It wasn’t too long ago that AI was the new kid on the block. Today, it’s how high-performing advertising gets done.
Shoppers are now embracing AI at a rapid pace, and marketers are right behind them.
A global McKinsey’s survey on the state of AI reports that 78% of companies use AI in at least one business function, and 71% regularly use generative AI – with marketing and sales among the top functions adopting it.
On the consumer side, recent studies show 71% of shoppers want generative AI integrated into their buying journeys. Translation: AI in advertising has shifted from nice‑to‑have to standard operating procedure.
This guide breaks down the basics, compares generative and predictive AI, and offers a roadmap with practical ways to use AI in advertising across the funnel.
We’ve got a lot to cover, so let’s get started.
What is AI in advertising?
AI in advertising is about using machine intelligence to help humans make better decisions across the ad lifecycle. That means everything from audience selection and bidding to creative, measurement, optimization, and the rest.
The real goal of AI in the context of advertising is to automate repetitive work and surface insights that humans would (probably) miss.
A quick refresher on the terms you’ll hear (or you can read a more in-depth comparison here):
- Artificial intelligence. An umbrella term for systems that can perform tasks that typically require human intelligence: learning, pattern recognition, decisioning, and content generation.
- Machine learning. A subset of AI that learns patterns from data to make predictions. You give it examples; it finds the signal.
- Deep learning. A subset of machine learning that uses neural networks with many layers to learn complex representations, especially useful for images, audio, and language.
So where does this fit into your stack?
One way to think of AI is as the decision-making and content generation layer that sits on top of your existing data. That data feeds into models, which then inform your activation channels – programmatic, retail media networks, social, search, video, CTV – and your analytics.
For a deeper dive into day‑to‑day workflows in the context of performance advertising, take a look at AI for performance marketers: How to cut through the noise (and boost results).
Generative AI vs. predictive AI in advertising
Unless you’ve been hiding under a rock, you’ll already be familiar with generative AI – it powers the most popular AI tools today, like ChatGPT.
But alongside these generative tools you’ll also find specific machine learning models designed for specific tasks – just like the predictive AI which powers many of Criteo’s solutions.
Both of these technologies matter – and they can be even more powerful when combined.
Here’s a quick comparison of what each of them does:
| Generative AI | Predictive AI | |
|---|---|---|
| Goal | Produce content variations and ideas | Predict outcomes and choose actions |
| Input data type | Prompts, brand guidelines, product feeds, images/video | Event and conversion logs, user behavior, catalog and price data, context |
| Output | Copy, images, video, dialog flows, structured summaries | Scores, bids, audiences, recommendations, pacing and allocation |
| Uses in advertising | Creative ideation, asset scaling, chatbot ads, localized variations | Targeting, bidding, DCO, budget and channel optimization |
How AI works in advertising
If you’re most familiar with AI in the context of tools like ChatGPT, Claude, and Gemini, it can be tough to transpose that to the advertising world – so let’s clarify.
At a very high level, most AI systems follow the same flow:
Data signals → Feature engineering → Model training → Decisions in‑market → Measurement → Feedback loop
That’s a bit of context, but it could still do with a bit more explanation, so here’s what’s happening under the hood:
- Data collection. The “data signals” referenced above include user behavior on websites and apps, product catalog attributes, inventory and pricing, content and context, and anonymized ad interactions. Strong consent and privacy controls are table stakes – you want high‑quality, privacy‑safe signals.
- Model training. Models learn patterns that matter for advertising: purchase intent, creative responsiveness, propensity to churn, or predicted value. Deep learning helps with complex signals like images and language, while gradient‑boosted trees and logistic regression still shine for structured performance data.
- Automation and optimization. The models power automated bidding, audience selection, dynamic creative selection, and budget allocation. The system tests, learns, and re‑allocates in near real time, so every impression has a better shot at doing useful work next time.
Practical use cases of AI in advertising
That’s how AI works in advertising. The next question you might ask is why you’d want to use it.
That might seem like a softball question, but there’s more nuance here than you might think.
Audience targeting and segmentation (identifying the right customers)
Predictive models build audiences based on conversion probability, intent signals, and product affinity. Retailers and brands also use first‑party data to reach loyalty members or high‑value cohorts across the open internet.
Real-world example: Sephora used Criteo Audience Match and Dynamic Retargeting together to engage loyalty members and drove a 725% ROI across the journey.
Dynamic Creative Optimization (DCO) (showing the right creative at the right time)
DCO pairs predictive decisioning with modular (read: dynamically created) creatives. The system chooses headline, image, product, and price to match the shopper and context. Criteo’s DCO technology personalizes ads in real time using insights from billions of shopping signals, while keeping brand standards intact
Real-world example: See Shopify’s performance boost with Criteo product recommendations and dynamic ads.
Chatbot ads and interactive experiences
Generative AI powers conversational formats that let shoppers ask questions, compare products, and get guided recommendations inside the ad unit. Beyond that, our team is working on agentic solutions which will transform the way consumers shop and interact with the products they love (or those they will love).
Find out more about our vision of the agentic future from our Chief Product Officer, Todd Parsons.
How to start using AI in your advertising strategy
We all know that AI moves at lightning speed. By the time you read this, there’ll be another model, another feature, or another tool vying for your attention. But there are still some fundamentals which you can adopt in your day-to-day campaign planning to help leverage AI to best effect today.
Here are four steps to get you off on the right track:
- Step 1: Define your campaign goals. Be specific about the KPI – incremental sales, new‑to‑brand customers, qualified leads, app LTV – and where you’ll measure it. Clear goals tell the models what to optimize.
- Step 2: Choose the right AI tools and platforms. Look for partners that combine strong predictive models with transparent controls – and those that have access to unique data to power them. If you’re looking to start quickly and scale fast, consider automated advertising solutions like Criteo’s GO! campaigns.
- Step 3: Test and iterate on a small scale. Start with a pilot that has clean measurement – a defined audience, clear control vs. exposed groups, and a read on incrementality. Use a handful of creatives (or modules in the case of DCO) so the system can learn without getting noisy.
- Step 4: Integrate into broader marketing automation. Build a simple data loop: pipe consented first‑party audiences and conversion events from your CDP and analytics into Criteo, keep a clean product feed updating daily, and connect conversion APIs so signals stay consistent across web and app. Use a goal‑based setup to let AI handle bidding, targeting, and placement while you set guardrails for ROAS, CPA, and frequency.
Putting it all together
AI in advertising works best as a unified system – predictive models pick the moment and audience, generative models shape the message, and feedback loops keep improving results over time as the system learns.
Want to go deeper on AI, machine learning, and deep learning for performance marketing? Explore our latest articles on the Criteo blog – starting with AI for performance marketers – and if you’re evaluating commerce media and retail media partners, let’s talk about how to put these ideas to work on your next campaign.