Picture two homes on the same street. Same square footage, same year built, same tidy lawns. One of those owners is going to call a listing agent in the next four months. The other is staying put for another decade.

From the curb, you cannot tell them apart. But in the data, they look nothing alike.

That difference is the whole game, and most agents never look. They wait until a sign goes up or a home valuation form gets filled out, which is the moment every other agent in town finds out too. By then you are not predicting a seller. You are competing for one.

The agents who stay busy in a tight market do something different. They read the quiet signals a homeowner gives off long before they list, and they show up first. This is how that actually works, what the signals are, and why the data behind them is more reliable than most agents assume.

Why "they just listed" is already too late

Here is the uncomfortable timeline. By the time a homeowner fills out a valuation form, requests a market analysis, or interviews an agent, they have usually been thinking about selling for months. The decision happened in the kitchen weeks ago. The form is the last step, not the first.

And the data on what happens next is brutal for the agent who arrives late. According to NAR, 80% of sellers contact only one agent before signing a listing agreement. Not three. One. Whoever is in the room first, with a real reason to be there, usually walks out with the listing. We broke that dynamic down in detail in our piece on why 80% of sellers hire the first agent they meet.

So the question is not "how do I get more seller leads." Plenty of articles answer that, including our own ranking of lead generation companies by cost per closed deal. The better question is "how do I know who is about to sell, before they tell anyone." That is a data problem, and it has a data answer.

The signals that actually predict a sale

A "signal" is any data point that correlates with a homeowner being more likely to sell soon. Some are obvious in hindsight. Some hide in public records that anyone can pull. Here are the ones with real predictive weight, roughly ordered by how strongly each one moves the needle on its own.

Read those again and notice something. Each one, alone, is a weak filter. Loads of people have high equity and never move. Plenty of homes sit absentee for twenty years. A single signal hands you a giant list of maybes. Which brings us to the part that actually matters.

The math changes when signals stack

One signal is a hunch. Several signals on the same property is a pattern. And the research on what happens when you stack them is genuinely striking.

BatchData analyzed predictive models and found that moving from single-signal targeting to multi-signal targeting lifted conversion from 3.4% to 22.6%. That is a 6.6x improvement, from the same raw pool of homeowners, purely by requiring more than one signal to line up before you reach out. SmartZip's three-year study reinforced it from a different angle: areas targeted with predictive scoring saw 27% annual turnover, against a 5% national average. The homes were not special. The targeting was.

Think about why this works. Picture an owner who has high equity, AND has owned the home for nine years, AND lives in another state, AND recently inherited the property. Any one of those describes millions of people. All four describing the same person describes a very short list, and nearly everyone on that list is going to sell soon. You have filtered a haystack down to a handful of needles.

This is exactly what the expensive predictive platforms do under the hood. They ingest public and property data, look for overlapping signals, and hand you a score. The mechanism is sound. The markup is the only thing worth questioning, which we will get to.

Where the data actually lives

None of this requires inside information. The signals sit in places anyone with the patience to look can reach.

• Equity and tenure come from county records and tax data. Purchase date, purchase price, current assessed value, and any recorded mortgages tell you roughly how long someone has owned and how much equity they hold.
• Absentee status shows up as a mismatch between the owner's mailing address and the property address in those same public records.
• Expired and withdrawn listings live in MLS history, which you already have access to as a licensed agent.
• Vacant status surfaces through postal data, utility flags, and on-the-ground observation in your farm area.
• Life events appear in public filings: probate court records, divorce filings, recorded deed transfers, and obituaries for inherited property.

The expensive part of predictive platforms is not the data. It is the convenience of having it scored and stacked for you. SmartZip and similar tools start at several hundred dollars a month for exactly that. The data underneath is the same raw public record that a one-time data pull can deliver for a fraction of the price. We ran the full cost comparison across channels in our 2026 real estate lead generation statistics report, and signal-stacked outreach lands at $300 to $700 per closed deal, against $2,500 to $8,000 for portal leads.

Reading the signals before the competition shows up

Timing is the part that turns this from a research exercise into closed listings. Some signals are slow and some are fast, and they call for different speeds of response.

Equity, tenure, and absentee status are slow signals. They are visible months ahead and they do not expire. These let you build a watch list and warm up relationships over time, so that when the owner is ready, you are already the familiar name.

Expired listings and life events are fast signals. An expired listing is being worked by every agent with the same data feed by the next morning, which is why speed wins it. The same urgency applies to fresh probate or divorce filings. Reaching these owners early, with empathy rather than a pitch, is the difference between being first and being forgotten.

The agents who win consistently are not the ones with secret data. They are the ones who reach the right owner during the quiet window between "thinking about it" and "listed it." That window is where the entire advantage lives, and it closes the moment the home hits the market. The pattern shows up again and again among the agents who are thriving right now: they stopped chasing listings and started predicting them.

It is worth saying plainly that AI search is starting to play in this space too. As we covered in how agents get found in ChatGPT, more sellers now start with an AI assistant. But even there, the winning move is the same one this whole article is about: reaching the homeowner before they open the search bar at all.

Frequently Asked Questions

Can you actually predict which homeowners are going to sell?

Not with certainty for any single home, but the data shows you can predict it well enough to act. Homeowners carrying several seller-intent signals at once sell at dramatically higher rates than the general population. SmartZip's three-year study found 27% turnover in predictively targeted areas versus a 5% national average, and BatchData found that layering signals lifted conversion from 3.4% to 22.6%. No single signal is a guarantee, but stacked signals turn guessing into informed targeting.

What are the strongest signals that a homeowner is about to sell?

The signals with the most predictive weight are an expired or withdrawn listing, high equity of 40% or more, long ownership tenure of 7 or more years, absentee or out-of-state ownership, vacant status, and life-event triggers such as inheritance, probate, divorce, or a change in household size. The predictive power multiplies when several appear on the same property at the same time.

Why is stacking signals better than using one seller list?

A single signal casts a wide net with low accuracy. Plenty of people have high equity and never move. The odds change sharply when signals overlap. An owner with high equity who has also owned for nine years, lives out of state, and just inherited the property is far more likely to sell than someone flagged by any one of those alone. BatchData's research showed multi-signal models converting at 22.6% versus 3.4% for single-signal targeting, a 6.6x improvement.

How far in advance can these signals predict a sale?

Predictive signals typically surface months before a home hits the market. Ownership tenure, equity position, and absentee status are visible in public and property data long before an owner calls an agent. Life events and expired listings are more time-sensitive and call for faster outreach. Reaching an owner during this pre-list window is what lets an agent become the first, and often only, agent the seller speaks with.

Where do agents get the data to find these signals?

The data comes from public records, MLS history, and property data providers, the same raw inputs that expensive predictive platforms repackage and resell. Equity and tenure come from county and tax records, absentee status from mailing-address mismatches, expired listings from MLS history, and life events from public filings. Agents can pay several hundred dollars a month for a platform that scores it, or pull and stack the signals themselves at a fraction of the cost.