Mortgage / Home Loans pay per call & leads

Who's calling your loan officers

Mortgage inbound is a mix that refuses to sit still. Some callers are under contract and need a pre-approval yesterday. Some are rate-watching a loan they already have. Some want cash out of a house they don't actually have equity in. Some are first-time buyers who still think a 580 middle score and no down payment is a product you can 'just find.' Your LO's job is origination. The campaign's job is to stop the last group from eating the calendar.

Purchase callers usually have a timeline: an offer, a realtor, a listing they walked through last weekend. They can name a property type and a ZIP. Refinance callers can name an approximate balance and a current rate. The dangerous middle is the 'just checking rates' caller with no property, no credit picture, and no timeline. They will talk. They will not file.

Credit-bruised callers are not automatically junk. FHA and non-QM desks exist for a reason. The junk is the mismatch: a conventional-only shop taking 540 credit calls all morning, or a non-QM desk paying for pristine 780 purchase traffic it doesn't need. Write the credit band you originate. Don't write the credit band you wish you originated.

Self-employed, ITIN, investment-property, and jumbo callers show up inside generic 'mortgage' media whether you want them or not. If those are your products, great, filter for them. If they aren't, they are expensive conversations. A self-employed caller with two years of returns is a file for some desks and a polite decline for others. The campaign cannot guess. The brief has to say.

Real-estate agents, wholesale flippers, and 'I help my clients' third parties also leak into consumer campaigns. Decide if you want those. A broker shop might. A consumer-direct lender often doesn't. Third-party callers will stay on the line and still never be the borrower. That's a qualification item, not a surprise.

Rate moves flip purchase vs refinance mix overnight

This is the mortgage-specific fact that wrecks campaigns copied from last quarter. When rates drop, refinance and cash-out interest wakes up. When rates jump, purchase and 'I need to know if I can still buy' stays, and refi shoppers go quiet. Your media mix, your talk track, and your LO seating have to follow that, or you'll pay purchase CPAs during a refi wave and wonder why lock rates look terrible.

Don't run a single 'mortgage calls' bucket through one greeting if purchase and refi need different questions. Purchase needs occupancy, property type, timeline, and whether they're working with an agent. Refi needs approximate balance, current rate, occupancy, and cash-out versus rate-and-term. Mixing those IVRs produces callers who pass a generic 'are you interested in a home loan' and fail every product your desk actually sells.

Rate-watchers will redial. They called you in March. They call again in June because a headline said rates dipped. Duplicate windows and 'existing customer / existing pipeline' rules matter more here than in verticals where the consumer has one crisis. If your CRM already owns that ANI, you should not pay full freight for another transfer. Put that in the brief.

Creative has to match the environment you're actually in. A 'refinance and save' ad in a purchase-heavy rate world will still get clicks from dreamers, and your LOs will hate the phone. A 'get pre-approved to buy' ad during a refi surge will miss the people who would have filed this week. We can shift supply. We cannot make a stale offer perform. You have to tell us when your product priority changes.

None of this means you chase every rate headline with a new campaign. It means you schedule a mix review when the 30-year moves enough that your lock desk already changed pricing. If your capital-markets team knows the world flipped, your media brief should know it the same week, not at the monthly QBR.

Credit band and loan purpose protect LO time

Loan-officer time is the scarce inventory. A twenty-minute call with a 520 score and a cash-out request on a house they don't own is not 'a lead that didn't convert.' It's a scheduling theft. Credit band is how you stop that theft. Purpose is the other half: purchase, rate-and-term refinance, cash-out, and sometimes a HELOC or second that your shop doesn't even originate.

Set credit as a range you can actually price, not a humblebrag. If you start conversations at 620 for conventional and 580 for FHA, the screen should say that. 'Good credit' is not a filter. Callers with a 600 will claim good credit. An IVR that asks for a band (under 580, 580-619, 620-679, 680+) is blunt and useful. Human screens can ask for a recent pull or a self-report and still be honest that self-report is messy.

Loan purpose mismatches are as costly as credit mismatches. Cash-out shoppers who want to consolidate debt are not purchase callers. They also aren't automatically debt-settlement overflow; some of them are valid cash-out files if equity exists. Ask purpose early. Route purchase to purchase LOs. Route cash-out to the people who know overlay and equity math. One queue for all purposes is how junior LOs get crushed.

DTI and occupancy can wait until the LO, but bankruptcy, foreclosure, and housing event timing often cannot. If your overlays kill anyone with a foreclosure in the last three years, ask it. If you have a product for that, don't ask it as an auto-fail. Same logic as debt settlement's floor: the hard disqualifier belongs before the expensive human.

Protecting LO time also means concurrency and hours. Mortgage callers cluster around lunch, early evening, and Sunday real-estate browsing. A branch that staffs 10 to 4 and buys 8 a.m. to 9 p.m. will miss the calls it paid for and then call the traffic 'bad.' Cap to staffed hours or pay a call center to take overflow with a booked appointment into the LO calendar. Unanswered qualified calls are still a cost. They're just a cost you don't see on the publisher invoice the same way.

Purchase, rate-and-term, and cash-out are different jobs

Purchase files live and die on timeline. Under contract is a different animal than 'I might buy next year.' If your desk needs contracts or a 30-day close, the campaign should prefer that language. If you're building a nurture pool of first-time buyers, CPL may beat live transfer. Paying transfer rates for a 12-month shopper trains everyone to think mortgage pay per call is overpriced.

Rate-and-term refinance is a math conversation: current rate, remaining term, break-even, and whether a fee-heavy refi even helps. These callers can be sharp and still not proceed. That's origination, not traffic fraud. What is a traffic problem is the caller who has a 2.75% loan from years ago and wants you to 'beat it' in a 6% world. Screen current-rate band if your desk cannot help those files. Yes, you'll cut volume. You'll also cut wasted lock-desk conversations.

Cash-out is where debt-consolidation stories walk in. Some are solid equity plays. Some are people who need debt settlement, not a larger first mortgage. Your LO should not spend a half hour discovering there's no equity. A rough home-value and balance question up front is allowed. It will be wrong sometimes. It will still catch the renter who thought cash-out meant a personal loan.

Government products (FHA, VA, USDA) versus conventional versus non-QM should be a filter when your license and overlays demand it. VA callers with COE questions are not USDA acreage callers. A 'we do all loans' greeting is fine for a true broker shop with the seats to match. It is not fine for a captive FHA team of four. Product honesty in the brief is cheaper than LO burnout.

Investment and second-home intent needs its own gate if you don't want it. Media that says 'home loan' will attract house hackers and Airbnb math. That's a real business for some buyers. For a primary-residence retail desk, it's a talk-track detour. Ask occupancy. Primary, second, investment. Three buttons. Save the stories for the LO who can use them.

CPL, pay per call, transfers, and traffic into your funnel

Pay per call fits when an LO or a licensed team is sitting there ready to run credit and talk product. You pay when the conversation meets credit, purpose, geo, hours, and duration rules. You do not pay for a funded loan. Anyone who sells you 'guaranteed closings' is selling something else. A qualified mortgage call is a shot at a file. Your pull-through is still your pull-through.

CPL fits application starts, rate quotes on your site, and booked appointments. It's the right model when the consumer will not take a live transfer (they're at work, they're with an agent, they want to type numbers first). Keep the same credit and purpose filters on the form. A CPL that only asks email and 'interested in mortgages' will look cheap and price out like garbage.

Warm transfers should mean someone already captured purpose, credit band, state, and occupancy. Live transfers can be click-to-call with a short IVR. Both are valid. What you should not do is take raw number connects with no screen and then judge RidgeRise (or any supplier) on LO pull-through. That's not a quality program. That's hope.

Exclusive vs shared is sharp in mortgage because borrowers application-shop. A shared purchase caller may already have two AUS runs in the last hour. Some broker shops live on that. Consumer-direct teams that need a clean credit pull often don't. If you buy shared, your LO has to be faster and clearer than the last person, not offended that a competitor exists.

Qualified traffic means clicks or redirects into your own pages, your own disclosures, your own call offer. Useful when you have a rate engine and want to own the experience. Useless when the landing page is a thin 'get a call' form that skips credit and purpose. Hybrid supply (our media buying plus vetted publishers) can sit behind any of these models. We'll tell you the mix on a campaign. We won't pretend it's all in-house, and we won't pretend we have no control over partners.

Filters that stop junk from hitting the LO queue

State licensing is the first geo rule. If you're not licensed in a state, do not buy it. Mortgage is not like some home-services verticals where you can 'refer it out' casually. Unlicensed origination conversations are a compliance event. Map NMLS coverage to the campaign. Update it when you add states, not six weeks later.

Hours should follow borrowers, with a cap at staffed capacity. Evening purchase calls from working buyers are often better than 11 a.m. rate-window shoppers. That's desk-specific. Measure it. Then set hours on purpose. Buffer after each transfer so the LO can run the next step instead of flipping to a second live caller with a half-written 1003.

Duration rules in mortgage should reflect a real qualification-plus-needs conversation, not a voicemail. If your LO's good call is twelve minutes, a 90-second billable gate will invoice a lot of 'what's your purchase rate' hang-ups. Pair duration with dispositions: wrong credit band, wrong purpose, unlicensed state, not the borrower, duplicate. Duration alone is a blunt instrument. You still need it. You shouldn't worship it.

Concurrency caps protect pull-through. Mortgage shops love to 'turn it up' after one good week. Then three publishers spike on a rate headline and your five LOs each have two calls waiting. Everyone's conversation gets worse. Caps exist so the calls you paid for can be handled like origination instead of a call-center dump.

Do-not-contact and existing-customer filters belong in the routing logic if you can support them. Calling your own servicing customers as if they were new internet leads is how you generate complaints. If you can't hash-match, at least use a duplicate ANI window and a script line that checks 'are you already working with us.' Cheap. Effective.

Qualification that matches how your desk originates

A practical mortgage screen: state, purchase vs refinance vs cash-out, self-reported credit band, occupancy, and whether they are the borrower. Add 'under contract / shopping / 3+ months out' for purchase if timeline matters. Add VA/FHA/conventional only if you truly cannot mix. That's enough to protect LO time without running a full TRID application in the IVR.

Don't pull credit in the IVR. Don't ask for Social Security numbers before a human. Qualification is routing, not origination. If a publisher's path is collecting full 1003 data before transfer, that's a different product (and a different consent conversation). Most live-transfer programs should not do that.

Self-employed and income-type questions can be a single fork if your desk is W2-retail only. 'Are you paid on a W2, self-employed, or both?' Three options. Transfer W2 to retail. Transfer self-employed only if you have a desk for it. Parking everyone in one queue because 'the LO will figure it out' is how your best purchase LO spends Thursday on a bank-statement file they aren't allowed to do.

Language, third-party, and 'I'm just getting a number for my client' should fail or reroute. Realtor referral programs are a separate campaign with a separate brief. Don't blend them. You'll ruin consumer consent language and you'll confuse payout logic for publishers.

Write pass/fail answers down. 'Credit around 600' might pass your FHA band and fail conventional. The IVR should not require the consumer to know product names. It should capture the band and let routing pick the desk. When we QA a call, we need that mapping, not a post-hoc story about what the LO felt.

Quality monitoring when a long call still isn't a file

We monitor sources: duration, qualification answers, geo, duplicate rate, and what your dispositions say. A source that suddenly tilts to 540 credit and cash-out-on-a-rental will get paused. Recordings are the evidence. Your LOs' Slack complaints are a signal, not a dataset. Send dispositions if you want speed.

A long call can still be a bad buy. An LO who is kind will educate a first-time buyer who is two years away. That call might bill on duration. It might still be the wrong inventory for a transfer campaign. Fix it with timeline filters and a cheaper CPL path for early shoppers, not by accusing the source of fraud. Fraud is fake intent, wrong geo, IVR manipulation. Education is a filter miss.

TCPA-aware, campaign-specific consent is the standard language. Mortgage marketing has its own statutory pile on top of that. We will not say guaranteed compliant or zero risk. If you have a required consent string, a recording disclosure, or a state-specific oral notice, put it in the brief so the path that generates the call can match the path that receives it.

Dispute process should separate invalid traffic from 'didn't pull credit' and 'didn't lock.' Pull-through is your origination. Invalid is our filter miss. If you mix them, you'll cut your best purchase source because your LO was slow on a Tuesday. We'll work disputes against the written rules. Keep the rules tight enough that the argument is possible.

Source-level cutoffs are how hybrid supply stays usable. In-house paths and publishers get the same scorecard. A partner who won't fix creative when the mix goes rotten loses the campaign. That's the control. It's not a promise that every remaining call funds.

Seasonality that isn't only spring homebuying

Spring and early summer still matter for purchase. More listings, more under-contract panic, more evening calls. If your purchase team is already at capacity from realtor referrals, maybe you don't buy more purchase transfers in May. Buy refi, or buy a different geo, or don't buy. Adding paid calls on top of a full pipeline just creates abandoned qualifieds.

Year-end is a closing-calendar problem. Callers who want to close for tax reasons show up late. Your processors are already underwater. Concurrency should go down in the weeks you cannot open new files, even if media is cheap. Cheap calls into a frozen ops team are not cheap.

Refinance waves don't follow the listing calendar. They follow rates and, sometimes, a headline. You cannot staff a full refi floor 'just in case' all year. You can have a brief ready: credit band, purpose, current-rate filter, hours, exclusive vs shared. When you want it on, we turn supply toward that brief instead of inventing a new vertical from scratch.

Weather and local economies change purchase more than national averages show. A factory closing, a military PCS cycle, a regional price drop. Geo filters should be allowed to move. National mortgage media with no state-level pause is how you keep buying into a dead MSA while your licensed LOs sit in a living one.

First-of-month and mid-month payday patterns show up on cash-out and credit-sensitive purchase. It's not a law of nature. It's worth watching on your own dispositions. If Friday night cash-out is consistently below band, stop buying Friday night cash-out. That's an hours filter, not a vertical problem.

Lenders, brokers, and mortgage call centers

Retail lenders with licensed LOs buying for a branch or a region need tight state and product filters. Your advantage is brand and pricing. Your risk is paying for broker-style shoppers who want four quotes. Exclusive and a clean purpose screen help. Shared may still work if your price wins and your answer rate is high.

Brokers and mini-mags often want broader credit and product mix, and they can sit on shared traffic if the LO is a closer. Their risk is compliance consistency across a floor of independent originators. If half your LOs won't take FHA, don't buy FHA. Internal acceptance is part of the brief.

Mortgage call centers and net branches buying central intake need concurrency math more than anyone. Central intake can take volume. Central intake can also destroy brand experience if the appointment never shows up on the local LO's calendar. Define the handoff. A warm transfer to intake plus a booked LO callback is a different product than a transfer straight to a licensed originator.

Credit unions and community banks often have overlay and membership gates. Put membership and footprint in geo. A national campaign into a five-county CU is waste. If you have a membership workaround, say so. If you don't, don't buy outside the footprint and hope the closer 'explains the family membership.'

Agencies buying for a lender client must bring overlays, licensed states, and whether the client wants purchase, refi, or both this month. A stale IO from a rate environment two moves ago is how everyone gets mad. Refresh the brief when pricing changes. Come through /buyers if you want the model explained, then /contact with the current overlay sheet.

Publisher note for mortgage traffic

We buy and we partner. Publishers with rate, homebuying, refinance, and real-estate intent can fit if the path can collect purpose and a credit band without lying about the offer. 'See if you qualify' is fine. 'You're pre-approved' is not, unless that's actually true, which it usually isn't on a media path.

Search, native, and call units all appear in this vertical. Email remnant and incentivized 'rate alert' lists are where quality goes to die. If your traffic cannot survive a recording review against credit and purpose, it will not stay live. Payouts get confirmed in the partner process, not here.

Consent has to match a mortgage offer, not a generic finance checkbox. Campaign-specific, TCPA-aware, no 'guaranteed approval' claims. If you run publishers, you already know this. If you're a buyer reading this section, it's the 20%: supply is hybrid, partners are vetted, junk paths get cut.

Mistakes unique to buying mortgage calls

Running one campaign through a rate-environment flip without changing purpose mix. You'll measure 'mortgage quality' while actually measuring a product you stopped wanting. Split purchase and refi. Reweight when rates move.

No credit band. LOs then play amateur underwriter on every call and duration looks healthy because they're explaining why a 500 score isn't a conventional 30-year. That's education, not origination. Screen the band.

Buying unlicensed states because 'volume was there.' Volume in a state you cannot originate is a complaint waiting for a license check. Map NMLS to geo. Keep it current.

Treating cash-out debt consolidators as equivalent to purchase under-contract files on the same CPA target. They are different jobs, different talk time, different pull-through. Price and filter them apart or stop being surprised.

Judging sources on funded units in week one. Mortgage files take time. Early signals are qualification match, credit-band accuracy, appointment keep rate, and application start. Funded is lagging. If you cut a good purchase source on day six because nothing closed, you taught yourself the wrong lesson.

Skipping duplicate and existing-customer rules in a vertical where people rate-watch for months. You will pay for your own pipeline. You will also annoy borrowers who thought they already had an LO.

How to brief so we don't burn your origination calendar

Send licensed states, products you will actually originate this month, credit bands by product, purchase vs refi vs cash-out mix, occupancy rules, hours, concurrency, exclusive vs shared, duration, and duplicate window. Add self-employed handling and third-party handling. Add the IVR map.

Tell us what a call you're happy to pay for sounds like. Better: send a recording. 'Under contract, 680, primary, licensed state' is a brief. 'High-intent mortgage consumers' is a wish. We can screen the first. We cannot screen the second.

Name the model. Live transfer to licensed LOs during staffed hours. CPL for early purchase shoppers. Traffic into your rate engine if that engine asks real questions. Hybrid supply can support the mix. One definition of qualified should cover all three or your dashboard will lie.

Call out compliance needs: recording notice, consent string, state oral disclosures, any brand claims we cannot make in ads. TCPA-aware and quality-monitored, campaign-specific consent. No 100% compliant stamp. No zero-risk pitch.

When the rate world moves, send a short update, not a new RFP. Purpose mix and credit overlays change faster than legal entities do. For the buying models, see /buyers. To put a desk on a campaign, use /contact and include the overlay sheet and the hours you can actually answer.