I asked an agent last month how many people were in her database. She said about twelve hundred.
I asked how many of those twelve hundred she had talked to in the last ninety days.
She thought about it for a second. Then she said maybe thirty.
Eleven hundred and seventy people who know her. Who have done business with her, or were referred to her, or sat across from her at a closing table at some point in the last several years. Almost none of them have heard from her this year.
That is not a database. That is a list of names that used to be a database, sitting in a CRM, waiting for someone who isn't coming back to it.
The List That Calls Itself a Database
Here is the test. Pull up whatever you call your database right now. Look at the last contact date on the first twenty names. Not the date you added them. The date you last talked to them, texted them, mailed them something, anything.
If you are like most agents, a chunk of those dates do not exist at all. The field is blank. You have never logged a single touch with someone who is sitting in your database right now.
The National Association of Realtors has been tracking this for years, and their research consistently shows that somewhere around forty percent of sellers either used the agent they worked with before or were referred to one by someone they trust. That is the size of the repeat and referral pie. Most agents I talk to are converting a fraction of that, not because their past clients forgot them, but because their past clients never heard from them again after closing.
A real database is not a list of names. It is names plus context plus a plan for staying in touch. Take away any of the three and what you have left is a phone book.
Context means you remember that Sarah's daughter just started kindergarten, that Mike is two years from retirement and has mentioned downsizing twice, that the Johnsons loved their experience so much they sent a card after closing. A plan means those facts turn into a reason to reach out, on a schedule, whether or not you feel like it that week.
Most agents have the names. Almost none have the other two pieces. That gap is bigger than most agents want to admit.
Why Databases Go Dead in the First Place
It is not that agents refuse to add people to their database. They add people constantly. Every closing adds the buyer, the seller, sometimes both agents on the other side. Every open house adds a stack of sign-in sheets. Every referral conversation adds a name and a number typed into a phone before the lead goes cold.
The database grows. That part works fine. What never gets built is the other half: a system for working back through the list on a schedule.
So the database becomes a one-way street. Names go in. Nothing comes back out. Three years later it has eight hundred or a thousand or fifteen hundred contacts in it, and the agent has talked to maybe two percent of them in the last year, almost always because that person called first.
This is the part that stings. The agent did the hard part. They built the relationships. They earned the trust. The actual database exists. It just sits there, DEAD, because nobody ever built the rotation that keeps it breathing.
Three years is the average. Some databases die faster. An agent who has been licensed two years and closed fifteen deals might already have a hundred and fifty contacts gathering dust, simply because the rotation never existed in the first place. The timeline is shorter when there was never a system to begin with. There is nothing to maintain, because there was nothing built.
I wrote about the rotation itself in detail in the twelve group SOI system, which breaks a database into manageable chunks so you are never staring down the whole list at once. But before any rotation works, the database has to be alive enough to rotate through. A dead database does not need a better rotation. It needs to be woken up first.
A system that works through your database on a schedule, pulls the context you already have on each person, and drafts the check-in message so reviving a dead database stops being a project you never start.
Get the Free SOI AI Employee →What a Dead Database Actually Costs You
Let's do the math nobody wants to do.
Say your database has eight hundred people in it. That is a conservative number for an agent five years into the business. Industry research consistently shows that referral and repeat clients close at a far higher rate than cold leads, and cost a fraction as much to generate. NAR's data puts the referral and repeat share of all transactions somewhere close to forty percent in a normal year.
If even ten percent of an eight hundred person database has a real estate need this year, either their own or someone they know, that is eighty transactions worth of opportunity sitting inside your phone. Eighty. Most agents close somewhere between eight and twenty transactions a year, total.
The opportunity inside a dead database, even a modest one, often dwarfs the agent's entire current production. Not because the people changed. Because the agent stopped showing up, and when the moment came for someone in that database to buy or sell or refer a friend, they called whoever DID show up. An agent running ads. A friend's cousin who just got licensed. A name from a postcard that happened to land the same week.
None of those eighty people decided you were a bad agent. Most of them probably still like you. They just stopped thinking about you, because nothing reminded them you exist. That is the actual cost. Not lost trust. Lost visibility.
The Revival Sequence
Waking up a dead database feels like an enormous project, which is exactly why most agents never start. Twelve hundred names. Where do you even begin.
You begin by sorting, not contacting. Three groups. Group one is anyone who has done a transaction with you, or sent you a referral, in the last three years. This group is small, usually fifty to a hundred and fifty people, and it is the group most likely to do business with you again or send someone your way. Group two is everyone who knows you, has your number, would recognize your name, but has never transacted with you. Old neighbors, gym friends, people from your kid's school, past clients older than three years. Group three is everyone else. Cold contacts, old leads, names you collected somewhere and never followed up on.
Start with group one. Always. This is the highest value group and the lowest pressure conversation. A short message. Not a sales pitch. Something closer to a real check-in: how is the house treating you, are the new schools working out, anything coming up I should know about. Low pressure. Genuinely curious. The kind of message a friend sends, not an agent.
Twenty of those a day, five days a week, gets you through a hundred-person group one in about a week. Group two of five hundred takes five weeks at the same pace. By the end of two months, every person in your database who is realistically going to do business with you has heard from you at least once this year. For most agents, that has not happened in years.
This is not a complicated process. It is a SLOW one, and that is exactly why it works. The agents who try to revive their database in a weekend burn out by Tuesday and never touch it again. The agents who do twenty a day for two months come out the other side with a database that is actually alive.
What Keeps It From Going Dead Again
Revival is not the system. Revival is what you do because there was never a system. Once the database is alive again, the question becomes how it stays that way.
The answer is the same rotation mentioned earlier, but it only works once everyone in it has had at least one recent touch. A rotation built on top of a dead database just rotates through silence. A rotation built on top of a revived database keeps every relationship warm without you having to remember each one individually.
This is also where the gap between using AI and building a system shows up again. I wrote about that difference in a separate post, and the database is the clearest example of it. Anyone can ask ChatGPT to draft a check-in message. Almost nobody builds the structure that pulls the right twenty names for today, remembers what was said last time, and drafts a message that sounds like it came from someone who actually remembers Sarah's daughter started kindergarten.
That is what the SOI AI Employee is built to do. It works through your database on a schedule, pulls the context you already have on each person, and drafts the message so the only thing left for you to do is read it, adjust if needed, and send it. The FUNDAMENTAL work, deciding who to talk to and what to say, gets done automatically. The relationship part, the part that actually matters, stays yours.
Most agents will read this and agree with all of it, and then do nothing, because reviving twelve hundred contacts by hand still sounds like a project for a future version of themselves. The agents who actually do something about it usually start the same way. They pick the smallest group first. Group one. Fifty names. Twenty minutes a day. They start today, not on the Monday after the slow season ends, because that Monday has a way of never arriving.
Built for agents whose database has gone quiet. It runs the rotation, pulls the context you already have, and drafts the messages so the relationships stay warm without relying on memory.
Get the Free SOI AI Employee →