Build the Receipts: How a Solo Real-Estate SMMA Survives the 90-Day Realtor Churn Cycle

You did not get into this business to argue about whose fault it is.

You got in because running Facebook lead ads for a real-estate agent looked like a clean, repeatable service you could sell to one person at a time for $500 to $1,500 a month. And the ads part usually works. You can fill a lead form. You can hit a cost per lead that looks defensible. What breaks is everything that happens after the form is submitted — and almost none of it is under your control, which is precisely why it keeps ending your accounts at day 74.

This is not a post about building an inside-sales operation or a multi-team enterprise deployment. If you have a fourteen-person ISA floor, this is not for you. This is for the operator with four to eight realtor clients, no staff, no account manager, and a very specific recurring nightmare — an agent who signs up excited, goes quiet, ignores the leads, and then calls you in month three to say your leads are trash.

The fix is not better leads. The fix is receipts.

Why do realtors fire their marketing agency after 90 days?

Because at day 90 the agent has spent roughly $6,000 to $12,000 between your fee and ad spend, has no closed transaction to point at, and needs a story that explains the gap. The story that requires nothing of them is that the leads were bad. The story that requires something of them is that they never worked the leads. Without data, the first story always wins.

Look at the arithmetic from the agent's side. A typical Facebook buyer or seller lead campaign at $1,500 a month in spend produces somewhere between 60 and 150 form fills depending on the metro and the offer. In Dallas or Phoenix you might see a $12 to $20 cost per lead; in Toronto or Sydney, higher. Of those, a realistic share ever answer a phone. Of those who answer, a small fraction are within ninety days of transacting. Of that fraction, one might list or buy — and real-estate cycles being what they are, that closing may land in month five, not month three.

So the agent's honest expectation and the market's honest timeline are separated by two to three months. The relationship dies inside that gap. This is structural, not personal, and you cannot outrun it by generating more leads.

Now add the second problem. The agent judges you on volume, because volume is the only thing they can see. When 120 leads come in and nothing happens, the visible number is 120 and the invisible number is whatever they actually did with them. You are being evaluated on a metric that has almost no relationship to outcomes, using a dataset that contains no record of their own behaviour.

The third problem is the most frustrating. When you ask the agent whether they called the leads, they say yes. They believe it. They called some leads, in the first week, when they were excited. Human memory generalises from the vivid cases. Nobody remembers the eleven Tuesdays they meant to work the list and did not.

You cannot argue a client out of a belief with your own competing belief. You can only replace the belief with a record. That is the whole strategy in one sentence.

What actually happens to a Facebook lead in the first sixty minutes?

It decays, fast, and the shape of that decay is the single strongest argument for automating the first touch entirely.

A Facebook lead-ad submission is a low-friction, low-intent event by design. Someone was scrolling, saw a home-value estimate or a list of new listings, and tapped a form that pre-filled their name, email, and phone from their profile. They did not visit a website. They did not type anything. Twenty minutes later they have no strong memory of doing it.

That is why response time dominates every other variable in this category. Contact a lead within the first few minutes and you are talking to someone who still remembers the ad and still has the intent that prompted the tap. Contact them the next afternoon and you are a stranger cold-calling about something they half-remember. Contact them three days later and you are spam.

Here is the part that matters for your business specifically. Your client is a working real-estate agent. Between 9am and 7pm they are driving, at showings, at inspections, on the phone with a title company, or at a closing. The realistic probability that an individual agent picks up a new lead inside five minutes, during a workday, unassisted, is very close to zero. Not because they are lazy — because they are physically occupied.

So you have a service whose success depends on a five-minute response window, delivered by a person who is structurally incapable of hitting that window. That is a design flaw, not a client-quality problem, and it is your job to design around it.

The design fix is that the first response never depends on the agent. An automated SMS and email go out within sixty seconds of the form submission, under the agent's name and business number, sounding like a person, asking one simple question. The lead replies to a conversation that is already open. Only then does the agent get pulled in — with a notification, on a conversation that already has a warm reply in it.

This flips the agent's job from initiating cold outreach at speed, which they cannot do, to responding to warm replies, which they can. And crucially, it means the speed-to-lead metric you promised on the sales call is one you can actually deliver, every time, whether or not the agent is at a closing.

Why is "your leads are trash" almost never about the leads?

Because in nearly every account where an operator has actually measured it, the dominant failure is contact coverage, not lead intent.

Think about what the phrase is really describing. To know a lead is bad, someone has to have spoken to them and established that they are not a buyer or seller. A number that was never dialled is not a bad lead. It is an unknown lead. The agent is describing a population they never sampled.

There is a version of this where the agent is right, and you need to be honest about it. Broad interest targeting with a soft offer — free home valuation, list of homes under a price point — does produce a lot of casual curiosity. Some submissions are people checking their own home's value with no intention of selling for four years. Some are renters. Some are other agents. And a meaningful slice, in unfiltered campaigns, are simply invalid — mistyped numbers, deliberate junk, duplicate entries from the same person hitting the form twice.

So both things are true at once. There is genuine junk in the raw feed, and there is genuine neglect in the follow-up. The reason the conversation always collapses into "your leads are trash" is that only one of those two things is currently measured. You are measuring cost per lead. Nobody is measuring contact rate.

The moment you measure both, the conversation becomes tractable. You can say, with numbers, that of 94 raw submissions, 12 had invalid phone numbers and were filtered, 71 were never contacted by anyone at the brokerage, and of the 23 that were contacted, 4 booked appointments. That is a 17 percent booking rate on contacted leads — which is a functioning campaign being starved by a follow-up gap.

Without that breakdown, you are two people with opinions, and the one holding the credit card wins.

What does "build the receipts" actually mean in GoHighLevel?

It means every meaningful event in the lead's life leaves a timestamped record in one place, and one of the deliverables you send every month is a plain-language summary of those records.

Concretely, six things need to be true in the build.

Every lead enters the CRM automatically, with the source, the form, the ad set, and the exact submission timestamp attached. No spreadsheet exports, no email notifications as the system of record.

Every outbound message the system sends is logged against the contact with a timestamp. That gives you your speed-to-lead proof.

Every inbound reply is logged against the same contact. That gives you engagement data.

Every call, text, and email the agent makes goes through the platform, not their personal handset. This is the configuration decision that makes or breaks the entire report, and it must be settled during onboarding.

Every pipeline stage change is timestamped, so you can see whether a lead sat in "New" for six weeks.

And every appointment booked, kept, or no-showed is recorded on the calendar tied to the contact.

That is the raw material. The receipts themselves are the monthly artefact you build out of it — the contact-activity report — which we get to shortly. But notice what is required here. There is nothing exotic in that list. No custom development, no API work, no third-party tools. It is a correctly configured CRM plus the discipline to insist that client communication happens inside it.

The reason most solo operators do not have this is not difficulty. It is that they were sold on the ads being the product, so they treat the CRM as a place leads get dumped rather than as the instrument that measures the whole engagement.

How do you filter garbage Facebook lead-ad submissions before they reach the agent?

With four gates, applied in order, before anything appears in the pipeline the agent looks at.

Gate one — format validation. Check the phone number against the expected length and format for the country. A US mobile that is not ten digits after stripping formatting is invalid. Numbers that are obviously fake patterns — sequential digits, repeated digits, an area code that does not exist — get flagged. Same for the email field, which Facebook pre-fills but people do sometimes edit into nonsense. This alone typically removes a small but annoying percentage and costs you nothing but a workflow condition.

Gate two — duplicate suppression. Check the incoming phone and email against existing contacts. In an account running the same offer for months, repeat submitters are common — the same curious homeowner checks their value three times across a quarter. They should update the existing contact and re-enter nurture, not create a third row that inflates your lead count and makes the agent think they have more pipeline than they do. Inflated counts feel good for one month and destroy trust in month three.

Gate three — the reply gate. This is the meaningful one. The automated sequence goes out, and how the contact behaves determines where they land. A lead who replies to any message is a live human with at least minimal engagement, and moves into the agent's active pipeline immediately. A lead who never replies across the full sequence gets moved to a long-cycle nurture rather than sitting in the pipeline pretending to be an opportunity. They are not deleted — real-estate cycles are long and unresponsive contacts do resurface months later — but they stop occupying the agent's daily view.

Gate four — qualifying questions. Ask one or two things in the form or the first automated exchange that separate timelines. For sellers, something like whether they are thinking about selling in the next few months or just curious about the number. For buyers, whether they are already working with an agent and whether they have spoken to a lender. Tag the answers. A buyer who is pre-approved and not working with an agent is a fundamentally different asset from one who is neither, and both should be in the account, but they should not be in the same stack of work.

The net effect is that the agent opens their pipeline and sees a shorter, better list. The lead count in your report goes down, and this is the moment to get ahead of the obvious objection. Show both numbers. Raw submissions, and contactable leads. The gap is a feature you are delivering, not a shortfall you are hiding.

What does the speed-to-lead sequence actually look like?

It runs for roughly ten days, front-loads heavily into the first hour, alternates channels, stops instantly on any reply, and never sends outside local waking hours.

Here is the working cadence. Adjust the copy to the market and the offer, but keep the shape.

A few rules that matter more than the exact timings.

Send from the agent's business number, provisioned inside their sub-account, with the agent's name in the message. The lead should experience this as the agent being fast, not as an agency bot. That is also what makes the agent comfortable with it.

Write like a person texting, not a brand broadcasting. One sentence, one question, no links in the first message, no emoji, no all-caps. "Hi Sarah, it's Mike — just saw your request for the value on Oakmont Dr. Are you thinking of selling this year or just curious about the number?" outperforms anything that reads like marketing.

Kill the sequence on any inbound reply, immediately and globally. Nothing destroys trust faster than a lead replying "yes I'm interested" and receiving an automated follow-up two hours later asking if they got the message.

Respect quiet hours in the lead's timezone. Nothing before 8am or after 8:30pm local. A midnight text is a complaint and, in some jurisdictions, a legal problem.

And make sure the sequence always terminates somewhere defined. Every lead ends in one of three states — booked, in nurture, or filtered as junk. None of them ends nowhere.

That last rule is what makes the report possible, because a lead with no defined end state is a lead you cannot report on.

What pipeline will a realtor actually use?

A short one. Five or six stages, maximum, with names that match how agents already talk.

The single most common mistake a solo operator makes is building a beautiful fourteen-stage pipeline that maps a sophisticated sales process, handing it to an agent, and discovering three weeks later that every lead is still sitting in stage one. Not because the agent is incapable, but because a pipeline with fourteen stages requires a decision at every card and agents do not have the attention budget for that between showings.

Build two pipelines, one for buyers and one for sellers, because the questions are genuinely different and mixing them makes both worse.

The seller pipeline runs something like — New Lead, Contacted, Valuation Requested, Listing Appointment Set, Listing Appointment Held, Listed. The buyer pipeline runs — New Lead, Contacted, Qualified, Showing Scheduled, Actively Touring, Under Contract.

Automate every stage move you possibly can. When the automated SMS sends, the lead moves to Contacted by workflow, not by the agent remembering. When an appointment is booked on the calendar, the card moves automatically. When an appointment is marked as held, it moves again. The stages the agent has to move manually should be the two or three that genuinely require human judgement.

This is not just convenience. Every automated stage move is a clean data point for your report, unpolluted by whether the agent felt like updating a card that day. Manual pipeline data is aspirational. Automated pipeline data is evidence.

Layer appointment booking directly on top. The agent's calendar connects, availability rules get set once during onboarding, and the booking link appears in the SMS sequence, the email sequence, and the agent's own signature. Confirmation and reminder messages go out automatically at 24 hours and 1 hour before, because real-estate no-show rates on self-booked appointments are ugly without them.

Then the long-cycle nurture underneath everything. Nothing complicated — a monthly email with genuinely useful local market content, plus a quarterly SMS check-in, running for twelve to eighteen months. Most real-estate leads are not transacting this quarter. The nurture exists so that when they do decide, in month nine, your client's name is the one in the recent message history.

What goes in the contact-activity report?

Three sections — a summary, the contact-coverage breakdown, and a lead-level detail table. It goes out monthly, on a fixed date, whether the numbers are good or bad.

The summary is four or five numbers, stated plainly at the top. Leads generated. Leads filtered as invalid. Leads contacted by the automated system. Leads contacted by the agent personally. Appointments booked. Appointments held.

The coverage breakdown is where the argument gets settled. It answers one question — of the contactable leads this month, what percentage did the agent personally touch by call, text, or email?

The detail table is the receipts themselves. Something like this.

Read that table as a realtor and the story tells itself. Four leads replied to the system. One got a same-day text and became a held listing appointment. One got a call and a text and is touring. Three replied to the automated message, said they were interested, and were then never touched by a human being again.

That is the entire pitch. You do not have to editorialise. You do not have to say "you are not calling your leads." The rows say it, and the agent reaches the conclusion themselves, which is the only way anyone ever accepts an uncomfortable one.

Two practical notes on building this. Generate it from the CRM's own reporting and conversation data rather than reconstructing it by hand — if it takes you three hours a month per client, you will stop doing it by month two, and a report you stopped sending is worth nothing. And send it on a fixed schedule from the very first month, when the numbers are still fine, so it never looks like a document you produced defensively after things went wrong.

How did Marcus cut his realtor churn with a report?

Marcus runs a solo real-estate SMMA out of Phoenix. He signs individual agents and the occasional two-person team, charges $900 a month plus ad spend billed to the client, and until last year he was quietly failing.

Over five months he lost four of six clients. Every exit conversation was some version of the same thing. "The leads are trash." "Nobody answers." "I'm not seeing a return." One agent told him she had called every single lead and not one was real.

Marcus believed it. He rebuilt creatives. He moved from broad targeting to lookalikes. He tightened the geography. He added a qualifying question to the form. His cost per lead moved from $19 to $14, which he was proud of, and it changed nothing about his retention. He was optimising a variable that was not the bottleneck.

The turn came from a conversation with his fifth client, an agent named Dana, six weeks into the engagement. Dana said she was thinking about pausing. Marcus, out of something between curiosity and desperation, went into the CRM and did something he had never done — he went lead by lead through 94 submissions from the previous seven weeks and manually built a spreadsheet of what had actually happened to each one.

Twelve had invalid phone numbers. Eleven had replied to his automated first text. And of the 94 leads, Dana had personally contacted 23. Seventy-one had never received a call, a text, or an email from her. Ever.

He did not send it as an accusation. He rebuilt it as a clean one-page summary, put it on a screen share, and walked through it stage by stage without commentary. Then he stopped talking.

The silence on that call, he says, was about eight seconds long. Dana said, "I didn't realise it was that bad."

They did not have a churn conversation after that. They had a workflow conversation. Dana admitted she was checking leads on her personal phone, meaning to get back to them, and losing them in the scroll. So Marcus set up two things — the GoHighLevel mobile app with push notifications so replies hit her like a text from a friend, and a standing fifteen-minute slot at 8:15am and 5:30pm where her only job was to clear the Contacted column. Not call 40 people. Clear a column.

Over the next quarter Dana's personal contact coverage went from 24 percent to 81 percent. Booked appointments went from 3 a month to 9. She listed two properties from the campaign in month four, and she stopped talking about lead quality entirely.

Marcus then did the thing that actually changed his business — he productised it. He built the whole thing once as a reusable package. The filtering, the ten-day sequence, the two pipelines, the booking calendar, the eighteen-month nurture, and an automated monthly activity report that took him about twenty minutes per client to review and send rather than a full day to assemble.

Fourteen months later his numbers look different. He runs nine realtor clients. His average client tenure went from a shade under three months to just over nine. Two clients have been with him more than a year. He raised his setup fee from $500 to $1,200 because the onboarding now clearly delivers a system rather than just an ad account, and nobody has pushed back.

He still loses clients. Two of the nine will not last the year. But when they leave, it is a conversation about market conditions or budget, not an argument about whether he did his job. He has the receipts, and both parties know it.

How do you run the retention conversation with evidence?

Carefully, early, and without ever using the word "you" as an accusation.

The first rule is that this conversation happens on a schedule, not in a crisis. A fifteen-minute monthly review call, booked as a recurring slot at onboarding, is where the report gets walked through. If the first time an agent sees a contact-coverage number is during a cancellation call, it reads as a defensive document and you have lost anyway.

The second rule is to lead with the system's performance and the funnel, not the agent's behaviour. Start with leads generated, cost per lead, and speed-to-lead — the parts you own. Show that every single lead got a first touch inside sixty seconds. That is a genuine win, it is yours, and it establishes that you are reporting on yourself before you report on them.

Third, present coverage as a shared metric. Not "you only called 23 leads" but "our contact coverage was 24 percent this month — here is what that means and what happens when it moves." Frame it as a dial the two of you are turning together. That is not spin; it is accurate. You built the system that makes contact easy, they supply the human touch, and coverage is the joint output.

Fourth, always attach a specific, small next action. Coverage problems are almost never solved by asking someone to try harder. They are solved by removing friction. Push notifications on. A fifteen-minute slot morning and evening. A saved reply template for the three most common lead responses. Voicemail drops so the agent can leave twenty messages in six minutes while driving between showings. Small, concrete, achievable.

Fifth, when the agent does move the number, make a visible thing of it. Coverage went from 24 to 61 percent, booked appointments doubled, and here is the causal line between those two facts on the same page. Agents respond to evidence that their effort produced a result far more than they respond to being told their effort was insufficient.

And sixth — be prepared for coverage to go up and results to stay flat. That happens. When it does, you have earned the right to a completely different conversation, one about targeting, offer, geography, or price point, and you will be having it with a client who trusts your data because you showed them the uncomfortable numbers first. That trust is the actual asset. The report is just how you build it.

What should you charge, and what does this cost to run?

Setup of $500 to $1,500, retainer of $300 to $800 a month, ad spend billed separately and paid by the client directly to the platform. Never fund a realtor's ad spend through your own card.

Set the setup fee against what you are actually building. At the low end, the agent already has campaigns running and you are installing the follow-up system, the pipelines, the calendar, and the reporting. At the high end you are also producing the ad creative, writing the forms and offers, building the landing assets, and running the account. The setup fee also does real commercial work beyond covering your time — it filters out the tyre-kickers, and an agent who has paid $1,200 to get started is measurably more likely to actually engage with the system in week one.

Your running costs per client are modest and predictable. A sub-account on your agency plan. Phone number rental, which is a couple of dollars a month. Per-message SMS costs, which for a campaign generating 100 leads a month running the ten-day sequence typically lands in the low tens of dollars. Email is effectively free at this volume. Your real cost is time — the monthly report review and the fifteen-minute call, which after the first couple of months should be well under an hour per client per month.

That leaves healthy margin at $500 a month, and very healthy margin at $800.

The thing that destroys this model is not price. It is custom work. The moment you agree to build a bespoke pipeline for one agent because they have a particular way of tracking things, you have created a snowflake account that you have to maintain forever and cannot replicate. Ten snowflakes and you have no business, you have ten part-time jobs.

Hold the line on one package. The offer is that the system is proven, standard, and deployable in days precisely because it is not custom. Agents accept this readily when you frame it as a benefit — this is the exact build running for eight other agents, and here is what it produces — rather than as a limitation.

Where you do flex is copy and targeting. Message wording, market content, geography, price point, and offer are all client-specific and always will be. The architecture underneath never changes.

How do you deploy the same build to the next client without rebuilding it?

Snapshot it, then run a fixed onboarding checklist.

Build the whole thing once in a clean sub-account — pipelines, workflows, message copy with merge fields for the agent name and business, calendar template, nurture campaigns, custom fields, tags, and report configuration. Get it genuinely right, because everything you deploy afterwards inherits its flaws. Then save it as a snapshot and load it into each new client sub-account.

Use merge fields everywhere so the copy personalises itself. The first SMS references {{contact.first_name}} and the agent's name from the sub-account settings, not a hard-coded name you have to find and replace in nineteen workflows.

Your onboarding checklist should look roughly like this, and it should be the same every time.

Day one — collect the legal business entity details, EIN or equivalent, and website, and start phone registration immediately, because it is the longest lead time and it is outside your control. Provision the number. Connect the Facebook page and ad account. Connect the agent's calendar and set availability rules.

Day two — load the snapshot. Customise the message copy with the agent's name, market, and offer. Set the geography and price bands in the nurture content. Walk the agent through the mobile app, get push notifications enabled on their actual device while you are on the call, and get them to send one test text through it. Do not accept "I'll set it up later." Later means never, and this is the step the entire reporting layer depends on.

Day three — run five test leads through the full sequence yourself using real phone numbers. Check the timing, check the merge fields render, check quiet hours behave, check the stage moves fire, check the booking link works on a phone. Then go live.

Set the recurring monthly review call before you end the onboarding, and send the first report at day 30 even though the numbers will be thin. You are establishing a rhythm, not delivering insight, and the rhythm is what saves you in month three.

Realistically this runs three to five hours of your time spread over two or three days, most of it waiting on the client. That is a defensible cost against a $1,200 setup fee and a client who might now stay nine months instead of three.

What still breaks, and how do you get ahead of it?

Five things, and all five are predictable enough to handle during onboarding rather than in a crisis.

The agent uses their personal phone anyway. This is the big one, and it silently guts your reporting. Symptoms are contact coverage that looks impossibly low next to an agent who insists they are working leads. Get ahead of it by setting the app up live on the onboarding call, and by selling it on convenience — one number, personal line stays private, call recordings they can review, message history that survives a lost handset. Check the app usage numbers in week two, not month three.

Message deliverability quietly fails. Unregistered or improperly registered SMS traffic gets filtered by carriers, and the failure is invisible from your side — messages look sent, nothing arrives. A speed-to-lead system that silently does not deliver is worse than none, because you are reporting a first-touch that never happened. Complete registration properly, and monitor delivery rates rather than send rates.

The agent renegotiates on price in month two. Usually right before the results curve turns. Your defence is the report — showing a functioning system with a coverage gap reframes the conversation from value to execution. Your other defence is not signing agents who were price-shopping from the first call.

Lead volume drops and the agent panics. Facebook ad accounts fluctuate; a bad two weeks happens. Report on it before they notice it. An operator who flags a dip and explains the plan retains trust; one who gets asked about it looks caught out.

You stop sending the report. This is the most common failure and the most self-inflicted. Around month four, when things are going fine, sending the report starts to feel unnecessary. Then month seven happens and you have no baseline, no rhythm, and a client who has forgotten what you do. Automate the assembly, calendar the review calls as a recurring block, and treat the report as the product it actually is.

Notice that none of these are lead-generation problems. Your ads were probably never the weak point. The weak point was that everything after the form submission was invisible, and invisible work is work you do not get credit for.

Where this leaves you

The solo real-estate SMMA business is genuinely hard, and it is hard for a reason that has nothing to do with your skill at running ads. You are selling a service whose outcome depends on the daily behaviour of a busy, distractible person you do not employ and cannot manage — and until now you have had no visibility into that behaviour and no evidence when it goes wrong.

The system in this post does not fix realtors. Nothing fixes realtors. What it does is remove the two things that were killing your accounts. It makes speed-to-lead independent of the agent, so every lead gets worked in sixty seconds whether or not your client is at a closing. And it produces a record, every month, of exactly what happened to every lead — which turns your worst conversation from an argument you cannot win into a document you can walk through.

Marcus did not get better leads. He got better at proving what happened to them, and his average client tenure tripled.

If you want that built properly rather than assembled over six frustrating weekends, that is exactly what we do. GHL Spark builds the complete realtor package — lead filtering, the sixty-second speed-to-lead sequence, buyer and seller pipelines, booking, long-cycle nurture, and the contact-activity report — as one clean, snapshot-ready deployment you own and can load into every client you sign after it.

You bring the ads. We will build the receipts.