Guide
Mortgage pull-through and lead conversion: what the published numbers say.
About seven in ten retail mortgage applications close at independent mortgage banks, and just over half at banks: 69% and 55% in the first half of 2025, the lowest in more than a decade. Earlier in the funnel there is no independent benchmark. The few public figures for online leads run from about 3% to 8.5% reaching a funded loan, and they measure different things.
By Amit Ghole, President and CTO, inflooens · November 13, 2026 · Every figure links to its source.
Application to closing
The part of the funnel with solid numbers
Two sources publish it. They count different things, so read each against its own definition and do not average them.
| What is measured | Figure | Source | What to know |
|---|---|---|---|
| Applications that close, independent mortgage banks, retail | 69% | MBA and STRATMOR Peer Group Roundtables, first half of 2025 | The lowest since 2012. Counts applications as RESPA defines them. |
| Applications that close, banks and other depositories, retail | 55% | MBA and STRATMOR Peer Group Roundtables, first half of 2025 | The lowest since the survey began reporting it in 2000. |
| Home purchase applications that were originated, all lenders | About 70% | CFPB, 2023 Mortgage Market Activity and Trends, Table 1A | 3.25 million originations on 4.61 million applications in 2023. The division is ours. |
| Refinance applications that were originated, all lenders | About 46% | CFPB, 2023 Mortgage Market Activity and Trends, Table 1B | 0.80 million originations on 1.75 million applications in 2023. The division is ours. |
| Home purchase applications denied | 9.4% | CFPB, 2023 Mortgage Market Activity and Trends, section 3.3 | Refinance applications were denied at 32.7% in the same year. |
The industry survey is the closer match to how a lender thinks about pull-through. It covers the retail channel of the lenders that take part, and it counts an application only once the borrower has given the six pieces of information RESPA names. The survey’s authors note that the rate would be lower still if leads and preapprovals were counted.
The federal figures come from the data nearly every lender reports under the Home Mortgage Disclosure Act. They cover the whole market and include applications that were withdrawn or left incomplete. The CFPB report used here covers 2023 and is the latest edition we found.
Lead to funded loan
The part with weak numbers
We found no independent benchmark for how many mortgage leads become funded loans. These are the two public figures we could open and read, with the reasons to treat each with care.
| What is measured | Figure | Source | What to know |
|---|---|---|---|
| Online mortgage leads that ended in a funded loan, with any lender | 8.5% | iLeads.com press release, January 2018, on leads from the first half of 2017 | Nearly 2.75 million leads, matched to funded loans through property records. It counts a loan with any lender, not the lender that bought the lead. From a lead-data company, and old. |
| Online leads that became funded loans at the lenders studied | About 2.7% | National Mortgage News, 2026 Lead Vendor Analysis Report (sponsored) | 77,300 funded loans on 2.9 million online leads across 29 organizations. The division is ours, from the report’s own summary. The full report sits behind a registration form. |
The two are not in conflict. One asks whether the person who filled in a form took out a mortgage anywhere. The other asks whether the lender that bought the lead funded it. The difference in what they count is business a lender paid for and lost to someone else. The figures are years apart and come from different samples, so the size of that loss cannot be read off them.
What has no source
Four numbers people quote that we could not find published
If someone gives you one of these as a benchmark, ask where it comes from.
Referral leads
We found no published rate for leads that come from real estate agents and other referral partners, although purchase lenders depend on them. Lenders generally expect them to close far more often than bought leads.Lead to first contact
No independent figure for how many leads are ever reached. Consultants report that no response at all is a common failure, without publishing a rate.Lead to application
No independent figure. The preapproval counts in federal data cover only a small, formal subset of preapprovals.Speed of response
The claim that a faster call lifts conversion is widely repeated. We could not confirm a mortgage-specific study with a base rate, so we do not quote a number.
Your own numbers
How to measure pull-through so it can be compared
A published rate is only useful next to yours if both were counted the same way.
1
Fix the starting line
Decide what counts as an application. The industry survey uses the RESPA definition: name, income, Social Security number, property address, estimated value and loan amount. Count from an earlier point and your rate will look lower than the published one.2
Follow a group, not a month
Closings in a month divided by applications in the same month mixes different loans. Take the applications from one period and follow them to the end.3
Split purchase from refinance
They behave differently. In the 2023 federal data, about 70% of purchase applications were originated and about 46% of refinance applications were.4
Split by where the lead came from
A referral from an agent and a bought internet lead are different businesses. One blended rate hides both.5
Record why loans fall out
Denied, withdrawn, went elsewhere, never responded. The published totals do not say why, and the reason is what you can act on.
Questions to ask any vendor
When someone promises to lift your conversion
Including us. A number without these answers is a slogan.
- 1What exactly was counted at the start, and what counted as success?
- 2Is the figure from your customers’ own data, from a survey, or from an estimate?
- 3How many lenders and how many loans is it based on, and over what period?
- 4Is it purchase, refinance or both, and which lead sources?
- 5What was the rate before, measured the same way?
- 6Can we see the source?
How inflooens uses these numbers
One sourced rate, and the rest is yours to set
inflooens publishes no customer conversion results. The business-case calculator on our home page starts from the industry survey’s application-to-closing rate and lets you set every earlier stage yourself. The earlier stages are examples, because nobody publishes them.
Straight answers
Questions about pull-through and conversion
- What is pull-through rate in mortgage?
- Pull-through is the share of loan applications that go on to close. The Mortgage Bankers Association measures it as total closings divided by total applications in a period, counting applications as RESPA defines them.
- What is a typical mortgage pull-through rate?
- In the first half of 2025, retail pull-through was 69% at independent mortgage banks and 55% at banks and other depositories, according to the MBA and STRATMOR Peer Group Roundtables. Both were the lowest in more than a decade.
- What is a typical mortgage lead conversion rate?
- There is no independent benchmark. The public figures for online leads range from about 2.7% becoming funded loans at the lenders studied, in a sponsored 2026 report, to 8.5% ending in a funded loan with any lender, in a 2018 lead-data company release. They measure different things and should not be compared directly.
- Why do federal figures differ from the industry survey?
- They count different things. Federal HMDA data covers nearly all lenders and counts applications that were withdrawn or left incomplete. The MBA and STRATMOR survey covers its participating lenders’ retail channel and counts RESPA applications. Neither is wrong, and they are not interchangeable.
- Does inflooens publish its customers’ conversion results?
- No. The business-case calculator on the inflooens home page uses one sourced rate, the MBA and STRATMOR application-to-closing figure, and example rates for the earlier stages that a visitor can change. It is an illustration, not a benchmark and not a customer result.
Sources opened and figures re-read on November 13, 2026. Where a percentage is marked as ours, we divided two numbers the source publishes.
More guides
Keep reading
- Mortgage CRM vs POS vs LOS
- How many tools a loan officer uses, and what they cost
- Two-way sync with your LOS: what it should mean
- Loan officers on personal phones: what it costs
- How to track referral partners and who sent each loan
- Twelve questions to ask a mortgage AI vendor
- What an AI agent should be allowed to do on a loan
- Who should hold a lender’s loan data?
- How to pilot new software while under contract elsewhere
See where your loans fall out.
Thirty minutes. Your funnel, your numbers, a live file.