LinkedIn
Most small businesses shopping for a Meta ads agency are buying a 2018 job description. They ask about secret interest stacks and clever targeting tricks because those once mattered, and agencies still know how to sell them. But with Advantage+ Audience on by default, audience picks often act as suggestions while broad delivery finds buyers. The work worth paying for is less mysterious: fresh creative, a conversion signal Meta can use, and a budget concentrated enough to learn. The old hiring myths keep attention on everything else.
Myth 1: A percentage-of-spend fee means the agency wins when you win
It sounds fair on a sales call. If you spend more, the agency earns more, so surely it wants you to grow. What the deck calls alignment, I call a meter running while you pay for the fuel.
Agency pricing commonly runs 10% to 20% of spend, or $1,500 to $10,000 a month flat, plus setup. Say you spend $5,000 a month on Meta ads. At 15%, the fee is another $750. Double your spend to $10,000 and the fee doubles to $1,500, even if the extra work amounts to putting more budget behind one winning ad.
Spend is not profit. An agency paid on spend earns more when you scale a mediocre campaign and less when it tells you to pause and fix tracking. That does not mean every agency on this model gives bad advice. It means its fee does not prove its incentives match yours.
Ask two questions before you sign: What is your flat-fee alternative, and what would make you tell me to spend less? Listen for a concrete answer to the second one. If the only plan is to raise the budget whenever leads rise, nobody is minding what those leads are worth.
Myth 2: A good agency can guarantee your cost per lead
A guaranteed CPL sounds reassuring when you are about to hand someone your ad budget. It is also an awfully confident promise from someone who has not watched your offer convert.
WordStream’s 2026 benchmarks across 1,892 US campaigns put median leads CPL at $27.39, median CPC at $1.80, and median conversion rate at 8.54%. The CPL range across the cited verticals runs from $3.16 for restaurants to $76.71 for dentists. Those figures show why a single promised number tells you so little about your business. Your market, landing page, and follow-up speed all affect where you land.
The arithmetic is plain: cost per lead depends on what a click costs and how often a click becomes a lead. An agency cannot set your local auction prices or make a six-field form pleasant to fill out. It can improve the parts it controls, test the page, and tell you what the resulting leads turn into.
Ask for a plausible range and the plan behind it, not a magic number. What has the agency seen for businesses like yours? What would it change on the page? If it commits to a CPL without asking about close rate and ticket value, cheap form fills may be the easiest way to keep its promise. Cheap is not the same as useful.
Myth 3: Your budget is too small to manage properly
Agencies that only want $10,000-a-month advertisers repeat this one a lot. They are half right, which is enough to cost a smaller business money.
Meta’s spending floor is low: about $1 a day for impressions in the US and $5 a day when optimizing for clicks. That amount can spend. It does not follow that it can produce enough results for the system to learn. Using the rough target of 50 results per ad set per week, the daily-budget calculation is cost per result × 50 ÷ 7. At $20 per lead, that is about $143 a day. At $10 per instant-form lead, it is about $71.
I used to tell clients spending $2,000 a month that Meta was not for them. I was wrong about the conclusion and right about the arithmetic. If you cannot fund 50 expensive purchases a week, spreading the money across more ad sets will not rescue you. Consolidate. Then choose an optimization event you can afford to generate often enough, such as a qualified form submission or a booked call if it produces enough volume, rather than insisting the system learn from a rare closed deal.
Small budgets fail when they are spread too thin, not simply because they are small. Ask a prospective provider how many ad sets your budget can support and which event each one will optimize for. One honest ad set beats five starving ones.
Myth 4: More ad sets means more testing
I used to build sprawling structures myself: ten ad sets, one interest each, $20 a day split ten ways so every idea got a fair test. It felt rigorous. It was arithmetic suicide.
Meta needs roughly 50 optimization events per ad set per week to exit learning. Each ad set needs its own events. Spend $100 a day at $20 per lead and you have a budget for roughly 35 leads a week in total, before you divide it among ten ad sets. You did not buy ten meaningful tests. You divided one modest budget into ten accounts of wishful thinking.
For most small budgets, the working shape is boring on purpose:
- Under $3,000 a month: One campaign, with no more than three to five ad sets.
- $3,000 to $10,000 a month: A small testing setup feeding one scaling campaign-budget-optimization campaign, with two or three ad sets at most in that scaling campaign.
Differentiate the tests by creative angle, not by overlapping interests that compete for the same people. And treat those ranges as ceilings, not a challenge to use every slot. Ask to see the account map before you sign. Count the ad sets against the budget and the events each one is expected to receive. If the map needs more 50-event thresholds than your spend can fund, that is fragmentation with a report attached.
Myth 5: Boosting posts is a cheap version of ad management
The Boost button is easy to find. That is much of its appeal. A post gets attention, you put money behind it, and the numbers on the post get bigger. An owner who wants a full Saturday event might be perfectly happy with that.
But boosting is not a substitute for building a lead-generation campaign in Ads Manager. Boost lacks the same options for form-fill and call optimization, website-visitor or customer-list retargeting, and A/B testing. Its default emphasis on engagement makes likes and profile taps look like progress, even when the business needs booked jobs. Tell the system to find clappers and it will find clappers.
Match the tool to the outcome. Boost an already-winning post when visibility is the point. Build in Ads Manager when you need leads, calls, or bookings you can measure. If a provider’s management plan is mostly pressing Boost, ask what, exactly, it will manage after the button is pressed.
Myth 6: Creative is a one-off setup cost
Most proposals bundle three polished ads at launch and treat creative as finished. That made more sense when agencies could lean harder on audience selection. In broad delivery, the ad itself does more of the work of attracting the right person. Three launch ads are not a creative plan; they are three ads.
The volume problem is real. One industry guide estimates that only 5% to 8% of ads become winners and about half fade quickly. Its working cadence is three to five new ads a week below $15,000 a month in spend, rising to 10 to 15 above that. A small local audience may tire of an ad in 14 to 21 days. One January batch is unlikely to carry the account through March.
Ask who makes the next 20 ads, not who made the first three. A useful answer names a weekly cadence, where new angles come from—reviews, calls, lost deals—and what evidence gets an ad killed or scaled. A vague promise to ‘refresh as needed’ means your winners can fatigue while the fee keeps billing. Buy a creative conveyor belt, not a creative unveiling.
Myth 7: A monthly report proves the work was done
A PDF packed with impressions, reach, and CPC can look reassuring. It can also tell you almost nothing about whether the account is getting better. The familiar owner complaint is paying an agency a substantial cut and receiving vanity dashboards that omit cost per lead by ad set. Impressions do not pay rent.
I want the less decorative version. Show me CPL by ad set. Show me which ads were killed, which were scaled, and how much budget moved. Then show me what happened after the form fill. A lead count without lead quality tells only the first half of the story.
Signal work belongs in that account of what changed. Pixel alone can miss conversions because of Safari ITP, iOS protections, ad blockers, and in-app browsers. If the events feeding delivery are incomplete, the agency is steering with part of the map. Pixel plus Conversions API can help send those events, but the two need a shared event_id so Meta deduplicates them instead of counting one conversion twice. Installation is not the same as a clean signal.
Ask for the Events Manager view, not another reach chart. You want to see match quality, deduplication, and how the leads recorded by your business compare with the events Meta receives. Pair that with a record of budget and creative decisions. A report should prove what the agency did and what those decisions produced, not merely prove that ads ran.
Myth 8: You’re paying for clever audience targeting
This is the one I find hardest to kill, partly because I used to sell it myself. In 2018, a good interest stack could be worth the work. You were telling Meta more precisely who should see an ad, and a bad selection could waste the budget. That experience stuck around long after the job changed.
Today, Advantage+ Audience is on by default, and selected interests generally act as suggestions rather than hard boundaries. With enough conversion signal—roughly 50 or more conversions a week is the benchmark often used here—broad delivery has room to find buyers you would not have named in an interest list. Meta has also removed detailed interest and behavior categories and expanded Advantage+ delivery for leads and sales goals. An agency can still show you an intricate audience map. That does not mean the map is doing the work it claims.
So ask what happens if the system treats half those selections as suggestions. Then ask for what actually feeds delivery: distinct creative angles built around different reasons to buy, a conversion event that reflects a useful lead, and enough budget behind that event to learn. Targeting is increasingly the outcome; creative, signal, and budget discipline are the inputs.
The fee should reflect that work, too. Percentage-of-spend pricing rewards an agency when the budget rises, even when rising spend is not the right call. It is part of the old operating model that search automation has challenged. For a comparison outside Meta, groas offers flat-fee search management at $1,499 a month, with continuous execution and a record of changes rather than a cut of spend. Do not mistake that for a claim that every search service runs Meta campaigns. Do take the standard into your Meta agency interview: who makes new creative each week, who proves the conversion signal works, and who will tell you to spend less when the numbers demand it? The 2018 checklist asked who had the cleverest audience. Hire for the job that exists now.