctive brand cues support instant recognition. Content being shared amongst people sustains your brand in conversations. People’s viewing behaviour then feeds what’s “recommended” to others. Recommendations drive exposure. Exposure deepens familiarity. 

Each cycle reinforces the next.

The brands winning today are experts in building these reinforcing loops. 

From campaigns to infrastructure

This changes how we think about brand strategy.

For years, brands have largely been managed as campaigns – bursts of communication designed to refresh memory and recapture attention. But now, we need signals to operate across memory, social media and AI tools. 

What’s needed is something more like infrastructure — something that connects environments, reduces friction and keeps working between campaign bursts. 

Jet2 is a good example of this done well. Consistent sonic, visual and emotional cues built strong memory structures over time. Those cues then travelled socially, becoming a genuine cultural reference point — not because the brand chased that, but because the signals were coherent enough to spread.

Social interaction generated behavioural signals, increasing search demand and platform visibility. The signals are connected across systems rather than being fragmented.

Jet2’s strength wasn’t an accident of meme culture. It was a great example of coherence.

The commercial consequence

So, what does this actually mean for the business?

When memory and networked visibility reinforce each other, the commercial effects compound in ways that isolated campaigns simply can’t replicate.

Acquisition gets cheaper because you’re being surfaced rather than having to buy your way back into consideration every time. Pricing holds better because familiarity reduces the urge to switch. Visibility doesn’t fall off a cliff between campaigns. And demand becomes less volatile, because consistent signals generate consistent behaviour. 

The alternative is paying for attention you should already own. Resetting memory structures every campaign. Retraining algorithms that never quite learn. Rebuilding from scratch what should have been accumulating.

That’s linear effort in a compounding system. And it’s more common than most brands would like to admit.

The future of brand building isn’t about designing for machines. It’s about building brands that machines learn to surface because people already prefer them.  

Chetan Murthy is head of strategy and effectiveness at Havas Media Network