Mortgage Lead Conversion Rates: How to Measure Yours Correctly

There isn’t a single mortgage lead conversion rate worth comparing yourself to. The number depends entirely on where the lead came from, what you counted as a lead, and how long you waited before scoring the outcome. Published benchmarks blend all three away, which is why yours will never match them. Build your own instead.

That’s unsatisfying if you came here for a figure to drop into a board deck, and it’s the honest answer. What follows is how to build a conversion rate that means something, starting with the three things that quietly break most of the ones you’ve seen quoted.

The denominator problem

Conversion rate is closings divided by leads, and nearly all of the disagreement lives in the denominator. Two lenders can run identical marketing, close identical loans, and report rates that look nothing alike, purely because they draw the line around the word “lead” in different places.

Ask five shops what counts as a lead and you’ll get five answers. Some count every form fill, including the ones with a disconnected number and an email that bounces on the first send. Some count only the leads a loan officer actually reached. Some treat a past client filling out the same form a second time as a brand new lead, and some don’t.

None of those definitions is wrong. They’re just different, and the number they produce doesn’t travel. Answer three questions in writing before you compare anything to anything:

• What stage counts as a lead? Raw inquiry, verified contact, or qualified prospect. Pick one and apply it to every source.
• Which sources share the pool? Purchased leads, organic web forms, realtor referrals, past clients and walk-ins behave nothing alike and shouldn’t sit in the same denominator.
• What window are you measuring? Leads that arrived in a given month, or loans that closed in a given month. Those are two different reports, and mixing them is the most common reason a conversion rate moves for no visible cause.

Two sources, one average, no signal

The most expensive habit in this exercise is averaging unlike sources together. A purchased internet lead and a realtor referral are not the same product. One is a stranger who filled out a form on a rate table, very likely filled out several others the same afternoon, and has no particular reason to trust you yet. The other arrives with somebody else’s credibility attached and often a property already in play. The referral converts far better and costs a relationship to earn. The internet lead converts far worse and costs cash. Both can be perfectly good business.

Average them and you get a figure that describes your source mix rather than your performance. If the blended rate climbed last quarter, did your team get better at selling, or did referral share go up because one agent had a strong spring? The blended number can’t tell you, and it will quietly reward you for buying fewer leads even when those leads were profitable.

Report by source, every time. A blended figure belongs on one slide at most, and even there it needs the source mix printed beside it.

Lag, and why this month’s rate is fiction

Mortgage runs on a slow clock. A lead takes time to become a conversation, the conversation takes time to become an application, and the application takes weeks of processing before anything funds. The lead you paid for in March is closing in May at the earliest, and plenty of good ones sit longer than that before the borrower is ready to move.

So when you divide this month’s fundings by this month’s leads, you’re comparing two groups of people who have almost nothing to do with each other. In a growing month the math flatters you, because a small older cohort is producing closings against a large new cohort of leads. In a slow month it punishes you for the same reason in reverse.

Measure by cohort instead. Stamp every lead with its arrival date and its source, then score outcomes against the group it arrived with rather than the month it happened to close in. Recent cohorts will always be incomplete, so label them that way in the report instead of pretending they’re final.

You also need an attribution window you’re willing to defend. Pick a length that covers your normal cycle with room on both ends, write it down, and stop moving it, because a channel judged before its cohort has matured will always look like it failed. Leads that convert well past the window still count, but keep them on a separate long-cycle line rather than reopening old cohorts, so your channel comparisons stay stable.

The rates worth tracking

One overall conversion rate tells you whether the quarter went well. Stage-to-stage rates tell you what to go fix, which is the only real reason to track any of this.

Notice how much of that table is about record-keeping rather than marketing. A conversion rate is only as good as the discipline behind the stage changes underneath it.

Building your own baseline

Your baseline is the only benchmark that can tell you anything, and it costs one afternoon plus a few months of patience.

1. Write the definitions down. What a lead is, what a contact is, what an application is, and when a record gets closed as lost. Circulate them so everyone touching the pipeline uses the same rules.
2. Pull enough closed history to matter. You need cohorts old enough that nearly everything in them has either funded or died, and you need more than one of them.
3. Compute the stage rates per source. Never blended. If a source is too small to produce a stable rate, say so in the report rather than quoting a figure one loan could swing by half.
4. Record the range, not just the average. Knowing the band a source normally lands in tells you when a bad month is genuinely a bad month.
5. Set your floor from your own history. What triggers a conversation about a channel should come from what that channel has done for you, not from a figure you found online.
6. Re-baseline on a schedule. Rate environments move and referral partners come and go, and a baseline built in a refi wave will lie to you in a purchase market.

When a channel comes in under your baseline

Resist the urge to cut it that afternoon. Work the list in order, because the first two items are more often the culprit than the channel is.

Check whether the cohort is mature. If those leads are younger than your attribution window, you don’t have a result yet, you have a partial one.

Check whether the tracking survived. Confirm the source is still tagged correctly, that a form or integration didn’t quietly break, and that duplicates aren’t padding the denominator. Broken attribution looks exactly like a failing channel on a dashboard.

Find the stage where it falls off. A source that dies between lead and first contact is telling you about your response process and staffing, not about the leads. A source that dies between contact and application is telling you about targeting, intent or the offer. A source that dies after application is usually about qualification or operations, and cutting marketing spend won’t touch it.

Then compare cost per funded loan. Conversion rate alone will steer you wrong here, because a source with a low conversion rate and low acquisition cost can produce cheaper closed loans than a high-converting source you can’t scale.

Change one thing and wait a full cycle. Adjusting the offer, the routing and the follow-up cadence at once leaves you with no idea which move worked, and in a business with this much lag you don’t get many clean reads a year.

Getting the data to hold still

All of this depends on lead source, stage history and dates living in one system that everybody actually uses. When the source sits in a spreadsheet, the stages sit in the loan origination system, and the follow-up sits in somebody’s inbox, the conversion rate you report is a guess wearing a decimal point.