A Click Is Behavior. A Question Is Context.

One visitor opens a listing, scrolls through the photos twice and leaves. Another asks about the roof, renovation history, monthly ownership costs and whether a Saturday tour is available.

Both may appear as activity in an analytics system. But only one has explained what matters to them and what they want to do next.

Useful intent does not come from collecting the largest possible trail of behavior. It comes from preserving the context a visitor chooses to express.

That distinction is central to how real estate should think about buyer intent. The goal is not to decide who a buyer is. It is to help an agent understand what the buyer is trying to accomplish.

Attention is not intent

Real estate has become comfortable measuring attention. Listing views, photo clicks, repeat visits, time on page and email opens are all easy to count. They can help a brokerage understand aggregate performance, compare campaigns or identify friction in a digital experience.

At the individual level, however, these signals are often ambiguous.

A long visit could indicate serious interest. It could also mean the tab was left open. A repeat visit could reflect growing intent, a price check, professional research or simple curiosity. A gallery click tells us that someone viewed a photo, not what they noticed or whether it changed their decision.

Behavioral data is not useless. It is simply weaker than it often appears. The strongest signals tend to be the ones buyers provide deliberately.

A question carries more meaning than a score

Consider two notifications. The first says:

New lead. Jane Smith viewed 214 Oak Street. Lead score: 82.

The score looks precise, but the agent cannot see what produced it. Was it the number of page views? The visitor’s device? A marketing attribution rule? A model trained on a different market? Even if the score predicts conversion reasonably well across a large population, it gives the agent little help with the next conversation.

The second says:

Jane asked when the roof was replaced, whether the converted room was permitted and what the HOA fee includes. She requested a Saturday afternoon tour. The permit question remains unresolved.

This brief does not claim that Jane is high intent, financially qualified or likely to make an offer. It does something more practical. It tells the agent what Jane is considering, what she still needs and the next action she requested.

This matters in real estate because digital discovery and professional representation are not competing paths. They are connected parts of the same buying process. In NAR’s 2025 Profile of Home Buyers and Sellers, 88% of buyers purchased through an agent or broker. The most common reason was help finding the right home, followed by support with negotiations and paperwork. Buyers look to agents for help evaluating properties and moving the purchase forward.

The same report notes that the typical buyer searched for 10 weeks. Context accumulates across that period. Click-based analytics reveal activity. A system designed around questions can preserve more of the reasoning behind that activity.

Preserve the interaction, present the decision

Capturing useful context does not mean forwarding a complete transcript or an exhaustive activity log to the agent. The source interaction may need to remain available for accuracy, review and correction. But what the agent sees first should be shaped around the next decision.

A useful handoff can be concise:

• Property: the listing involved
• Buyer questions: what the visitor explicitly asked
• Priorities or constraints: only what the visitor volunteered
• Resolved information: the answer and its property source
• Open items: questions the system could not answer reliably
• Requested next step: tour, call, document or follow-up
• Timing: only when stated or selected by the visitor

Retention and presentation are not the same thing. A system may retain enough source context to support traceability while presenting only what helps the next person act.

Volunteering something in a conversation is not blanket permission for every later use of it. A visitor should be able to understand what will reach the agent, and what is retained should stay relevant to the follow-up they asked for. Context gathered for a tour request does not become a general profile because it was convenient to keep.

The agent should also be able to inspect the basis of the summary. If the brief says a buyer is concerned about monthly costs, the underlying questions should support that statement. If the visitor corrects an answer or changes a preference, the handoff should change with it. Summarization should reduce noise, not manufacture certainty.

The line between relevance and profiling

Once a system can interpret questions and behavior, it becomes tempting to infer more than the visitor actually said.

A question about a school commute does not establish whether the visitor has children. A request for a ground-floor bedroom does not reveal age or disability. A search across several neighborhoods does not justify assumptions about race, religion, family status or national origin. None of these inferences is necessary to answer the property question, and in housing they create real fair-housing and privacy exposure.

In a Pew Research Center survey of U.S. adults in May 2023, 67% said they understand little to nothing about what companies are doing with their personal data. That is a general attitude to company data practices rather than a measurement of home buyers, but it describes the environment an unexplained score enters.

The Fair Housing Act prohibits housing discrimination based on race, color, national origin, religion, sex, familial status and disability. HUD’s April 2024 guidance on advertising through digital platforms is direct about automated systems: intentional discrimination can include using a protected characteristic, or a proxy for one, as the basis for a decision, and that is true “even if that decision is made in whole or in part by an automated system, including a system using machine learning or another form of AI.”

Although that guidance specifically addresses digital advertising and audience delivery, its treatment of protected characteristics and proxy variables illustrates the risk of using inferred identity to shape any housing-related interaction. HUD describes proxies as variables closely correlated with protected classes, with ZIP code and race as one example, and names several that look innocuous on their face.

Several boundaries follow:

• Do not infer protected characteristics from questions, location, language, names or browsing patterns.
• Do not characterize neighborhoods in ways that steer buyers toward or away from an area.
• Do not use opaque qualification labels the agent cannot inspect.
• Do not present predicted intent as a fact about an individual.
• Do not collect information merely because it might become useful later.

NIST’s AI and privacy risk-management frameworks offer a principle that fits this workflow: risk is managed across a system’s design, use and outcomes, not bolted on as a disclosure at the end. In practice that means knowing what is collected, why it is needed, how an inference is produced, who can see it and how a person can correct it.

The responsible alternative is not to discard context. It is to use the minimum necessary for the buyer’s stated purpose, and to keep the meaning of that context visible.

What a responsible handoff looks like

Illustrative example.

A visitor arrives at a listing late in the evening. They ask whether the home has central air, when the water heater was replaced and whether the downstairs room could function as an office. The approved property information confirms central air and records the water heater installation date. The available material does not establish the legal use or dimensions of the downstairs room. The visitor then asks for a Sunday morning tour.

That gives the agent a productive opening:

Thanks for asking about Sunday morning. Let me check availability and confirm a time with you. In the meantime, I can find out the downstairs room’s dimensions and whether it is permitted as a bedroom, since you were considering it as an office. Anything else you would like me to look into?

That response is more relevant because the buyer supplied the context. It is also more human. The agent enters the conversation prepared to help, rather than asking the buyer to repeat everything or opening with a generic script.

Measure whether the context improves the conversation

If the purpose is a better handoff, page views and form submissions are not enough to judge performance. More useful measures include:

• First-response quality: did the agent address the stated question or requested action?
• Repetition avoided: did the buyer have to restate what they had already provided?
• Open-item resolution: were unanswered property questions followed through?
• Appointment progression: did a requested tour or call move forward?
• Correction rate: how often did the buyer or agent have to correct the brief?
• Net review burden: did the system save time after the effort of verifying it?

These keep the system accountable to the workflow. A summary that is elegant but frequently corrected has not improved the handoff. A score that predicts response while encouraging agents to ignore other buyers creates a different failure. A system that captures more data while making the conversation no better has optimized the wrong thing.

Better context, not more surveillance

The first conversation after a digital inquiry should not feel as if the buyer has been watched. It should feel as if they have been heard.

That is the standard buyer-intent systems should meet. Use explicit questions before ambiguous clicks. Use volunteered priorities before speculative inference. Give the agent a concise, inspectable brief rather than a black-box label. Carry the buyer’s requested next step into the human workflow.

The visitor who asks about the roof, the monthly costs and a weekend tour has already explained what matters.

The system does not need to invent intent. It needs to preserve it.

Sources

• National Association of REALTORS®, 2025 Profile of Home Buyers and Sellers, November 2025; primary-residence purchases, July 2024 to June 2025.
• Pew Research Center, How Americans View Data Privacy, 18 October 2023; survey of U.S. adults conducted 15–21 May 2023.
• U.S. Department of Housing and Urban Development, Office of Fair Housing and Equal Opportunity, Guidance on Application of the Fair Housing Act to the Advertising of Housing, Credit, and Other Real Estate-Related Transactions through Digital Platforms, 29 April 2024.
• National Institute of Standards and Technology, Artificial Intelligence Risk Management Framework (AI RMF 1.0), January 2023.
• National Institute of Standards and Technology, Privacy Framework: A Tool for Improving Privacy through Enterprise Risk Management, Version 1.0.