Spend five minutes in retail media and you'll hear the same thing: everyone has AI. Ad servers have it. DSPs have it. Analytics tools have it. Even dashboards now "use AI".
At this point, the term has lost all meaning. So the more useful question is:
How do you tell if the AI is actually doing something real?
"AI" Has Become a Label, Not a System
In many products, AI shows up as a recommendation toggle, a forecasting widget, a reporting summary, or a rule-based optimizer with a fresh coat of paint.
These features can be useful, but they are not the same as AI that changes outcomes.
In retail media, the difference matters because the system isn't just analyzing data. It's making economic decisions: which ads to show, which products to rank, how budgets are spent, how revenue is generated.
If the AI doesn't influence those decisions, it's not part of the core system. It's just an add-on.
A Simple Test: Does It Control the Auction?
One thing cuts through the noise: does it actually influence the auction?
If the answer is no, the AI is not driving the system. It's observing it.
Real AI Lives in the Feedback Loop
Real AI systems have a loop: signal → decision → outcome → learning → better decision.
If the system improves over time without manual tuning, the AI is real. If it requires constant human adjustment, it isn't.
Where AI Actually Matters
In a retail media system, AI should sit in the parts that directly affect revenue:
Auction ranking — deciding which advertiser or product wins a placement.
Budget pacing — distributing spend intelligently over time.
Cross-surface allocation — splitting demand between search, category, banners, and offsite.
Product and audience selection — determining who sees what, and when.
If AI isn't influencing these layers, it isn't really driving performance.
What "AI Washing" Looks Like
A few patterns show up repeatedly. Reporting AI summarizes performance after the fact but never touches delivery. Rule-based systems get rebranded as AI — but "if CTR drops, increase bid by 10%" is automation, not intelligence. Batch optimization runs on a schedule rather than in real time. Isolated models exist somewhere in the product but never connect to execution.
These approaches can still add value. But they don't compound performance over time.
The Difference You Can Feel
In a system where AI is real, campaigns stabilize faster, pacing becomes predictable, fewer manual interventions are needed, and performance improves over time
In a system where AI isn't real, teams constantly adjust campaigns, results fluctuate, reporting and delivery drift apart, and performance plateaus.
You don't need a whitepaper to spot the difference. You can see it in how the system behaves day to day.
Why This Matters Now
Retail media is moving from tools to infrastructure. In a tool-based world, AI is a feature you can toggle on or off. In an infrastructure-based world, AI is the decision layer, the thing the system can't function without.
As the industry shifts toward real-time auctions, automated systems, and agent-driven commerce, the systems that actually make decisions will matter far more than those that simply report on them.
The Better Question to Ask
Forget “Do you have AI?” — find out where it actually lives in your system.
- Is it in reporting, or in the auction?
- Is it optional, or essential?
- Does it assist decisions, or does it make them?
The answers to those questions tell you everything.
The Bottom Line
AI is not defined by what a product says. It's defined by what the system does.
In retail media, real AI sits inside the auction, learns from outcomes, improves allocation over time, and directly impacts revenue.
Everything else is just a layer on top.
Everyone Says They Have AI. How Do You Know If It's Real?
Spend five minutes in retail media and you'll hear the same thing: everyone has AI. Ad servers have it. DSPs have it. Analytics tools have it. Even dashboards now "use AI".
At this point, the term has lost all meaning. So the more useful question is:
How do you tell if the AI is actually doing something real?
"AI" Has Become a Label, Not a System
In many products, AI shows up as a recommendation toggle, a forecasting widget, a reporting summary, or a rule-based optimizer with a fresh coat of paint.
These features can be useful, but they are not the same as AI that changes outcomes.
In retail media, the difference matters because the system isn't just analyzing data. It's making economic decisions: which ads to show, which products to rank, how budgets are spent, how revenue is generated.
If the AI doesn't influence those decisions, it's not part of the core system. It's just an add-on.
A Simple Test: Does It Control the Auction?
One thing cuts through the noise: does it actually influence the auction?
If the answer is no, the AI is not driving the system. It's observing it.
Real AI Lives in the Feedback Loop
Real AI systems have a loop: signal → decision → outcome → learning → better decision.
If the system improves over time without manual tuning, the AI is real. If it requires constant human adjustment, it isn't.
Where AI Actually Matters
In a retail media system, AI should sit in the parts that directly affect revenue:
Auction ranking — deciding which advertiser or product wins a placement.
Budget pacing — distributing spend intelligently over time.
Cross-surface allocation — splitting demand between search, category, banners, and offsite.
Product and audience selection — determining who sees what, and when.
If AI isn't influencing these layers, it isn't really driving performance.
What "AI Washing" Looks Like
A few patterns show up repeatedly. Reporting AI summarizes performance after the fact but never touches delivery. Rule-based systems get rebranded as AI — but "if CTR drops, increase bid by 10%" is automation, not intelligence. Batch optimization runs on a schedule rather than in real time. Isolated models exist somewhere in the product but never connect to execution.
These approaches can still add value. But they don't compound performance over time.
The Difference You Can Feel
In a system where AI is real, campaigns stabilize faster, pacing becomes predictable, fewer manual interventions are needed, and performance improves over time
In a system where AI isn't real, teams constantly adjust campaigns, results fluctuate, reporting and delivery drift apart, and performance plateaus.
You don't need a whitepaper to spot the difference. You can see it in how the system behaves day to day.
Why This Matters Now
Retail media is moving from tools to infrastructure. In a tool-based world, AI is a feature you can toggle on or off. In an infrastructure-based world, AI is the decision layer, the thing the system can't function without.
As the industry shifts toward real-time auctions, automated systems, and agent-driven commerce, the systems that actually make decisions will matter far more than those that simply report on them.
The Better Question to Ask
Forget “Do you have AI?” — find out where it actually lives in your system.
- Is it in reporting, or in the auction?
- Is it optional, or essential?
- Does it assist decisions, or does it make them?
The answers to those questions tell you everything.
The Bottom Line
AI is not defined by what a product says. It's defined by what the system does.
In retail media, real AI sits inside the auction, learns from outcomes, improves allocation over time, and directly impacts revenue.
Everything else is just a layer on top.