Tracing a Closing Back to Its Source

A loan funds on a Friday. Eleven months earlier there was a postcard, then an ad, a portal inquiry, a call to an assistant, and a showing. Ask where the deal came from and you get five answers. Why attribution is harder in real estate than almost anywhere, and the principles I hold to when I trace a closing back to its start.

A steam locomotive with a red cowcatcher pulls into a desert station platform under a wide blue sky
For a Few Dollars More (1965), dir. Sergio Leone

A loan funds on a Friday.

Eleven months earlier, a postcard landed on that household’s counter. Then they saw an ad. Then they filled out a form on a portal. Then someone called a front-desk assistant and got a callback. Then they toured three houses with an agent, and the agent’s name ended up in a note somewhere.

Ask the team where that deal came from and you will get five answers. The marketer says the ad. The agent says the referral. The assistant says the call. Everyone is partly right, and almost nobody can prove it.

I have written about why last-click credit misreads a long journey. This is a different problem, and in real estate it is harder than in almost any other business I know. It is not about which model gives credit. It is about whether the chain from the first touch to the closing exists at all.

The sale ends where marketing cannot see

Most online businesses have a nice property. The purchase happens on the same screen as the marketing. The click, the cart, and the receipt all live in one system.

A home sale does not work like that. The decisive events happen offline and in systems that marketing never touches: a showing, a negotiation, an inspection, a lender’s file, a title company, a closing table. By the time money moves, the person is far from any website. A marketer sees a form fill in month 1 and then silence, while the real story runs for 8 more months in places their tools cannot read.

So the most important event in the whole journey, the closing, is usually the one that never makes it back to the people who paid for the ads. That is the first break in the chain, and everything else is downstream of it.

People re-key, overwrite, and forget

The second break is human, and I think it is the one nobody budgets for.

Between a lead and a closing, the record passes through many hands. An assistant creates it. An agent works it. A loan officer starts a separate file. Someone updates the CRM when they remember. Along the way, fields get retyped, merged, and replaced. A lead that began as a postcard response gets relabeled “referral” because the agent knew the family. The original source is gone, overwritten by a well-meaning person who was only trying to be accurate about the latest conversation.

None of those people is doing anything wrong. They have a job, and the job is closing a deal, not preserving a data trail for the marketing team. If the system does not protect the origin of each person on purpose, the origin quietly disappears, and no report can bring it back.

The phone and the showing are the real conversions

In a lot of real estate, the moment a lead becomes a real prospect is not a form. It is a phone call, a text, or a walk through a house. These are exactly the touches that leave the weakest digital trace.

A caller does not carry a tracking tag. A 20-minute showing leaves no pageview, and a callback in the CRM may be logged as “phone” with nothing about why they called. The channels that created the call, whether a search, a map listing, a mailer, or a video, may all be invisible by the time a human picks up. If you count only what a browser can see, you will undercount the channels that start conversations. And under a last-click rule, you will overcredit the ones that happen to be the last thing a person typed into a search bar.

One buyer, many people, many devices, long gaps

I keep coming back to the household as the account, because it makes attribution harder in a very specific way. The person who saw the ad is not always the person who filled out the form. The one who called is sometimes the other half of the couple, from a different phone. A parent who is helping with the down payment may have done half the research on a work laptop.

Stretch that across a sales cycle of 12 to 18 months, and the same household appears in your data as four or five unrelated visitors. The touches are real. The link between them is not there unless someone built it. This is the same identity problem that makes direct mail so hard to measure, just with more channels in the mix.

What I hold to

I will not lay out the wiring. How the pieces connect depends on the operator, the stack, and the rules that apply, and it is part of how we run accounts. But the principles are plain enough to say out loud.

The closing is the conversion that counts. A form fill is a step. A qualified stage is a step. The thing that matters is a funded loan or a closed transaction, so the system has to be built to carry that event back, not stop at the lead.

The CRM is the record, and someone owns it. Not a dashboard, not an ad platform, not a spreadsheet. One place is the source of truth for who a person is and what happened to them, and the process is designed around keeping it honest.

Protect the origin. The first source you observed for a person is information that must survive every later update. It should not be overwritten by the next conversation, however accurate that conversation is. If you later learn the real start was earlier, add that alongside the original instead of writing over it.

Record every later touch against the right person. Ads, calls, mail, email, and visits belong on that person’s record, linked to the others in the same household, with an honest bucket for the ones that cannot be matched.

Close the loop at the end. Closings get reconciled back to the people and the touches that preceded them. If you only reconcile at the top of the funnel, you are measuring activity, not results.

None of this is exotic. It is discipline applied to a data chain with a lot of weak links, and most of the work is in the unglamorous parts.

What it changes

Once you can trace a closing back through its touches, you stop arguing about which channel deserves credit and start asking a better question: what role did each one play?

Some channels are starters. They put a new household into the system. Others are finishers. They bring someone back at the moment of decision. A mailer might start a story and never get a click. A branded search might finish one it had nothing to do with creating. Treat those roles as a working hypothesis, not a verdict. Order is not proof of cause, so test it before you move real budget on it. But a report built on last-click or on raw leads will not even show you the roles, and it can point you the wrong way.

It also changes the conversation with a client. Instead of “we generated 90 leads this month,” the discussion moves to which channels are producing closings and how long they take to show up. That is a much better meeting to be in.

The honest limits

I want to be straight about this. It is never perfect.

Some closings have no trail at all. A buyer walks into an open house on a Sunday with no prior touch that you can see. Some trails are wrong, because a person at some point typed the wrong thing. Some links between devices and people are certain, like a login. Most are inferred, and some stay unresolved. Privacy rules, correctly, limit how far you can and should go. A long cycle means the full picture of this quarter’s marketing will not exist until next year.

So the goal is not a perfect answer, and I would be suspicious of a CRM report claiming 100 percent coverage. It is an honest one: a clear view of what you can trace, an explicit unknown bucket for what you cannot, and a refusal to pretend the unknown is zero. A marketing program that is honest about what it cannot see will make better decisions than one that trusts a dashboard because the dashboard is confident.

Tracing a closing back to its source is hard because the sale lives in places marketing cannot see. It is possible because someone decided it mattered, and built for it on purpose.