Each entry names who was measured, the year, and what the figure does not show.
Where no study exists, we say so.
The gaps are listed below in words, and no number stands in for them.
Every figure on the site is worked from these.
How we estimate shows each step, with the source beside it.
New leads
How fast lenders answer a new lead, how cold leads close, and what happens to pre-approved buyers.
Measured
21xbetter odds of converting
A mortgage lead reached within 5 minutes had 21 times the odds of converting of one reached after 30 minutes, an Insellerate speed-to-contact study of mortgage lenders found (2020).
Insellerate, a mortgage software company: mortgage leads at the lenders it studied, 2021 Comprehensive MBA Contact Study.
What it does not show
It describes lenders as a group, not any one loan officer, and a software vendor ran the study. It does not say how many of the uncontacted leads would have closed.
On LenderLogix’s platform in the third quarter of 2026, the average buyer spent about 93 days between pre-approval and application, after 8.3 pre-approval letters. Of its QuickQuals, 53% became applications.
“the average buyer spent about 93 days between pre-approval and application, and 53% of QuickQuals went on to become loan applications”
Who was measured
LenderLogix, a mortgage software vendor, Q3 2026, activity on its own platform. The unit is a QuickQual, its tool for pre-approval letters.
What it does not show
It counts QuickQuals, not individual buyers, and it does not follow the other 47%. LenderLogix says the extra time “didn’t show up as lost borrowers”. We use it only to show how long buyers shop.
“Loan officers, however, often only close about 3% of their leads.”
Who was measured
BankingBridge, a mortgage lead-capture software company, Mortgage Lead Conversion Part I, April 2023 (about 3% of leads close); leadPops, a mortgage marketing company (30 to 50% of leads from agents and other professionals close).
What it does not show
Both are vendors’ figures for leads in general, not a measured close rate for any one loan officer.
Who comes back for a next loan, who forgets their loan officer, and how many loan officers an investor keeps.
Measured
82%walk away for their next loan
82% of borrowers walk away and don’t return for their next loan, STRATMOR reports (2025). Its reading of why: after closing, “the communication drops off dramatically.”
STRATMOR Group, a mortgage consultancy, from its own data (no sample or period given), May 2025.
What it does not show
It counts borrowers who do not come back for their next loan; it does not say where they went. The reason given is STRATMOR’s reading, not a measurement.
“Refinance retention reached a 3.5-year high (28%) in Q3 2025”
Who was measured
ICE Mortgage Monitor, December 2025: every US refinance in Q3 2025, from ICE’s loan-level data.
What it does not show
It counts whether the company that collected the payments kept the refinance, not the loan officer, and it does not say which lenders won them. The share kept moves each quarter (35% in Q4 2025, 32% in Q1 2026), so the share lost moves too.
In August 2026, 0.64% of US mortgages were paid off in a single month, the lowest pace in 17 months. At that pace, about 7 in 100 are paid off in a year.
ICE First Look, August 2026, all US first-lien mortgages, reported by HousingWire on September 28, 2026. The yearly pace is the monthly rate compounded over twelve months.
What it does not show
It counts every payoff, including homeowners who sold and took no new mortgage, and it does not split sales from refinances. The monthly pace moves with rates.
Checked
Last checked
Used on
The home page card and Past clients stay yours
Measured
4.8million
On January 9, 2026, when the 30-year rate reached 6.04%, ICE counted about 4.8 million US borrowers who could have cut their rate by at least three quarters of a point, 20% more than before the drop. After a dip below 6% in February the count reached 5.4 million.
“On January 9, interest rates reached 6.04%... which put roughly 4.8 million borrowers “in the money” for a refinance... That drop effectively increased the eligible population by 20% overnight.”
Who was measured
ICE (Intercontinental Exchange, a mortgage data company), Mortgage Monitor, February and March 2026: a national count from ICE’s loan-level data. ICE calls these borrowers “in the money”: a refinance would cut their rate by at least three quarters of a point.
What it does not show
It is one dated national count, not a count of anyone’s clients and not a count of today: ICE’s Conforming 30-year Fixed Rate Index was 7.2% on September 24, so a count now would be smaller. It includes borrowers who may not qualify to refinance. We use it to show how fast the chance opens.
Checked
Last checked
Used on
The home page card and Rate drop retainment
Referral partners
Which loan officer agents use, why they switch, and what they send.
41.9% of high-producing agents (10 or more buyer-side sales a year) had a different #1 loan officer a year later, Mortgage Market Intelligence found across 88,823 agents (2026).
“Responsiveness and communication of the loan officer: 55.7%”
Who was measured
The Real Brokerage July 2025 Agent Survey, over 550 real estate agents across the United States and Canada, fielded July 31 to August 14, 2025; release dated August 20, 2025. Next in rank: rates 20.8%, speed and certainty of closing 13.3%, an existing relationship 10.3%.
What it does not show
A brokerage surveyed its own agents, and the sample includes Canada. It shows what agents say matters, not how many referrals a loan officer loses by going quiet.
“the average time taken decreased by 40% and output quality rose by 18%”
Who was measured
Noy and Zhang (MIT), “Experimental evidence on the productivity effects of generative artificial intelligence”, Science, July 2023: 453 college-educated professionals given occupation-specific writing tasks.
What it does not show
It measured writing tasks such as emails and short reports, not loan officers or mortgage messages.
Checked
Last checked
Used on
AI speed to lead, “For you” (hours saved)
Measured
1%per loan
Loan officers were paid about 1% of the loan amount per loan (101.3 basis points) in the second quarter of 2021. LBA Ware’s own measured average was $2,876 a loan.
LBA Ware compensation data, second quarter of 2021, reported by Scotsman Guide on July 28, 2021.
What it does not show
Pay varies widely, and refinances paid less (91.7 basis points). The feature pages count $3,500 a loan, about 1% of a $350,000 loan.
Checked
Last checked
Used on
Every feature page, “For you” ($3,500 a loan, about 1%)
What we looked for and could not find
Where a study does not exist, the page says so and prints no number.
How many calls and inquiries to loan officers go unanswered today?
We found no current study we could verify. The tests we found were run by lead-software vendors and their own reports disagree, so we print none.
How much does a faster first reply win a mortgage lead?
For mortgage leads alone, we found no figure we could verify. The 21x we print is Insellerate’s 5-minute against 30-minute comparison, and counts a converted lead, not a closed loan.
How often is a quiet lead won back?
No published figure we could verify measures it, so we print none.
How often does a referral partner stop sending business?
No published figure measures it. Lack of communication can cause referral partners to leave, so our system keeps them informed.
How much do stage updates keep an agent sending business?
No published figure we could verify measures it, so we print none.
How many pre-approved buyers never buy at all?
No source we could open says. LenderLogix measures how long buyers shop and how many QuickQuals became applications, not how many never buy.
How many borrowers never hear from their loan officer after closing?
We found no US survey that asks.
How many past clients come back to the same loan officer?
We found no published measure for one loan officer. The nearest is STRATMOR’s: 82% of borrowers don’t return for their next loan.
How many borrowers could save by refinancing today?
ICE last published a count in March 2026, when rates were near 6%. Its rate index was above 7% in September, so we print the dated count and no figure for today.
How comfortable are borrowers with AI?
We found no independent survey we could open. The site describes what the AI drafts and who approves it, and prints no comfort figure.
Figures we do not print
People quote these. We could not trace them to a study we can open, so they stay off this site.
A widely quoted open rate for text messages: we could not find the study behind it.
A claim that most buyers choose the first lender to respond: it is not in the study it is credited to.
A claim that half of agents drop a lender after one bad experience: we found no source.
A claim about how many pre-approved buyers never buy: we found no study that measures it.
Average production figures for loan officers from pages that cannot be opened.
A conversion gain of several hundred percent for calling a new lead fast: the figure is quoted from different studies and we could not confirm which one it belongs to or what it is compared with.
A 2011 mystery-shopper result for how many lenders called a new lead within a day: the test’s own reports give different figures, and larger tests exist.
A claim that most real estate agents want better communication from lenders about loan status: we could not open the survey sentence behind it.
A widely quoted share of agents said to have left their loan officer because they stopped hearing from them: we found no study behind it.
A claim that most buyers say at closing they would use their lender again: that survey asks about real estate agents, not loan officers.
How we choose figures, and fix errors
We print a figure only when we can open its source, and we name who was measured and what it does not show.
Think a figure is wrong? Email sales@mpowersourcing.com with the figure and the page. We check it against its source and correct it wherever it is printed.
Last reviewed .
Stop leaving loans on the table.
Fifteen minutes on your business, and an honest answer on fit.