Intent signals have become central to how digital audiences are understood and targeted. But observed behavior does not always reflect durable demand, and as signals fragment across environments, relying solely on intent can introduce uncertainty into planning and measurement.
The case here is for validation: grounding intent in stable, real-world conditions to improve audience clarity, sharpen targeting decisions, and give measurement something reliable to work against.
Targeting Strategies Need an Update
Today’s marketers are increasingly focused on intent: what is this user interested in, and are they a potential customer? Intent is typically inferred from observed behavior — a user reading about heat pumps or downloading a buyer’s guide is assumed to be interested in upgrading their heating system.
In some environments, that assumption holds. Paid search and retail media are built around high-intent moments: 89% of consumers begin their purchasing journey with a search prompt (SEO Guidelines). Intent signals and the point of decision are tightly coupled, making them relatively reliable.
Outside those moments, the picture is less clear. Behavioral signals can overstate demand. People research aspirational products, make one-time purchases — a baby gift for a coworker — and browse on behalf of others. As Intent Data, an intent-data vendor, noted in 2021, activity that looks like purchase research could just as easily be driven by education, entertainment, or curiosity. Nexus B2B warns that overreliance on intent data can lead to premature outreach and wasted spend, particularly when IP-based signals cannot identify the actual in-market person. A 2025 Demand Gen report found that 45% of available marketing data is incomplete, inaccurate, or out of date.
The answer is validation: a reliable way to determine whether observed interest corresponds to a plausible, persistent need. The capability already exists. A user reading about heat pumps in the Upper East Side — a neighborhood dominated by rental apartments — is unlikely to be a buyer. A spike in luxury SUV interest from dense urban cores with low vehicle ownership is more likely aspirational browsing than near-term purchase intent. Grounding digital signals in stable, real-world context filters the noise.
The question is where that stability should come from.
The Audience Definition Gap
Digital advertising is often framed through a set of debates: whether contextual targeting can stand in for intent, whether identity should be built with or without IDs, how to balance privacy with performance. These debates matter. But they tend to obscure a more basic problem: most campaigns still lack a clear, validated definition of who they are trying to reach — audiences defined by interest, ability to purchase, and a plausible, persistent need.
Context is often treated as a proxy for intent not because it is especially accurate, but because it is observable and measurable. It can explain what someone engages with. It cannot explain who they are, whether they can buy, or whether a real need exists. Without those constraints, contextual targeting optimises for engagement, not customers.
The ID debate has the same blind spot. Most IDs capture patterns of digital behavior — where someone goes online, how they move across devices — not whether they can realistically buy the product being advertised. Identity makes it easier to track behavior. Without real-world data to anchor it, that behavior may not translate into demand.
Privacy framing has it too. Treating targeting as a tradeoff between privacy and performance leads to complex solutions — clean rooms, for instance — that can add cost without improving audience clarity. Privacy constraints determine how data can be activated, not whether an audience is valid in the first place. When the audience definition is sound, privacy becomes an implementation concern, not a strategic limitation.
What these debates share is that they attempt to solve audience clarity at the point of execution. But without a method for validating audiences against real-world conditions, every tactic becomes a substitute for strategy. Measurement improves, precision increases, complexity grows — and the underlying question stays unanswered: was the intended audience clearly defined? Until it is, no combination of context, identity, or privacy controls can reliably produce buyers.
Defining a Targeting Model
A targeting model is a method for determining who matters for a campaign before deciding how to reach them. That definition must be grounded in durable needs, not momentary behavior.
Many current approaches rely on demographic proxies — age, gender — to infer demand. Those proxies can exclude real buyers while including people with no sustained need. Caregiving illustrates the problem: the need is not defined by gender or marital status. A gay man caring for an infant has the same ongoing demand for children’s products as any other caregiver.
Intent signals are valuable inputs, but only when evaluated against a stable understanding of who the audience is and whether their need persists over time. Models that prioritise identity labels or isolated behavioral events over durable need can obscure the audience rather than clarify it.
Because durable needs exist outside platforms and identity systems, a sound targeting model holds regardless of platform churn, regulatory change, or shifts in available signals.
Why Households Provide a Stable Basis for Targeting
What unit of analysis is stable enough to validate intent against real-world conditions? The household.
Individual-level targeting relies on signals and identifiers that change across devices, sessions, and platforms. Research, one-time purchases, shared browsing, and proxy actions all inflate apparent intent — producing behavior that looks actionable but may not reflect real demand.
Households reflect durable conditions that change slowly: whether a household rents or owns, lives in a single-family home or an apartment, has children or dependents, relies on a vehicle or public transport, and so on. Many purchasing decisions are household decisions, even when different individuals research, influence, or complete the transaction. The need is shared, even if the behavior is not.
Because households exist independently of devices, platforms, or media environments, their characteristics remain consistent across channels — making the household a stable reference point for evaluating intent without relying on fragile identifiers.
In practice, this changes how signals are interpreted. A user researching minivans may look like a strong prospect. But if that behavior is coming from a household without children, limited parking, or no recent vehicle ownership, the apparent intent is unlikely to convert. The interest is real. The demand is not. Similarly, household-level indicators — broadband infrastructure, technology adoption, digital engagement patterns — help distinguish genuine purchase intent from aspirational browsing when someone is researching premium laptops or smart home devices.
This isn’t about dismissing intent signals. It’s about asking whether that intent can realistically be acted on. Household signals act as a validation layer, filtering out activity that looks meaningful in isolation but lacks the capacity to result in a purchase.
The same logic applies to measurement. When the audience definition is stable, results can be compared across channels without redefining the target each time — measuring performance against the same underlying demand, not a shifting set of behaviors. Strategy, execution, and measurement are often built on different assumptions about who the audience is. Aligning those assumptions at the household level is where waste declines and performance improves.
Measurement as the Feedback Loop
Targeting begins with an audience definition: people who show interest, are able to purchase, and have a real, persistent need. Measurement confirms whether that definition held in practice.
The basic question is: how many of those people did you actually reach? This is where many intent-based approaches fall short. Intent data captures behavioral events, not confirmed audience membership. Much of what digital measurement reports reflects activity generated by advertising itself — not whether ads reached people who actually belonged to the intended audience.
Household-level attributes help close this gap. The same criteria used to define the audience — homeownership, family composition, living situation — can also be used to validate reach. You can confirm that the people exposed to the campaign were, in fact, members of the audience you set out to reach. And that validation doesn’t happen only after a campaign ends — it can occur as a targeting strategy is set, while a campaign is running, and again in post-campaign analysis.
A German travel brand used household-level measurement to validate whether its existing personas still reflected its real audience — personas that had been expensive and time-consuming to create. The result was a campaign launched quickly and economically, without starting from scratch.
Without that validation loop, optimisation relies on assumptions, leading to wasted spend and fragile performance gains.
What Changes When You Reset the Model
Validating intent against a stable audience definition changes more than targeting accuracy. It changes how campaigns are planned, executed, and evaluated.
Planning shifts away from channel-specific tactics toward an audience-first question: who actually matters for this campaign? Because the audience definition is consistent, the same logic can be applied across campaigns, verticals, and time periods. Personas can be reused, tested, and refined rather than rebuilt for each new effort.
Channels become delivery mechanisms, not strategy drivers. The audience definition stays the same whether intent shows up in search queries, display behavior, video consumption, or CTV viewing. What changes is how the message is delivered, not who it is meant to reach. That makes channel performance easier to interpret — differences can be evaluated without redefining the audience for each environment.
The deeper advantage is proof. With a stable audience definition in place, intent-based hypotheses can be tested directly. Marketers can assess whether intent signals correspond to real buyers, not just higher engagement. Because validation occurs throughout the campaign lifecycle, targeting can be continuously refined rather than periodically reset. Over time, this builds a cumulative understanding of what works — and evidence that campaigns reached the right people, not just performed.
Moving Forward
Focusing on intent is a rational strategy. Reaching people when they are actively researching or expressing interest will always matter. But intent alone can overstate demand, and uncertainty introduced at the audience definition stage cannot be fixed by optimisation alone.
Validation is the missing step — confirming whether observed interest maps to a viable audience before activation, during execution, and after campaigns conclude. Household-level attributes derived from verified public sources — census records, national statistics, tax data — provide a stable, real-world foundation for that confirmation: the same criteria used to define the audience can be used to verify who was actually reached.
The result is planning that transfers across campaigns, measurement that reflects real demand, and evidence that campaigns reached the people who actually mattered.