Algorithmic double jeopardy: AI advertising and the brand divide

When Mark Zuckerberg announced that we’ll see fully automated AI advertising by 2026, it painted an exciting future for companies of all sizes. With Meta’s AI Ad Engine, imagine this: all it needs is your brand’s URL and your budget, and AI takes care of the rest. It generates visuals and copy, determines the right targeting, optimizes campaigns and reports along the way.

Of course, it’s an efficiency breakthrough: creativity and production become commodities. But where does that leave the human touch? While execution is automated, strategic thinking is not. End-to-end campaign creation is a compelling vision, but AI-driven platforms – without clear objectives, signals and governance – can actually optimize brands into irrelevance. We look at the gaps this creates and how you can navigate this new era with proactive customer orchestration.

The optimization gap

AI-driven ad platforms are optimization engines that relentlessly pursue the goal you set. Tell them to optimize for conversions, and they’ll do it, even at the expense of your brand’s long-term value. If you set profit as your goal, the system might push low-priced, easy-to-convert products, sidelining your premium positioning. This is the heart of the AI advertising challenge: the gap between what you intend and what you instruct. Closing it isn’t about better tools, it’s about sharper thinking.

“The machine maximizes what you asked for at scale, not what you meant. In other words, it understands what ‘more’ means. It has no concept of ‘better’.”
Ernst Hiemstra, Chief content and advertising strategist at Empathy Lab

Algorithmic double jeopardy: when signal loss goes haywire 

The Law of Double Jeopardy, rooted in Andrew Ehrenberg’s research and made famous by Byron Sharp’s How Brands Grow, states that smaller brands face a double whammy challenge: fewer buyers who buy less often. In today’s algorithmic ad world, that disadvantage multiplies. What we like to call 'algorithmic double jeopardy’ works against smaller brands in four ways:

• Smaller brands start with a limited customer base.
• Those customers buy less frequently, starving algorithms of the conversion signals they need to learn and adapt.
• Privacy legislation fragments the signal infrastructure that ad algorithms depend on even more. These include Apple's ATT framework, the EU's Digital Services Act, the Digital Markets Act and GDPR's consent requirements.
• Finally, when signals weaken, the algorithm stops exploring new audiences and falls back on the familiar, compounding the growth problem.

Large brands absorb signal loss through first-party data pipelines and server-side tracking. Constrained by budget, smaller brands don’t have this option. The outcome? Smaller brands pay more to reach fewer people, while the system confidently optimizes away from growth, learning only where the light is brightest.

Platform response: more AI, less transparency

The major platforms have responded to signal loss not by restoring transparency but by deepening their reliance on AI inference. Meta rebuilt its advertising infrastructure around Andromeda – a next-generation retrieval engine that inverts the traditional targeting model. Rather than beginning with advertiser-defined audiences, Andromeda evaluates historical engagement, creative signals and format preferences to infer likely converters. Creative has become the new targeting lever. Google has moved in the same direction with Performance Max and AI Max Search campaigns, abstracting audience controls behind AI-driven delivery.

The paradox: privacy rules mean less data for advertisers, so platforms double down on AI. Advertisers lose visibility on both ends, less data in, less transparency out. This is the fog your teams are navigating.

The cold start problem

Nowhere is the tension clearer than in campaign cold starts. Meta's algorithm needs volume to learn, so early results are messy. Most teams panic and intervene, cutting spend or tweaking settings when the system needs space to explore. Campaigns get killed before they can succeed, and the algorithm gets blamed for a governance failure.

The opposite is equally risky. Hands-off management lets the algorithm drift, optimizing for efficiency rather than strategy.

“Managing these systems is like learning to ski in fog. The fundamentals haven't changed, but visibility has.”
Ernst Hiemstra, Chief content and advertising strategist at Empathy Lab

The competitive advantage

The brands that consistently outperform in this environment share three characteristics:

We get it. The instinct is to reach for the old media plan again: the defined audience, the predictable funnel, the end-of-month report. That model is gone. The effective marketing leader does not try to restore it. They learn to exercise judgment inside a system that never stops moving, knowing which decisions belong to them, and which ones the machine has earned the right to make.

At Empathy Lab, this means an important shift: from reactive marketing to proactive customer orchestration. It’s about designing the conditions under which AI systems learn deliberately, transparently and in alignment with brand intent.

The best teams know when to trust the process and when to step in. That judgment is your competitive edge, and it can't be automated. If you want to learn how AI can be operationalized to drive brand performance, download our white paper: The Growth Operating System: a blueprint for growth in an AI-mediated world. Or get in touch to chat to an expert.