Is pay per lead cheaper than an agency retainer?

It is more predictable, which is not the same thing. Pay per lead fixes the cost of one record: public 2026 tables post $10 to $50 shared, $50 to $120 exclusive, and $80 to $200 and up for a live transfer. A retainer fixes a month of effort at a price none of the agencies surveyed publish. In a good month the retainer can be cheaper per application, because the fee is spread over more of them. In a dry month it is infinitely more expensive per application, because the fee is owed and the applications are not there. Which one is cheaper depends on how many dry months a shop can absorb.

What does a $60 lead have to do to beat a $0.05 aged record?

Fund about one deal for every twelve hundred the aged file funds, and that is the arithmetic every shop should run on its own numbers rather than on a vendor's. The published ladders make the input side easy: aged records at $0.05 to $0.50 on a 2026 wholesaler ladder, exclusive real-time records at $50 to $120 on a vendor table. The output side is the part no table holds. A marketplace index measures real-time inventory at 10 to 50 times the aged price, which is the market's own estimate of the difference, and it is an estimate about willingness to pay rather than about funded deals.

Can a small ISO run Meta ads for merchant cash advance?

It can, under a policy that never names the product. Meta's prohibited financial products policy bans ads for short-term loans of 90 days or less, payday loans and paycheck advances. Its restricted financial services policy repeats that wording and adds an 18-plus targeting rule and disclosure requirements for permitted finance ads. Merchant cash advance appears in neither list. The practical consequence is that approval is not permission: an account can run for months and be read the other way later, and the spend and the pixel history go with it. A shop that builds the channel should assume the account is rented and keep a second platform warm.

Why do agencies not publish a monthly retainer price?

Because the work is scoped per shop, and the number the agency needs depends on the vertical, the geography and how much creative it has to build from nothing. That is a real reason rather than an evasion, but it has a practical cost for the buyer: a retainer cannot be compared before a call, while a per-lead price can be compared in a browser tab. The nearest published cousin is a pay-per-meeting model at a $300 setup fee and $200 to $400 per booked meeting, which is worth using as a reference point because it prices an outcome rather than a month.

Does exclusivity matter more than the buying model?

Usually, yes. Exclusivity is priced across every tracked vertical at 2.0 to 2.5 times the shared version, and that premium is not a markup so much as a description of what your closer walks into. A shared record puts three or four rooms on the same merchant within the hour, so the conversation starts with the merchant already tired of the subject. An exclusive record starts cold and stays yours. That difference survives every buying model on this page: an agency that generates shared applications and a vendor that sells shared records hand your room the same problem at different prices.

What is the real risk of building the channel in-house?

Losing the account, not overspending. The spend is visible and can be capped in an afternoon; the policy exposure cannot. Both of the published platform policies that apply here ban short-term lending ads without naming merchant cash advance, which leaves every account in the category running on an unwritten interpretation. When an interpretation changes, the account, the audiences and the pixel history go together, and the salary that built them is already spent. That is why shops that do build in-house treat it as one channel among several rather than as a replacement for bought inventory.

Whose consent covers the call when leads are bought?

Not the vendor's, and this is the item most agreements leave vague. A marketplace's published methodology states it plainly: all purchased lead inventory is brokered consumer data, and the original consumer's consent does not transfer to the buyer along with the data. That applies to a bought file, to an agency-generated file and to a filing or trigger record equally. The question to ask before any of the three models starts is whether the consent record names your shop, and what the vendor retains as evidence of it. A cheap file is only cheap until the first complaint.

How should a shop test two buying models against each other?

Run them at the same time, on the same closers, and compare on funded deals rather than on connect rates. Splitting the room is the only way to hold the sales process still, and the sales process is the variable that ruins every cross-shop comparison. Give each model enough volume to fund at least a few deals, because a comparison at two deals apiece measures luck. Then divide total channel spend by funded deals for each, and note that no provider can guarantee that number: conversion depends on the offer and the sales process, which is exactly why the test has to happen inside your own room.

Pay Per Lead, Agency Retainer or In-House: How Should an MCA Shop Buy Applications in 2026?

What are the three ways an MCA shop buys applications?

Buy leads when you want the cost fixed per record: public 2026 tables post $10 to $50 shared and $50 to $120 exclusive. Pay a retainer when you want a channel built and can carry a dry month. Run ads in-house only when policy risk and salary are affordable losses.

Buy the record, rent the team that finds it, or build the team yourself. Pay per lead prices one merchant at a time, and the public 2026 tables run from $10 to $50 for a shared record up to $50 to $120 for an exclusive one, with live transfers at $80 to $200 and up. A retainer prices a month of somebody's effort. In-house prices a media budget, a salary and a platform account that somebody else can switch off.

The three are usually compared on price per application, which is the wrong axis. They are three different contracts, and what actually changes between them is who absorbs the month where the ads run, the spend clears, and nothing worth dialling arrives. That is the question underneath the price, and it is the one this page answers first.

What does pay per lead actually cost in 2026?

It depends entirely on which rung of the ladder the quote sits on, and the published ladders span four orders of magnitude. One 2026 wholesaler ladder, dated 7 May 2026, prices business data at $0.02 to $0.20, aged records at $0.05 to $0.50, filing and trigger files at $0.20 to $0.75, applications and submissions at $20 to $100 and up, and real-time leads at $50 to $200 and up. A separate vendor table dated 25 August 2026 prices shared leads at $10 to $50 and exclusive leads at $50 to $120.

Why do the published bands overlap so heavily?

Because the sources price three variables separately: age, exclusivity and how much qualification was done. A marketplace index puts an exclusive live transfer at $35 low, $50 median and $75 high, while a shared internet form on the same index runs $0.50, $2 and $5. A $50 invoice line can therefore describe either a shared real-time record or an exclusive one, and nothing on the invoice tells you which. The agreement does.

What each buying model costs, what it guarantees, and who carries a bad month, from public 2026 sources read on 4 September 2026.
Buying model	What you pay	What the payment guarantees	Who carries a bad month
Pay per lead, shared	$10 to $50 a lead on one vendor table	A record arrives. Several rooms may hold it	You. The record is spent whether or not it answers
Pay per lead, exclusive	$50 to $120 a lead on the same table	A record arrives, held by one buyer	Split, if a credit term is written down. Otherwise you
Pay per booked meeting	$300 setup and $200 to $400 a meeting on one rate card	A calendar slot with a merchant on it	The vendor, up to the point of the meeting. You after it
Agency retainer	Not publicly posted by the agencies surveyed	Hours and a media plan, not an outcome	You. The retainer is owed whether or not the ads work
In-house media team	Ad spend plus salary, and a policy risk with no price	Nothing. You own the account and the account can be read against you	You, twice: the spend and the pixel history
Pay per lead, here	$60 flat at every volume, no setup fee, no ad budget	An exclusive live lead to one buyer only, never resold, credit back with no time limit	Split by the credit rule. Conversion stays your room's number

What does an agency retainer buy that a per-lead price does not?

Control and continuity. A retainer buys a team that builds creative, holds the ad accounts, tests offers and keeps working when the numbers dip, which is exactly the work a per-lead vendor keeps for itself. It also buys the asset: the audiences, the creative library and the conversion history stay with the shop that paid for them, rather than living inside a vendor's account.

The trade is that the invoice is owed whether or not the phone rings. None of the agencies surveyed for this page publish a monthly figure at all, which is a finding rather than an omission: it means the price is negotiated per shop and cannot be compared before a call. The nearest published cousin is a pay-per-meeting model at a $300 setup fee and $200 to $400 per booked meeting, which prices an outcome rather than a month and sits between the two models.

Can an MCA shop run its own ads in 2026?

Yes, and the risk is not the one most shops plan for. It is policy, not cost. Meta's published prohibited financial products policy bans ads for short-term loans of 90 days or less, payday loans and paycheck advances. Its restricted financial services policy repeats that wording and adds the rules for permitted finance ads: 18-plus targeting and disclosure requirements on the advertiser.

What does the silence about merchant cash advance mean?

Neither page names merchant cash advance in either direction. That silence is where the exposure sits: an account can run for months and then be read the other way by a reviewer, and when it goes, the media spend and the pixel history go with it. Shops that build the channel anyway spread the buy across several platforms rather than one, and treat the account as rented rather than owned.

What does the in-house channel really cost to build?

Three line items, and only one of them is the ad budget. There is the spend, there is at least one salary for somebody who can write, test and read a report, and there is the learning period, during which the spend produces data rather than deals. None of the three is publicly benchmarked for this vertical, so any shop quoting a confident in-house cost per application is quoting its own account and not the market.

What is measurable is the gap the in-house team has to beat. A marketplace index finds real-time leads costing 10 to 50 times more than aged records, which means an in-house team that generates real-time applications is producing the expensive end of the market at cost. That is the whole case for building. It is also why the case only closes at volume: the fixed costs do not care how many deals the room funds this month.

Which model wins on cost per funded deal?

Whichever one your room can actually work, and no published table can answer it for you. Cost per funded deal is total spend on a channel divided by the deals that funded from it, and the denominator lives in your CRM. Every vendor publishes the numerator. None can publish the other half, which is why the honest ones say so: no lead provider can guarantee funded deals, because conversion depends on the offer and the sales process.

Two things move that number more than the price does. The first is dial speed, since a fresh record worked in the afternoon is the expensive end of the market bought and then wasted. The second is exclusivity, priced across every tracked vertical at 2.0 to 2.5 times the shared version, because a shared record puts your closer in a race rather than a conversation. A cheap channel that funds nothing is more expensive than an exclusive lead that funds once a week.

What should be in writing before any of the three starts?

Four things, and they are the same four whichever model is signed. What exactly is delivered, in fields rather than adjectives. How many other buyers receive it. What happens when a delivered record fails, meaning the credit rule, the window and the process for claiming it. And whose consent the outreach rests on.

The unit being billed, written as a definition rather than a label
The exclusivity term, with a buyer count and a duration
The credit or replacement rule, with the window and the claim process
The consent basis, and whether it names your shop
The exit, meaning what is owed the month you stop

The fourth item is the one shops skip. A marketplace's own methodology page puts it bluntly: all purchased lead inventory is brokered consumer data, and the original consumer's consent does not transfer to the buyer along with the data. That sentence applies to a bought file and to an agency's file equally. The fifth item is where a retainer and a per-lead agreement genuinely differ, and it is worth reading before the first invoice rather than after the third.

More on this site: the posted price, how live delivery works, the 3 costs inside a sourced deal, 12 questions to ask before the first order.

Where does PayPerMerchant sit among the three?

PayPerMerchant is the per-lead model with the terms written down: exclusive merchant cash advance leads at $60 a lead, flat at every volume, delivered live to one buyer only and never resold. No setup fee, no minimum term, no ad budget on your side. A lead with a fake number or junk name is credited.

Pricing and terms in full. Get Leads.

Questions funders ask about how to buy applications?

Is pay per lead cheaper than an agency retainer?: It is more predictable, which is not the same thing. Pay per lead fixes the cost of one record: public 2026 tables post $10 to $50 shared, $50 to $120 exclusive, and $80 to $200 and up for a live transfer. A retainer fixes a month of effort at a price none of the agencies surveyed publish. In a good month the retainer can be cheaper per application, because the fee is spread over more of them. In a dry month it is infinitely more expensive per application, because the fee is owed and the applications are not there. Which one is cheaper depends on how many dry months a shop can absorb.
What does a $60 lead have to do to beat a $0.05 aged record?: Fund about one deal for every twelve hundred the aged file funds, and that is the arithmetic every shop should run on its own numbers rather than on a vendor's. The published ladders make the input side easy: aged records at $0.05 to $0.50 on a 2026 wholesaler ladder, exclusive real-time records at $50 to $120 on a vendor table. The output side is the part no table holds. A marketplace index measures real-time inventory at 10 to 50 times the aged price, which is the market's own estimate of the difference, and it is an estimate about willingness to pay rather than about funded deals.
Can a small ISO run Meta ads for merchant cash advance?: It can, under a policy that never names the product. Meta's prohibited financial products policy bans ads for short-term loans of 90 days or less, payday loans and paycheck advances. Its restricted financial services policy repeats that wording and adds an 18-plus targeting rule and disclosure requirements for permitted finance ads. Merchant cash advance appears in neither list. The practical consequence is that approval is not permission: an account can run for months and be read the other way later, and the spend and the pixel history go with it. A shop that builds the channel should assume the account is rented and keep a second platform warm.
Why do agencies not publish a monthly retainer price?: Because the work is scoped per shop, and the number the agency needs depends on the vertical, the geography and how much creative it has to build from nothing. That is a real reason rather than an evasion, but it has a practical cost for the buyer: a retainer cannot be compared before a call, while a per-lead price can be compared in a browser tab. The nearest published cousin is a pay-per-meeting model at a $300 setup fee and $200 to $400 per booked meeting, which is worth using as a reference point because it prices an outcome rather than a month.
Does exclusivity matter more than the buying model?: Usually, yes. Exclusivity is priced across every tracked vertical at 2.0 to 2.5 times the shared version, and that premium is not a markup so much as a description of what your closer walks into. A shared record puts three or four rooms on the same merchant within the hour, so the conversation starts with the merchant already tired of the subject. An exclusive record starts cold and stays yours. That difference survives every buying model on this page: an agency that generates shared applications and a vendor that sells shared records hand your room the same problem at different prices.
What is the real risk of building the channel in-house?: Losing the account, not overspending. The spend is visible and can be capped in an afternoon; the policy exposure cannot. Both of the published platform policies that apply here ban short-term lending ads without naming merchant cash advance, which leaves every account in the category running on an unwritten interpretation. When an interpretation changes, the account, the audiences and the pixel history go together, and the salary that built them is already spent. That is why shops that do build in-house treat it as one channel among several rather than as a replacement for bought inventory.
Whose consent covers the call when leads are bought?: Not the vendor's, and this is the item most agreements leave vague. A marketplace's published methodology states it plainly: all purchased lead inventory is brokered consumer data, and the original consumer's consent does not transfer to the buyer along with the data. That applies to a bought file, to an agency-generated file and to a filing or trigger record equally. The question to ask before any of the three models starts is whether the consent record names your shop, and what the vendor retains as evidence of it. A cheap file is only cheap until the first complaint.
How should a shop test two buying models against each other?: Run them at the same time, on the same closers, and compare on funded deals rather than on connect rates. Splitting the room is the only way to hold the sales process still, and the sales process is the variable that ruins every cross-shop comparison. Give each model enough volume to fund at least a few deals, because a comparison at two deals apiece measures luck. Then divide total channel spend by funded deals for each, and note that no provider can guarantee that number: conversion depends on the offer and the sales process, which is exactly why the test has to happen inside your own room.

Pay per lead, agency retainer or in-house:how should an MCA shop buy applications in 2026?