# Lead Scoring for Loan Brokers: Stop Calling Leads Who Can't Qualify

> Most brokers sort leads by when they came in. The ones who close more sort them by whether they can actually fund. Here's a simple, written scoring model built on a consented soft pull, with worked examples.

Source: https://phonenumbercreditscore.com/blog/loan-broker-lead-scoring  
Author: Jordan Avery, Head of Content  
Published: 2024-10-10  
Topic: Lenders & Dealers  
Publisher: Phone Number Credit Score (https://phonenumbercreditscore.com/)

Loan lead scoring means ranking every new inquiry by how likely it is to become a funded loan, then working the list from the top. For a broker, the inputs that matter are simple: is this a real, reachable person, does their credit fit a product you can place, can they afford the payment, and do they want a loan soon. Get those four answers in the first minute and your team stops spending afternoons on people who were never going to close.

Most shops don't do this. They work leads in the order they arrive, call everyone three times, and let the best borrowers sit in the queue behind people with a 520 score and a bankruptcy from last spring. It feels fair. It's expensive.

## What a lead actually costs you

Run the numbers on your own pipeline. Say you buy or generate 400 leads a month, and a loan officer spends an average of 25 minutes per lead across calls, voicemails, texts and a partial application. That's about 167 hours a month. If 6% of those leads fund, 24 loans are carrying the cost of 376 conversations that went nowhere.

You'll never get the dead-end share to zero. But if a minute of checking up front tells you that 120 of those 400 leads have no realistic product today, you've handed your team roughly 50 hours back. That's the whole case for loan lead scoring. It isn't about rejecting people faster. It's about calling the right people first.

## The five signals worth scoring

### 1. Is the person real and reachable?

Start with the phone number, because it's the one thing every lead form collects. Does the name tied to the number match the name on the form? Is it a mobile line, a landline, or a VoIP number? What's the fraud score on the line? A mismatch or a high-risk VoIP number doesn't mean "fraud," but it does mean "verify before you invest an hour." Our posts on [mobile, landline and VoIP numbers](https://phonenumbercreditscore.com/blog/mobile-vs-landline-vs-voip) and [pre-qualifying leads by phone number](https://phonenumbercreditscore.com/blog/pre-qualify-leads-with-phone-number) go into the details.

### 2. Does their credit fit a product you can place?

Score bands only mean something relative to your product menu. For mortgages, HUD's rules allow FHA financing with 3.5% down at a credit score of 580 or higher, and [borrowers between 500 and 579 are limited to 90% loan-to-value](https://www.hud.gov/sites/documents/10-29ml.pdf), meaning 10% down. Most conventional programs start at 620. Personal loan and business lenders each have their own floors. Map your lenders' minimums first, then score leads against them, not against some generic idea of "good credit."

### 3. Can they afford the payment?

A 740 score with a 55% debt-to-income ratio is a harder file than a 660 with 28%. Look at total monthly debt payments next to stated income. The CFPB's [debt-to-income explainer](https://www.consumerfinance.gov/ask-cfpb/what-is-a-debt-to-income-ratio-en-1791/) has the simple formula, and each loan program sets its own ceiling.

### 4. Is there recent trouble?

Old problems fade. New ones don't. Late payments in the last 12 months or collections opened this year often tell you more than the score itself does. Read inquiries with care, though. A handful of mortgage or auto inquiries in a couple of weeks is usually just rate shopping, and FICO's models treat those as one inquiry within a shopping window. A string of new card and personal loan applications is the pattern worth a closer look.

### 5. Do they want a loan soon?

Ask on the form. "Within 30 days," "1 to 3 months," "just researching." Timeline is the only signal here that comes from the borrower's own words, and it's the easiest one to act on.

## A point system you can copy

Keep it written down and boring. Here's a starting model. Where two rows cover the same signal (score, debt-to-income), a lead gets one or the other, not both. Adjust the score bands to your actual lender minimums.

| Signal | Points |
|---|---|
| Name on the number matches the form | +15 |
| Mobile or landline (not high-risk VoIP) | +5 |
| Fraud score under 30 | +5 |
| Score clears your best program's minimum by 40+ points | +30 |
| Score clears at least one program's minimum | +15 |
| Debt-to-income under 36% | +15 |
| Debt-to-income 36% to 45% | +5 |
| No 30-day lates in the last 12 months | +10 |
| No collections opened in the last 12 months | +5 |
| Fewer than 3 hard inquiries in the last 90 days | +5 |
| Wants to borrow within 30 days | +10 |

Then route by total:

- **70 and up:** Call within five minutes. These are the loans you close this month.
- **45 to 69:** Call today, with a specific product in mind before you dial.
- **25 to 44:** Second-look lenders, a smaller loan, or a credit-building plan with a follow-up date.
- **Under 25:** A polite, useful email. Don't burn a rep's hour.

## Three leads, scored

All three are fictional, submitted on the same Tuesday morning.

**Dana Reyes**, (555) 555-0142. Name matches the mobile number, fraud score 6. FICO® Score 4 of 712, comfortably over your 620 conventional floor. Debt-to-income around 31%, no lates, two inquiries this year. Wants to buy in the next month. **Score: 100.** She gets a call before she finishes her coffee.

**Tom Albright**, (555) 555-0187. VoIP number registered under a different name, fraud score 41. Score of 598, which clears the FHA floor but nothing else. Debt-to-income near 44%. One 30-day late in March, no new collections, one recent inquiry. Timeline: "1 to 3 months." **Score: 30.** He's not a no. He might be an FHA borrower after a few months of clean history, or a fit for a second-look lender. He goes on a follow-up list, after someone confirms the number is really his.

**Keisha Monroe**, (555) 555-0163. Mobile number, name matches, low fraud score. Score of 655, clear of the conventional floor but not by 40 points. Debt-to-income 27%. No lates, no collections, one inquiry. Just researching. **Score: 75.** Strong file, soft intent. A fast call that's helpful rather than pushy probably wins this one, because whichever broker gives her a straight answer first usually gets the application.

![A laptop screen full of bar charts and line graphs, the kind of view a loan lead scoring dashboard gives you](https://phonenumbercreditscore.com/media/blog/01a1366f-70c3-79ba-9f9b-72148001c8dc-1200.webp)

## The rules that keep loan lead scoring legal

None of this works without the borrower's permission, and none of it should. A few points to build in from day one. This isn't legal advice, just the parts brokers most often get wrong.

**Get consent to check credit on the form.** Under the [Fair Credit Reporting Act](https://www.law.cornell.edu/uscode/text/15/1681b), you need a permissible purpose to pull a consumer's credit. The cleanest is the consumer's written OK. A clear checkbox on your lead form works: "I authorize [Brokerage] to obtain my credit report and score to review my loan options. This won't affect my credit score." Save who agreed, when, and the exact wording. Our post on [consumer consent for credit checks](https://phonenumbercreditscore.com/blog/consumer-consent-credit-check) has more sample language.

**Bought leads need your own consent.** If a lead vendor collected the form, make sure the consent language names you, or get it yourself before you pull anything. Screening a purchased list without consent isn't a permissible purpose. Sending unsolicited "firm offers of credit" to a list is a separate, regulated process called prescreening, and it isn't the same thing as scoring inbound leads.

**Credit consent isn't calling consent.** Permission to check credit says nothing about whether you can autodial or text someone. That's governed by the TCPA, [47 U.S.C. § 227](https://www.law.cornell.edu/uscode/text/47/227), and the FCC's new one-to-one consent rule, scheduled to take effect January 27, 2025, tightens how lead forms can collect it. Have your counsel look at your form before that date.

**Score on credit, never on who someone is.** The Equal Credit Opportunity Act and [Regulation B](https://www.consumerfinance.gov/rules-policy/regulations/1002/) prohibit discrimination based on race, color, religion, national origin, sex, marital status, age and receipt of public assistance, among other things. Keep your model to the signals above, apply it the same way to every lead, and stay away from inputs like ZIP code that can act as a stand-in for a protected trait.

The [permissible purpose guide](https://phonenumbercreditscore.com/blog/permissible-purpose-fcra) and our [compliance page](https://phonenumbercreditscore.com/compliance) go further if you want the detail.

## Making it automatic

Scoring by hand works for 20 leads a week. Past that, wire it into your intake. A common setup: the lead form posts to your CRM, the CRM sends the phone number and the consent purpose to our [API](https://phonenumbercreditscore.com/api), the response comes back with name, address, phone details, FICO Score 4 and the credit profile, and a few lines of logic turn that into a points total and a queue. Some teams push the result to a rep's phone with a webhook, so a 90-point lead gets a call while they're still on your website.

The API is $49 a month or $149 a year plus $7.50 per report, and lookups that don't find a match aren't charged. Our [credit score API guide](https://phonenumbercreditscore.com/blog/credit-score-api-guide) walks through the request, the response and the webhook setup.

Start with the table above, run it for a month, and compare funded loans against scores. The bands will need adjusting. That's the point. A written model you tune every quarter beats gut feel every time, mostly because you can see where it's wrong.