What ‘intent’ means in 2026, and why the machines won’t wait for your product to catch up
BY HAZEL BROADLEY, BEELER.TECH
Picture the scene. A man walks into a meeting in a business suit, a tutu, and a scuba tank. On paper, he defies every demographic label you could reach for. But what he cares about, and what he is trying to do right now, is completely legible – if you know how to read the signals.
If you look closely at this recent London Underground ad, that’s exactly the scenario Nano Interactive’s recent intent campaign brings to life:
And it’s also the central problem with how advertising has talked about ‘intent’ for the past decade.
Intent has been treated as a targeting label rather than a human truth. The gap between the two is where publisher revenue leaks, buyer trust erodes, and campaigns quietly underperform.
Intent remains the strongest measurement signal in the room, with 97% of marketers agreeing that “intent data brings a competitive advantage.” But that edge disappears fast when nobody in the room can agree on what they’re actually measuring.
And it’s about to get harder. Agentic AI is reshaping how ad buying works: bots don’t browse, and they don’t click ads. Intent signals that can’t be read by a machine won’t survive that shift. More on that shortly.
Stop using ‘intent’ as a magic word
Most of us have sat through the same meeting. Someone talks about ‘intent’ charts. Everyone nods, because nobody wants to be the person who asks exactly what that means. Then the deck moves on, the line item gets created, and the word ‘intent’ quietly becomes a substitute for thinking.
Intent has been treated as a targeting label rather than a human truth. The gap between the two is where publisher revenue leaks, buyer trust erodes, and campaigns quietly underperform.
Niall Moody, Chief Revenue Officer at Nano Interactive, puts it plainly:
“Intent is one of those buzzwords that gets dropped into conversations all over the place. Everyone talks about it, but not everyone means the same thing. Without that shared definition, intent has become a catch-all label for a wide range of signals, leaving buyers struggling to identify audiences who are genuinely ready to act at that moment in their customer journey. That lack of clarity matters, because when intent can mean everything, it risks meaning very little.”
In practice, intent almost never shows up as one clean signal, but rather a bundle of signals with different freshness, provenance, and rights, which makes it worth defining before you try to measure it.
Why demographics were never enough
The industry spent decades using identity as a proxy for intent, and the proxy was always leaky. Traditional demographic targeting would drop someone into a box – say female, 30-44, suburban – and push luxury beauty products. But real people are more complex. Their passions, priorities, and intentions shift constantly, and no label captures that.
The more interesting challenge is understanding why people engage with content, not just what they clicked. Seedtag answers this through neuro-contextual advertising, which uses neuroscience principles to go beyond the click and ask what emotional state the viewer was in when they engaged, and what that tells you about their next move.
According to Seedtag, the three-principle framework is grounded in how attention and memory actually work:
- Genuine interest captures attention.
- Real emotion improves recall.
- Active intent is what actually moves someone to act.
Each layer is a better predictor than the demographic label it replaced. The catch is that better signals come with a shorter shelf life, and that’s where most publishers are currently losing the argument.
Intent is one of those buzzwords that gets dropped into conversations all over the place. Everyone talks about it, but not everyone means the same thing. Without that shared definition, intent has become a catch-all label for a wide range of signals, leaving buyers struggling to identify audiences who are genuinely ready to act at that moment in their customer journey. That lack of clarity matters, because when intent can mean everything, it risks meaning very little.
Not all intent signals age the same way
Someone searching for ‘mortgage rates’ or ‘best running shoes’ is in a conversion-ready (or ‘short-horizon’) mode. That signal is valuable precisely because it sits close to an action, but target it a month later, and you’re simply targeting the memory of intent.
Consideration (or ‘long-horizon’) signals like affinity categories, consistent topic interest and repeat visits still matter, but they should be labeled honestly. Buyers who pay ‘in-market’ prices for them will eventually notice.
Freshness should be a product requirement, not a nice-to-have. Be ready for buyers to ask how a segment was made, how often it refreshes, what the lookback window is, and which ID currencies are involved. If you can answer those questions quickly and consistently, you’re selling something that holds up under scrutiny.
If you can’t, you’re selling a label and hoping nobody calls your bluff.
Separate what users say from what they do
There are four types of intent signals, and they are not interchangeable:
- Declared intent is the cleanest form: registration fields, preference centers, surveys, newsletter choices, and explicit ‘notify me’ behaviors. It’s the hardest to scale but the most honest.
- Observed intent comes from the logged behaviors most publishers actually have: pages viewed, dwell time, scroll depth, repeat visits, video completions and product clicks.
- Inferred intent turns that behavior into a probability through modeling like ‘read three mortgage articles in seven days, probably in-market.’ Useful, but requires more honesty about what you actually have.
- Modeled extension is where trust gets tested. Lookalikes and propensity scores extend your reach, but if you can’t explain the inputs or the refresh cadence, you’re asking a buyer to accept a black box and pay a premium for it.
It’s also worth mentioning that buyers have intent, too. An RFP, a brief, or even how a buyer describes your site, is a signal. If you don’t capture and normalize those signals, you end up guessing what buyers want and calling it a product strategy.
Map intent to the journey without lying to yourself
Publishers usually get pulled into intent conversations when a buyer wants performance. The trap is equating intent with ‘buying right now,’ because that’s where the market tends to price the highest.
However, only 5% of any market is genuinely ‘in-market’ at a given moment. Price yourself around that 5%, and you shrink your value to the last click. A full-funnel view of intent matters.
- Awareness intent is about contextual alignment, broad interests, and top-of-funnel signals.
- Consideration intent is active research: on-site search, repeat visits to comparison content, and deeper engagement.
- Decision intent is high precision – pricing pages, add-to-cart behavior, deterministic signals in authenticated environments.
- Retention intent, tied to repeat engagement and churn risk, is its own category and often worth more than people realize.
The open web is where discovery and comparison happen. It’s where early-stage public intent lives. If you let the market define intent as only the last mile, then you can’t take credit for what happens before that point. Worse, you’re handing the walled gardens the high ground by default.
Make peace with contextual again
Contextual is back, and it’s back for a reason.
When third-party IDs stopped being reliable, the page itself became the targeting layer again. Historically, ‘contextual’ meant a thin layer of keywords taped onto a URL. Take ‘salsa’: keyword matching can’t tell you whether it’s a dance or a food. But concept-based classification can, and that distinction matters. Today’s systems successfully interpret meaning, not just match words.
The more sophisticated platforms go further. They build an intent graph – a live map that pulls in billions of signals, from previous campaign performance to live contextual signals, content sentiment and emotion analysis. Those combined dimensions model outcomes for advertisers, moving beyond the single dimension of cookies and identity.
As Moody explains:
“Marketers are now leveraging technology that performs complex data modelling to build bespoke audience segments from no less than 4.9 billion live intent signals across the open web – all without leaning on personal IDs. Metrics such as brand suitability, time of day, device, context, sentiment and emotion are all evaluated in real time and converted into a live decision that delivers a defined outcome for the advertiser. Together, these inputs help define intent as it exists in that exact moment.”
Productize intent without leaking it everywhere
Having intent signals is not the same as having an intent product. That distinction is where most publishers quietly lose the argument, and the margin goes with it.
The mechanics are important, because intent travels through auction-level signaling via standard bid objects. Header bidding can pass first-party data with permissions, and curated audiences scale first-party intent without data leakage or reliance on deprecated IDs. But if you don’t know how your inventory is being packaged and described by the curators and SSPs downstream, someone else is defining your product before it reaches a buyer.
The open web is where discovery and comparison happen. It’s where early-stage public intent lives. If you let the market define intent as only the last mile, then you can’t take credit for what happens before that point. Worse, you’re handing the walled gardens the high ground by default.
The fix is less technical than it sounds: it’s all about consistency.
And it’s the same answer every time someone asks how the segment was built, how often it refreshes, whether it was observed or modelled. That consistency is what turns a signal into something worth a premium. Without it, you’re just hoping the buyer doesn’t ask the right questions.
Automate so intent doesn’t die in a spreadsheet
Automation and AI often get conflated, but they’re not the same thing. Automation makes the known process efficient. AI helps you deal with messy inputs, surface patterns, and find what you’d otherwise miss. Agents orchestrate tasks across systems, and they’re coming whether you like it or not.
When RFPs arrive in inconsistent formats, sales notes live in inboxes, and the logic behind your best deals exists only in the head of the person who closed them, you can’t see the patterns. And if you can’t see the patterns, you can’t build a product around them.
Automate the grunt work, like mapping the buyer language to how inventory actually performs. Then your team can focus on what actually requires judgment: what to build, what to price, and what to walk away from.
Prepare for the agentic era, because the clock is speeding up
The next wave of intent will be shaped by machines talking to machines.
Agentic systems are already emerging that can plan and optimize campaigns with minimal human involvement. Protocols like AdCP, UCP, and ARTF are emerging as the shared language for those agents. Whether these standards win or not, the direction is clear – more decisions will happen faster, closer to the impression, and with less patience for ambiguity.
As Moody puts it:
“The evolution of intent‑driven strategies is closely tied to advances in machine learning and, increasingly, agentic AI. We’re already seeing the impact of AI‑fuelled engines that can analyse content, capture engagement signals, assess sentiment and deliver contextual alignment at scale.”
The structural challenge for the open web runs deeper. Zero-click search is pushing answers into AI summaries, with some reports citing web traffic declines by nearly 60%
As I mentioned earlier, bots don’t browse, and they don’t click ads. They hit APIs, compare structured information, and make decisions quickly. Influence moves upstream toward trusted brands, clear claims, and machine-readable data. If you want your content to keep driving demand, you need to think about what an agent can understand, not just what a human will read.
This is where intent stops being a targeting feature and turns into a survival tactic. The clock speeds up, and the value of fresh signals goes up with it. If your intent product can’t be inspected, governed, or constantly refreshed, it won’t be usable in an agentic workflow.
How the industry is building intent activation
The gap between intent theory and intent activation is where a lot of good strategy goes to die. Four platforms are doing interesting work to close the gap:
- Nano Interactive focuses on ID-free intent signals, analyzing billions of live contextual inputs across brand suitability, device, time of day, sentiment, engagement, converted into real-time targeting decisions. Nano’s agentic media planner, launched in March 2026, streamlines planning workflows by using autonomous agents to interpret campaign goals, brand parameters, and sector signals from a natural language brief – before recommending targeting strategies. The aim is to reduce manual planning time while keeping human control over audience composition and activation.
- Seedtag has taken a complementary route with Liz Agent, also launched in March 2026. Built on top of Seedtag’s neuro-contextual intelligence engine, Liz acts as a conversational planning interface that connects research, audience mapping, and campaign activation. Rather than relying purely on static contextual categories, it analyzes emotional and cognitive signals to identify moments when audiences are most receptive. Planners can move from insight to activation through natural conversation, with recommendations.
- DoubleVerify’s MRC-accredited solutions combine keyword analysis, URL classification, and suitability controls to help advertisers scale privacy-safe reach while maintaining brand alignment. As a result, Miele saw a 50% higher CTR rate than the benchmark, FanDuel a 41% reduction in cost per action, and Vodafone a 54% lower cost per acquisition.
GumGum brings together contextual analysis, attention measurement, and creative optimization within a single system. Its Mindset Graph, a predictive engine designed to map consumer mindsets to advertising goals by analyzing the intersection of content context, user attention, and creative format – the latter of which is vital as 41% of placements ‘miss the mark’ without creative alignment.
Intent is a product, not a dataset
If you take nothing else from this, take this: intent is a product family, not a single dataset.
Contextual intent gives you broad reach with lower personal data dependency. Engagement and behavioral intent give you mid-funnel lift if you disclose how you built it. Commerce and purchase-adjacent intent can be powerful when you have the right partnerships and measurement, and deterministic authenticated intent can command premium pricing when you can activate it safely.
Once you accept that, the operating model is practical: collect signals with clear permission, classify them in ways you can explain, package them to fit buying workflows, and govern them like the product line they are.
Now comes the part that requires backbone: decide what you’ll demand from partners
And what you won’t accept. Understand how segments were built, how often they refresh, whether they were inferred or modeled, and what happens when an agent makes a decision you disagree with.
The man in the tutu doesn’t fit in a box. He never did. The industry has spent years pretending otherwise – and losing revenue in the gap between the label and the reality. Intent, done properly, closes that gap. It reads what people actually care about, in the moment they care about it.
You can either build a product around that truth, or keep selling a demographic proxy and calling it targeting. The clock is running. Which one are you choosing?