Real Estate Lead Source ROI: Count Closings, Not Leads

The lead source that costs you nothing per lead can be the most expensive one you pay for.

Pay-at-close referral programs work that way. Zillow Preferred (formerly Zillow Flex), for example, sends no invoice when the lead arrives. The agent pays a success fee, a percentage of their side of the commission, when a deal with that connection closes, and the rate varies by market and price (The Close, June 2026). On a cost-per-lead report that source shows up as free. On a cost-per-closing report it can be the priciest line you have.

That gap is the whole argument of this post. A lead is not the thing you get paid for. A closing is. So the useful question is what each closing cost you, by source, once every fee and every month of lag is counted.

Why cost per lead points you the wrong way

Cost per lead is easy to calculate, which is most of its appeal. It goes wrong in three predictable ways.

• Cheap leads that rarely close. A source can deliver plenty of contacts for little money and still produce almost no deals.
• Costs that arrive at closing. Referral programs charge on the back end. Leave those fees out and the source looks free.
• Sources that never send an invoice. Your sphere and past clients cost you events, gifts, postage and hours, but no vendor bills you for them, so they tend to drop out of the comparison entirely.

That last one matters more than most teams admit. In the National Association of REALTORS® 2025 Profile of Home Buyers and Sellers, 43% of buyers used an agent they found through a referral and 18% went back to an agent they had worked with before, and 66% of sellers found their agent through a referral or a past agent (as summarized by Virginia REALTORS®). If those relationships are a real source of your business, they deserve a row in the report, with a real cost next to them.

The four numbers you need per source

Pick one period, a quarter or a year, and fill in four numbers for every source you pay for or put time into.

From those four you get the only two ratios you need. Cost per closing is total cost divided by closings. Net return is commission minus total cost. Cost per lead can stay on the sheet as a diagnostic, but it should never decide a budget.

A worked example (the numbers are invented, the arithmetic is not)

Here are three made-up sources for one team over one year, chosen to show how the rankings flip. The 35% success fee is also invented for the example. Real rates vary, so check yours.

Read the columns one at a time. If you leave the success fees out, cost per lead tells you the referral program is free and the sphere is the most expensive source on the sheet. Cost per closing puts the sphere first and the referral program last. Net return puts the sphere first and the portal ads last. Three columns, three different rankings. Only the last two are measuring the thing you get paid for.

The example is also a warning about small numbers. Three or four closings per source is normal for a small team, and one deal more or less would reorder this table. Treat a single year of a single source as a hint, not a verdict.

Measure by cohort, not by calendar

The most common mistake here is timing. If you count this quarter’s spend against this quarter’s closings, every new source looks like a loss and every source you just cut looks like it was working. Many of the deals closing this quarter came from leads you paid for long before it.

The fix is to measure by cohort. Group leads by the month or quarter they were created, and credit each closing back to the cohort its contact came from. Then compare a cohort’s closings with the spend that bought that cohort.

1. Find your own lag first. Take your closings from last year and, for each, note the days between the contact’s creation date and the close date. The middle value of that list is your typical lag. Do it separately for your biggest sources, because a portal lead and a past-client referral rarely move at the same speed.
2. Don’t grade a cohort until it has had that long. A source you started three months ago with a lag longer than that has not had a fair test yet. Track its early signals instead: response rate and appointments set, which is what Follow Up Boss points to for judging sources before deals close (Follow Up Boss reporting).
3. Re-run old cohorts. Last year’s cohort keeps producing closings this year. Update it each quarter instead of freezing it the day the year ended.

Fix the source field before you trust any of this

All of this depends on one field: the lead source on each contact, which is usually messier than anyone thinks. Start with our Follow Up Boss Pixel attribution guide for website leads, then do a cleanup pass.

• One source, many spellings. “Zillow”, “Zillow Flex” and “zillow.com” split one source into three small ones. Merge them, then decide on purpose whether two programs from the same company deserve separate rows. Our CRM cleanup guide covers the mechanics.
• Tools that are not sources. Home-search portals, home-value reports and AI texting tools often write their own name into the source field. They worked the lead. They did not create it, and crediting them with the deal hides where the lead actually came from.
• Deals that never get marked won. A closing that sits in a stage nobody recognizes as closed never reaches the report. If your stage names have drifted, the deal stages naming audit is the fastest fix.
• Referrals with no source at all. Sphere deals are the ones most often entered by hand with the source left blank. Our guide to tracking referrals in Follow Up Boss keeps them countable.

What this method can’t tell you

This is single-touch attribution: each closing is credited to the one source stamped on the contact. Real buyers touch several things before they hire an agent, and marketing research has found that the credit a channel gets depends heavily on how you measure it. In a study of multichannel online marketing, Li and Kannan (2014, Journal of Marketing Research) found channels’ contributions differed significantly from what commonly used metrics suggested. That was online retail, not real estate, so take the direction rather than any number from it.

In practice that means three cautions. A source can be worth keeping even when its own row looks weak, because it feeds your sphere years later. A past client you met through a portal will be credited to the portal on their next deal unless you decide otherwise. And no report will tell you whether an agent worked the leads well. When two reports disagree, our post on why Follow Up Boss team reports disagree walks through where the numbers usually split.

A quarterly checklist

1. Pull every vendor invoice and every referral or success fee paid this quarter, and tag each one to a source.
2. Merge duplicate source names and move tool names out of the source field.
3. Confirm every closing this quarter is marked won and has a source on its contact.
4. Update each open cohort with its new closings and commission.
5. Rank sources on cost per closing and net return, only for cohorts older than your lag.
6. For younger cohorts, check response rate and appointments set instead.
7. Make one budget decision, write down why, and check it against the same cohort next quarter.

If you want the rest of the scorecard around it, our real estate KPI checklist sets this next to response time, conversion and pipeline numbers.

Where Ace Trove fits

Ace Trove, our account-wide add-on for Follow Up Boss teams, does part of this work for you. Its Pipeline view includes a Source ROI card that ties each won deal back to the lead source on the deal’s primary contact. Per source, it shows won deals, GCI won and average GCI per deal, with the dollar figures labeled as estimates. Depending on the data available for your account, it can also show a deal win rate (only once a source has at least two decided deals) and the value of that source’s open deals weighted by each contact’s model win probability. That last figure is labeled as a forward-looking estimate and kept apart from realized GCI.

Here is the limit. The Source ROI card has no column for spend. It knows what each source produced. It does not know what you paid for it, so cost per closing and net return still come from your own books, using the four-number sheet above. Trove gets you the closings and commission half without a spreadsheet join. The cost half is yours.

Frequently asked questions

How do you calculate ROI on a real estate lead source?

Add up everything the source cost for a period, including any referral or success fees paid at closing. Count the closings that came from leads created in that same period, and the commission they produced. Cost per closing is total cost divided by closings, and the return is commission minus total cost.

Why is cost per lead a bad way to compare lead sources?

Because a lead is not the thing you get paid for. A source can sell cheap leads that rarely close, and a pay-at-close source shows a cost per lead of zero even though it takes a share of every commission it produces. Cost per closing compares sources on the outcome.

How long should I wait before judging a new lead source?

At least as long as your own typical gap between a lead’s first contact and a closed deal, which you can measure from your past closings. Judging a source on a 90-day window when your leads usually take longer than that to close will make every new source look like a loss.

One question for the operators reading this: how do you credit a repeat client whose first deal came from a paid portal? Back to the portal, or to your sphere? We have heard good arguments both ways.