For decades marketing strategy began with a familiar line: “Our target audience is…” But the rise of behavioral signals, AI search and algorithmic media buying is quietly replacing audience definitions with something far more dynamic—real-time behavioral data that reveals what people are about to do, not just who they are.

In the past, I have written about the shift from demographics to personas. For decades marketers targeted people using census categories—age, income, gender, geography. Then along came personas, which at least forced us to think about motivations instead of birth certificates.

Personas were an improvement. And they’re still useful in many contexts—particularly when planning legacy media campaigns, crafting influencer programs, ideating creative experiential ideas and, of course, for all manner of marketing storytelling.

Personas haven’t stopped working. But the platforms delivering our marketing don’t really use them.

When you run campaigns today on Google, Meta, Amazon or TikTok, you can upload targeting parameters and audience definitions. But behind the scenes those systems are mostly reacting to behavioral signals—search activity, browsing patterns, content engagement, purchase history and dozens of other variables.

The ad isn’t shown because someone matches a persona. It’s shown because the system has detected behavior suggesting that person might soon need what the brand sells.

This means something subtle but important has happened in marketing strategy.

We still write plans that begin with the sentence: “Our target audience is…” But the systems delivering our campaigns increasingly operate on a different logic entirely. They respond to signals. And signals tell you not who someone is—but what they might be about to do.

The First Wave: Demographics

For most of the twentieth century marketing strategy began with demographic segmentation. Television advertising bought audiences such as “Women 25–54.” Print publications were segmented by income, gender or household composition and media buying decisions largely followed those categories.

Research published in the Journal of Advertising Research estimates that more than 70% of media planning decisions in the 1990s relied primarily on demographic segmentation. 

The system worked because media itself was segmented. If you bought ESPN you were buying men. If you bought a home magazine you were buying homeowners. It was tidy. It was measurable. It was also wildly imprecise.

Two people of the same age and income can have almost nothing in common besides the fact that they were born in the same decade. Demographics helped marketers find audiences. They did not necessarily help them understand people.

The Second Wave: Personas

Personas emerged as an attempt to fix that problem.

Instead of targeting “Men 35–50,” marketers created narrative archetypes describing motivations and lifestyle—urban professionals, ambitious founders, health-conscious parents.

Today 93% of companies report using buyer personas in their marketing strategy according to research from ITSMA and Cintell, and 56% of marketers say personas improve campaign effectiveness according to HubSpot’s State of Marketing report.

Personas improved how marketing teams think about messaging, storytelling and creative strategy. They remain particularly useful in areas such as influencer partnerships, experiential marketing and social storytelling where understanding motivations and identity still matters.

A good persona helps a creative team understand how a brand should sound, what it should say and where it should show up culturally. But personas describe who someone is. They don’t capture what someone is doing right now.

The Third Wave: Signals

Signals are behavioral clues that reveal curiosity, interest or emerging demand that are derived from search queries, browsing patterns, content engagement, product comparisons, location data, purchase history, etc. Instead of inferring behavior from identity, signals allow marketers to infer intent from behavior.

Research from Google Consumer Insights shows that 70% of consumers now use multiple channels during the path to purchase, while Think with Google reports that more than 60% of purchase journeys begin with online research rather than advertising exposure.

In other words, people often begin making decisions long before brands know they are in the market.

Consider travel. Someone searching for flights, comparing luggage sizes and reading packing guides is sending a pretty clear signal about what might be coming next. 

Travel platforms have learned to respond to those signals quickly. Booking.com, for example, runs more than 1,000 concurrent experiments across its platform, continuously adjusting recommendations based on search behavior, browsing patterns and booking signals. The system is responding not to a demographic profile but to behavior.

Signals and the Rise of AI Discovery

AI search and recommendation systems accelerate this shift. Algorithms interpret signals across thousands of behavioral variables simultaneously.

Salesforce’s State of the Connected Customer reports that nearly 60% of consumers now discover new brands through social feeds or algorithmic recommendations. Google research shows more than 40% of Gen Z consumers begin product discovery on TikTok or Instagram rather than traditional search engines.

These systems operate less like advertising channels and more like behavioral radar. When someone watches multiple running shoe reviews, reads marathon training articles and compares footwear specifications, platforms infer a high probability of interest in running gear.

Retailers that understand those signals can respond quickly. The sporting goods retailer Decathlon, which operates a digital ecosystem with more than 100 million active users globally, analyzes browsing behavior and content engagement to detect emerging sports interests before purchases occur.

Similarly, Zalando reports that recommendation systems influence more than 60% of transactions on its fashion platform, largely driven by behavioral signals such as browsing patterns and saved products.

Signals Across Entire Customer Ecosystems

Behavioral signals are now embedded across many digital ecosystems.

When someone uses IKEA’s online room planner, saves furniture combinations or browses design inspiration, those actions signal an upcoming home furnishing project. IKEA reports that more than 60% of customers begin their home furnishing journey online before visiting stores.

Design platforms see similar patterns. Canva, which now serves more than 170 million monthly users globally, tracks template usage, design behavior and editing activity to recommend new products and templates based on signals of creative intent.

Retail platforms respond to signals in real time as well. Sephora reports that more than 80% of online purchases involve personalized recommendations triggered by browsing behavior, product comparisons and engagement with reviews.

Even transportation platforms rely on behavioral signals. The Southeast Asian super-app Grab, serving more than 35 million monthly users, triggers offers and service recommendations based on location signals, purchase patterns and time-of-day behavior.

These systems operate less like traditional advertising campaigns and more like behavioral detection engines.

Why Signals Move Marketing Upstream

Signals often appear long before explicit purchase intent.

Someone researching kitchen renovations may browse design ideas months before contacting a contractor. A traveler may explore destinations weeks before booking a flight.

A widely cited consumer study from GE Capital Retail Bank found that 81% of shoppers conduct online research before making major purchases.

Brands that detect these early signals have an opportunity to influence decisions before the final purchase stage.

Plant-based food brand Oatly has leaned heavily into this dynamic by focusing on content engagement signals related to plant-based diets and recipe exploration, using those signals to introduce its products during early stages of consumer curiosity rather than waiting for explicit dairy-alternative searches.

The difference is subtle but important. Traditional marketing often waits for consumers to declare intent. Signal-driven marketing identifies curiosity earlier.

That is how brands enter the consideration set before competitors even know a customer exists.

Where Most Marketing Teams Actually Are

The reality is that marketing departments now operate across multiple targeting models simultaneously.

Demographics still dominate legacy media planning such as television and outdoor advertising. Personas still guide storytelling, influencer partnerships and experiential campaigns. Signals increasingly shape digital distribution. That shift is reinforced by how digital advertising is bought.

According to the Interactive Advertising Bureau, more than 70% of global digital advertising is now purchased programmatically, meaning algorithms use behavioral signals to determine which ads appear to which users.

Marketing strategy still begins with audience definitions. But the systems distributing those messages increasingly rely on signals.

The New First Question in Marketing

So if most marketing plans begin with: “Our target audience is…,” signals suggest a different starting point: “What behaviors suggest someone may soon need what we sell?”

Demographics helped marketers understand populations. Personas helped marketers understand people. Signals help marketers understand moments. And in an era increasingly shaped by AI search, algorithmic discovery and behavioral data, moments are where marketing influence increasingly lives.

Sources: Journal of Advertising Research, HubSpot State of Marketing, ITSMA / Cintell Persona Research, Google Consumer Insights, Think with Google Research, Salesforce State of the Connected Customer, PwC Global AI Survey, IAB Programmatic Advertising Report, GE Capital Retail Bank Consumer Study, Booking.com Engineering Blog, Sephora Innovation Reports, Decathlon Digital Ecosystem Data, Mercado Libre Investor Reports, Canva Company Data