Paying your Meta agency per acquisition sounds like accountability. I think it is the fastest way to buy a pile of leads your sales team cannot close.

I made the Google Ads pricing comparison in Percentage of Spend vs Pay-Per-Acquisition vs Flat Fee: The $20k Google Ads Test. Meta deserves a second look because the agency often decides what counts as a conversion, then the algorithm chases the cheapest version of it. If you sign a per-acquisition deal for Facebook or Instagram to escape percent-of-spend, the billable event can cost you more than the fee you were trying to avoid.

The per-lead pitch hides who defines a lead

If you just fired a percent-of-spend agency, per-lead pricing feels like justice. Under percent-of-spend, you paid more when they spent more, whether the extra spend worked or not. Agencies commonly bill Meta work at 10 to 20% of spend, alongside flat retainers and hourly rates, so the conflict is easy to see on the invoice when pricing follows spend. Pay per acquisition flips it. No billable acquisition, no fee. Finally, you pay for outcomes instead of activity.

Except that pitch assumes an acquisition is a fixed thing. On Meta, someone defines the event, builds the form or pixel that fires it, and tells the algorithm to find more people who will complete it. In a per-lead deal, that someone may be the agency getting paid each time the event fires. Before you negotiate the price per lead, negotiate what a lead is.

Whoever defines the conversion controls what gets billed

On Google, someone types a problem and clicks. On Meta, someone is scrolling and gets interrupted. That makes the path from ad to form matter. An Instant Form is pre-filled, with no page load, and completion runs 2 to 3x higher than a website form. Intent runs lower because that missing friction also filtered the curious from the serious. I used to tell clients Instant Forms were a volume play. I was wrong about what that volume was worth.

Lead Forms can lift conversion rates up to 50% versus sending traffic offsite, yet lead-to-sale conversion often runs 20 to 30% lower. The extra step of leaving the app filters casual browsers. That does not mean every Instant Form lead is junk. It means a form submission and a sales opportunity are different events, and a contract that treats them as identical makes the difference your problem.

When the agency picks the event and gets paid per event, guess which container looks attractive. If you do not define qualified, the invoice will define it for you.

Meta optimizes for the event, not your sales team’s verdict

Give Meta a soft conversion and broad delivery, and it will do exactly what you asked. It will find people likely to complete that event at the lowest cost. That is not the same instruction as finding people likely to buy. A person who taps a pre-filled button at 10pm and never picks up the phone can still count as a successful form submission.

Say you spend $10k a month and pay $40 per raw lead. The agency switches the ad set from a website quote form to an Instant Form and changes the performance goal from qualified to volume. Reported lead cost drops to $22, and the report now shows more than 400 leads. Sales still has the same 25 real conversations. The cost per real conversation has not improved, even though reported lead cost has nearly halved. You are buying more of an event that did not move the outcome.

The setting matters here: Maximise Leads versus Qualified or Conversion Leads changes who Meta goes looking for. So does the signal you send back after a lead arrives. If the platform sees only the form submit, the agency can present a cheaper number while your team does the work of discovering which submissions were worth anything. Ask which event the campaign optimizes for, not just what the dashboard calls a lead.

Per-lead pricing makes a failed creative test the agency’s loss

Testing is the job on Meta. Say you spend $20k a month and the agency earns $35 per lead. A new angle bombs for ten days, and the cost to generate each lead hits $58. Under a flat fee, that is a test to evaluate. Under pay-per-lead, each lead costs the agency more to produce than it earns. So the account settles on two tired winners, frequency climbs, and nobody launches variant three because variant three costs the agency money before it makes the client any.

Fatigue is the bill for that caution. The same audience sees the same hook for eight weeks, click-through sags, and the CPA you thought you fixed drifts back up. Shops may insist on $1,500 to $3,000 in minimum spend so the algorithm has enough volume to test creatives. A per-lead partner can have that volume and still have a reason not to risk it on the next idea.

A flat fee does not magically produce good creative, either. The difference is that a losing test does not directly cut the manager’s revenue per lead. If the proposal has no written testing cadence, you are buying stagnation regardless of the pricing label.

Scale stops where the agency’s margin stops

Marginal cost rises as you scale. The first $5k finds the easy buyers. The next $10k reaches the maybes. Say your blended CPA on $10k is $30. Push spend to $25k, and the incremental $15k clears at $55 per acquisition. Your blended CPA rises above $40. That increase is worth discussing if those additional buyers still profit.

Under a flat fee, it is a discussion about whether the $55 buyers are worth it. Under pay-per-lead at $35, generating them costs the agency more than it earns on each one. So it has a reason to cap you, even if you have a reason to grow.

I have watched that cap arrive as a sudden concern for quote quality. The same partner who loved volume at $10k starts warning that broader audiences will hurt close rates. Sometimes the warning is honest. Often the math changed before the advice did. The partner cannot say the fee stopped working, so the strategy stops instead. The client’s profitable ceiling and the agency’s profitable ceiling are not necessarily the same.

Ask for a flat fee with a qualified-CPA target

A flat fee alone fixes nothing. A cheap retainer with no target buys neglect instead of gaming. What you want is a fee decoupled from spend, a written CPA target for a qualified event you define, and proof of testing. The agency has no fee incentive to inflate budget and no per-lead reward for junk. For ecommerce, flat retainers cluster at $1,500 to $5,000 a month, and groas runs Meta Ads at $1,499 a month, whatever you spend on ads. The fee is boring. That is the point.

If the fee stays put when spend or lead count moves, the agency does not earn more merely by pushing either number up. That predictability removes the fee incentive to inflate spend. The teeth come from what you staple to it. An agency that accepts your qualified-CPA definition in writing has agreed to be judged on something closer to your business outcome. One that insists on defining qualified itself is asking you to trust its own scoring.

Start with the event. You decide what counts: a booked call that showed up, a CRM stage marked qualified, or a paid purchase. Not simply a submitted Instant Form. Then wire that outcome back so Meta can optimize toward it. Meta’s Conversions API for CRM lets you send downstream stages such as qualified lead or closed deal server-to-server. That closes the gap where the platform sees the form but not what happened next. Advertisers that close the loop see up to 15% lower cost per quality lead and 44% higher lead-to-conversion versus volume optimization. A flat-fee partner should make that plumbing part of the job. A partner paid for raw submissions has little reason to prioritize it.

Put these terms in the proposal, not in a reassuring conversation you cannot find six months later:

  • A flat fee decoupled from spend. No percentage and no lead-volume bonus. Write down the number that stays the same when the budget moves.
  • A CPA target on your qualified event. Name the CRM stage, the measurement window, and who has authority to mark a lead qualified.
  • A testing cadence with a log. Put new hooks on a two-week schedule; record what launched, what lost, and what earned more budget.
  • Ownership of the assets. Keep the ad account, Page, and pixel or dataset in your Business portfolio. Accounts and pixels created inside an agency portfolio stay there when you leave, along with the history you paid to build.
  • Month-to-month terms with an exit trigger. If the qualified CPA misses the written target two months running, you can leave with those assets intact.

That is a deal you can inspect. You know what gets measured, who controls the measurement, what work continues when a test loses, and what you keep if the relationship ends. A low monthly number without those terms is just a cheaper way to avoid the argument.

Per-acquisition works only when the event is hard to fake

I am not saying every pay-per-lead deal is a con. There are three narrow conditions under which I would consider one, and they share a trait: the event you pay for cannot be generated cheaply while the real outcome disappears.

  1. The event is downstream and verified. Pay for a purchase, a funded account, or a CRM stage your team marks qualified after a real conversation. Do not call a form fill an acquisition merely because a contract says so.
  2. The contract handles bad leads. Put refunds or replacements in writing for duplicates, bad numbers, and out-of-area contacts identified within seven days. Otherwise, you pay to clean the list yourself.
  3. The offer economics stay stable. One service, one geography, one price: $45 per qualified lead should mean the same thing in March as it does in July. If the underlying offer keeps changing, the agreed price tells you less.

If you cannot get all three, do not sign per-lead. A home-services client I worked with years back paid $38 per call and loved the first month. Then calls drifted toward renters outside the service area and price shoppers who never booked. Call count held. Booked jobs fell. We paid for the phone ringing, not for work sold.

That is what a loose acquisition definition does. The agency can keep hitting the number on its invoice while your team absorbs everything that happens after the phone rings. A verified downstream event makes that harder. It also makes the contract harder to agree on, which tells you something useful before you sign.

Five questions to take into the first agency call

Copy these into the call. You are not looking for polished answers. You are listening for who controls the definition, who bears the cost of tests, and who keeps the assets when you leave. A straight agency can answer plainly. A gamer will spend five minutes explaining why the question is complicated.

  1. What counts as a billable acquisition, and who builds the event? If the agency bills on Instant Form submissions it sets up itself, it prices an event it controls. Ask who verifies whether that submission became a qualified lead.
  2. Will you use Instant Forms or my site form, and optimize for volume or qualified leads? Volume finds people likely to complete the easier event. A qualified signal gives Meta a different target. If the agency cannot explain which one it uses, do not accept a cheap lead forecast as an answer.
  3. How many new hooks and variants launch every two weeks, and what happens when a test loses? If testing stops whenever CPA spikes, creative ages and you pay later. A bare retainer without a testing promise can fail the same way retainers fail growth-stage advertisers: the work never gets written down.
  4. If I double spend from $10k to $20k and marginal CPA rises, what changes in your fee and your plan? Per-lead sellers have a reason to cap the account. Percent-of-spend sellers earn more as the budget grows. You want a plan that can still buy the $55 buyer if that buyer still profits.
  5. Who owns the ad account, Page, and pixel, and what do I keep if I cancel next month? If the history you paid to build lives in the agency’s portfolio, the cheap quote was expensive. Get the ownership terms before the first ad runs.

If you came here asking who charges per acquisition instead, you have my answer: almost nobody should be your first call. Ask for a flat fee that does not move with spend, a qualified CPA you define, a testing log you can read, and ownership that lets you walk. groas runs Meta Ads that way: one flat fee, audiences and budgets worked around the clock, new ads tested and winners scaled, all inside accounts you own. That is accountability a lead report cannot manufacture.

Sign a per-lead Meta deal without those protections, and you may get exactly what you ordered: cheaper leads, more of them, and a sales team that stops answering the agency’s reports. You escaped percent-of-spend and bought a cheaper way to waste spend.