They Said Not Right Now. That Was Three Years Ago.

In the spring of 2023 somebody filled in the home valuation form on your website. They wanted a number. You called twice, they said they were still thinking about it, and the thread went quiet the way most threads go quiet.

They are still in your CRM. Their number still works. Nobody has dialed it since.

Last year they sold that house. Somebody else listed it.

Nothing was neglected in that story, which is exactly what makes it expensive. No ball was dropped and nobody was lazy. It is that following up with a person who said not right now, three years later, in the one month it stopped being not right now, is not a thing a human being is built to do.

What is actually sitting in that database

Everybody says the money is in the database. Almost nobody says what the money is made of, so here is the honest version.

The National Association of REALTORS surveys recent buyers and sellers every year and asks, among a hundred other things, how they found the agent they hired. The most recent generational cut of that survey is worth looking at properly, because the two biggest answers are not marketing.

Two thirds of sellers hired somebody who was already known to them: referred by a friend, a neighbor or a relative, or an agent they had worked with before. Four percent found their agent on a website.

Read the small bars too, because they are the honest half. Four percent of sellers hired an agent who had contacted them directly. Cold outreach is not what fills a listing calendar, and this piece is not going to pretend otherwise.

What that leaves is an interesting middle. The people sitting in your CRM are not strangers and they are not friends. They are the group that raised a hand once, at a moment when the answer turned out to be later. They already know your name, they contacted you first, and they are the only group anywhere near the top of that chart that you actually hold a list of.

Why nobody works it

Ask an agent when they last went through their old leads properly and you get a wince.

It is not laziness and it is not a discipline problem. It is that the work is genuinely awful. Hundreds of conversations to find the handful of people whose life has changed, most of them going to voicemail, a few of them annoyed, and the majority of the ones who do answer saying no for the second time. The reward for doing it properly is getting to do it again next quarter.

There is a quieter trap in it too. Human follow-up dies at the second attempt, because a third starts to feel like pestering. But the person who did not pick up at two on a Tuesday was driving, not deciding, and the difference between those two is invisible from your end. So the list gets a burst of attention in a slow month, produces nothing that week, and gets closed again.

Meanwhile the contacts in it keep aging in two directions at once. Some are drifting further out of reach. Some are quietly becoming ready. Nobody is watching which.

What database reactivation actually does

The last one is the part that decides whether any of it was worth doing. An appointment that arrives with the reply attached is a different object from an appointment that arrives as a name and a time. You already know there is a house to sell, and you already know they said so themselves.

The part nobody selling you this will mention

Reactivation is the one AI use case in real estate where the technology is the easy half and the rules are the hard half. Everything above this line is a build problem, and it is a solved one. Everything below it is the reason a large share of these campaigns should never have been sent at all.

I am not a lawyer and none of what follows is legal advice. Read it instead as the list of things you are entitled to ask about before anybody sends a message with your name on it, ourselves included. There are three of them, they are all answerable from your own records in an afternoon, and a vendor who cannot answer all three has told you something important.

There is a clock on your old lead, and it is shorter than you think. Federal rules exempt calls to somebody you have an established business relationship with from the national do-not-call registry. That exemption is defined with dates in it. Under 47 CFR 64.1200(f)(5), the relationship runs from "the subscriber's purchase or transaction with the entity within the eighteen (18) months immediately preceding the date of the telephone call or on the basis of the subscriber's inquiry or application regarding products or services offered by the entity within the three months immediately preceding the date of the call". A closed deal buys you eighteen months. A form fill buys you three. Your 2023 lead ran out of both a long time ago, and if that number is on the registry, warmth is not a defense.

An automated call or text is a separate question with a stricter answer. The do-not-call registry is one rule. Using an autodialer or an artificial voice is another one entirely, and clearing the first does not clear the second. The same regulation, at paragraph (a)(2), bars a telemarketing call or text to a mobile number placed with "an automatic telephone dialing system or an artificial or prerecorded voice" unless you have the prior express written consent of the person you are calling. Written consent is defined narrowly: a signed agreement, with a clear and conspicuous disclosure that the person is authorizing automated calls, and a statement that agreeing is not a condition of buying anything. An electronic signature counts. A checkbox on an IDX registration page that said "by registering you agree to be contacted" almost certainly does not.

A no has to be easy, and it has to stick. The rules were tightened here recently and the current version is unusually specific. A person can revoke consent "by using any reasonable method", and stop, quit, end, revoke, opt out, cancel and unsubscribe are all listed as reasonable by definition. If somebody replies with different words, you have to treat that as an opt-out too whenever "a reasonable person would understand those words to have conveyed a request to revoke consent". And the request "must be honored within a reasonable time not to exceed ten business days from receipt". You cannot make people use one specific channel to get out.

Every one of those is checkable in an afternoon, and the answers live in your own systems: the form your leads filled in, the dates on the records, and whatever your CRM does with the word stop. If a vendor cannot tell you how their campaign handles all three, the campaign is not ready to send.

Once you know how much of that list you are actually allowed to contact, the rest is arithmetic, and it is arithmetic nobody can do for you.

What it costs when it goes wrong

We do not publish a price for this and I am not going to invent one here, because it depends entirely on the size and state of the list. What is worth publishing is the other side of the ledger, because it is a fixed number and almost nobody quotes it.

The Telephone Consumer Protection Act carries a private right of action. Under 47 U.S.C. 227(b)(3), a person can recover their actual loss "or to receive $500 in damages for each such violation, whichever is greater", and a court that finds the violation was willful or knowing may treble it. Per message. There is a whole plaintiff's bar that does nothing else, and a list of thirty thousand contacts is a large multiplier attached to a small mistake.

While we are here: you will see agents quote a figure of sixteen thousand dollars per text, and it is the wrong number twice over. Sixteen thousand was the Federal Trade Commission's ceiling for an unfair or deceptive practice until the June 2016 adjustment raised it to forty thousand, and 16 CFR 1.98 now puts that ceiling at $53,088. It is also a penalty an agency assesses rather than the claim a person brings against you. The number in a private lawsuit is five hundred, and five hundred multiplied by a list is the one that should worry you.

The second cost is less dramatic and far more likely, and it has a threshold you can actually look up. Send messages people did not want and some of them reply stop, which is their right and takes one word. Twilio, whose pipes a great many of these campaigns run through, publishes where that becomes a problem: an opt-out rate "under 1% is considered healthy; over 3% may lead to carrier filtering". Filtering is not a bounce. Your messages to real clients simply stop arriving, silently, with no error, and you find out a fortnight later when somebody mentions they never got your text about the inspection.

Both of those costs are avoidable, and both of them are avoided in the same place: before the first message goes out, not after the first complaint.

What to ask before you let anybody text that list

Everything above is theory until you are sitting across from somebody who wants the export. Four questions separate the people who have done this properly from the people who are about to use your database as their sandbox, and none of them require you to know anything technical.

Show me the consent evidence for one specific record. Pick a name yourself, at random, and ask what date that person gave permission, through what form, and where that date is stored. If the answer is a general assurance about the CRM rather than a date on a screen, nothing has been checked and the first message is a guess.

What happens to a message that would land at nine at night. You want to hear a rule that is enforced by the system, not a promise about scheduling discipline. Ask what happens when somebody replies at 9:40pm, too, because the answer to that one is usually more revealing.

Show me a real opt-out, end to end. Have somebody send stop from a phone in the room, and then try to send that number another message. If the second message goes out, the suppression list is decoration. This takes four minutes and almost nobody asks for it.

Whose number is it. If the campaign runs from a number the vendor owns, then the replies, the reputation attached to it and the ability to keep using it all belong to the vendor, and moving to anybody else means starting again. That is not a reason to say no. It is a reason to know before you sign rather than after.

What it does not do, and should not pretend to

It does not manufacture intent. Most of that list is going to say no again, because most of them meant it. What reactivation changes is that somebody finally asked, and that the small number whose circumstances moved get found in the same week they moved rather than eighteen months later. Anybody quoting you a conversion rate on a cold database before they have seen the database is quoting you a number they made up.

It does not fix a list that has no consent behind it. If the records do not carry a date and a source, the honest first project is not a campaign, it is a cleanup.

It is not a subscription, and you should be suspicious of anybody who sells it as one. A database is finite. You work it once, you find the people whose situation changed, and then the same list is a year away from being worth working again. Vendors in this category run out of inventory and quietly pivot the client onto something else, and the ones who are straight with you say so at the start. Reactivation is a harvest, not a lead source, and it does not replace generating new business.

It does not replace you calling the people who actually matter. Your past clients, the neighbor who sold with you in 2019, anybody whose relationship is the real asset: those calls are yours. A machine reintroducing itself to somebody who came to your daughter's christening is worse than no contact at all.

It also does not survive being run twice in a row. The first pass takes the accumulated years off the list, and whatever it finds is genuinely there. The second pass, three months later, is fishing in a pond somebody already emptied, and the temptation at that point is to widen the definition of who counts as a lead until there is a list again. That is the moment a reactivation campaign quietly turns into cold outreach, with none of the consent behind it that made the first one defensible.

And it does not close. It finds the conversation. Everything that happens after somebody says yes to a Thursday is the reason you have a job.

Where it goes wrong

Common questions, answered honestly

Do old real estate leads actually convert?

Some of them do, and that is the whole business case rather than a sales line. A person who said not right now two years ago was giving you a timeline, and that timeline has since run out. The job is finding the small share whose circumstances changed, which is exactly the kind of patient, repetitive work software does well and people do not.

Is it not annoying to text people who went quiet years ago?

It is, if the opener is generic. The difference between reactivation and a blast is whether the first message references what that specific person actually asked about, asks a real question, and takes no for an answer the first time. It also matters enormously that the person hears from you rather than from a stranger, which is why the consent and identification questions above are not paperwork. They are the reason the message lands as a follow-up instead of as spam.

How does it decide who is worth calling?

It reads intent from what the person says back: whether the move is still on, whether the timeline shifted, whether there is now a house to sell. Everybody in scope gets an attempt. Only the ones showing real intent reach your calendar.

Can I just do this myself with a mail merge?

You can, and for a list of two hundred you probably should. The reason it stops working is volume and follow-through, not sophistication. A merge sends one message. What actually produces appointments is the second and third exchange, held at the hour the person happens to reply, across a few thousand contacts, without anybody getting bored. That is the part that is hard to staff and easy to automate.

What happens to the people who say no?

They get recorded as a no, with whatever reason they gave, and they are not contacted again by the campaign. That is not only politeness. A no with a reason attached is the most useful thing the whole exercise produces, because it tells you which part of your list is genuinely dead and which part is just early.

What to do about it

Start with one record instead of with a campaign, and do it before you talk to a vendor. Open your CRM, sort by oldest, and find your own version of the person who asked what their house was worth in the spring of 2023. Then look at what you are actually holding: is there a date on the record, is there a source, and is there anything anywhere that says how that person agreed you could contact them. Four minutes on one record will tell you which of the two projects in this article you are really looking at, a campaign or a cleanup, and no consultant can tell you that faster than your own database can.

The system is laid out on the RealtyLT AI page, and the database reactivation page sets out what it reads, what it sends, and what happens to a reply once somebody sends one.

Two earlier moments in the same story are written out elsewhere: the missed call at 9:42 on a Sunday is this problem before the lead ever goes cold, and the buyer messaging your website at 11:40pm is where the record you just opened came from in the first place.