Should You Fire Your Meta Ads Agency? 12 Questions Growth Leaders Should Ask

Score each statement from 0 to 2.

• 0: Rarely true.
• 1: Sometimes true.
• 2: Consistently true.

13–16: The relationship probably needs optimization, not replacement.

8–12: Reset the expectations, scope, and operating cadence in writing.

0–7: You may not have a temporary performance problem. You may have an agency problem.

This scorecard is a decision aid, not an objective industry benchmark. A severe issue with ethics, access, or trust can outweigh the total.

Look at the agency dashboard. Then look at the underlying business.

Is Meta reporting a 4x return while revenue is flat? Is customer acquisition getting more expensive? Is contribution margin improving, or are the ad metrics being polished while the economics weaken?

Meta explains that off-ad actions such as website purchases are attributed to ads when they happen within the selected attribution window. That makes Ads Manager a useful attribution view, not the company’s complete financial record. (Meta Business Help Center)

Platform ROAS is directional. The business is the final score.

A weak call sounds like this: “Here is what happened.”

A useful call sounds like this: “Here is what happened, here is the likely cause, here are the available options, and here is the one we recommend.”

At First Spark, we think about this through a Pyramid of Personal Agency:

• Identify the problem.
• Identify the problem and its cause.
• Find possible solutions.
• Recommend the best solution and explain why.
• Act without raising it first, where authority and risk make that appropriate.

Client teams should usually operate at level four. If your agency consistently stops at level one, it is reporting the work instead of leading it.

If ads miss their target, the ads might be the problem. So might the landing page, product page, offer, product, email flow, or retention experience.

A good agency does not accept blame for work it does not control. It also does not use scope as an excuse to ignore an obvious constraint.

Media cannot fix an offer nobody wants. If the offer is the problem, the agency should say so, even if changing it is outside the engagement.

The question is not, “Will they own everything?” It is, “Will they help us find the real constraint and get it to the right owner?”

Accountability goes both ways. A capable agency should tell you no sometimes.

If tracking was broken, it should refuse to draw confident conclusions from bad data. If you want to pause the highest-spending ad solely because its CPA is worse than several low-spend ads, it should explain why Meta may be allocating volume to that ad and assess performance at the appropriate level before acting.

If you treat platform ROAS as unquestionable truth, it should challenge that. If the offer is weak, it should raise the issue rather than quietly billing for media that cannot solve it.

You are not paying for agreement. You are paying for judgment.

After months of managing your account, the agency should know more than it did on day one.

It should have a point of view on the hooks that earn attention, the motivators that create demand, the proof that reduces doubt, the formats that communicate the idea, and the parts of the funnel that limit scale.

Not every test will win. Failed tests should still sharpen the next decision.

Ask the agency to name its five most important learnings from the last quarter. Then ask how those learnings changed the roadmap. If the answers are vague, the testing process may not be creating organizational knowledge.

Performance will decline sometimes. Creative fatigues. Costs change. Demand moves. Offers weaken. Inventory creates constraints. Tracking breaks.

The real test is what the agency does next.

A level-one response says: “CPA increased last month.”

A level-four response says: “CPA increased because conversion rate fell after the offer changed, while CPM and click-through rate remained stable. We have three options. We recommend testing the previous offer against a revised version before changing campaign structure, and here is how we would run it.”

One identifies the problem. The other helps you solve it.

That difference is often more important than the bad month itself.

Most failed relationships do not start with one catastrophic mistake. They start with a pattern.

Communication slows. New creative stops entering the account. Old campaigns keep running because they worked months ago. Reporting becomes a recap. Performance declines, but nobody can explain why beyond seasonality, the algorithm, or the market.

Then blame moves around the funnel.

The problem is rarely one forgotten setting. It is usually that nobody owns the connections. There is no shared North Star, no clear testing strategy, no useful record of learnings, and no person responsible for turning those learnings into the next decision.

The biggest red flag is not declining performance. It is the inability to explain the cause, present the options, and recommend the next move.

Do not turn frustration into a rushed handoff.

Before giving notice:

• Confirm the company controls its Meta business portfolio, ad account, Page, Instagram account, pixel or dataset, catalog, and other core assets.
• Review who has full or partial access and how permissions will be removed after transition. Meta manages business-portfolio asset permissions through Meta Business Suite. (Meta permissions guidance).
• Export historical reports, creative files, briefs, testing logs, naming conventions, and key learnings.
• Document active campaigns, budgets, attribution settings, exclusions, and automated rules.
• Clarify ownership and usage rights for creative and creator content.
• Review notice periods, final invoices, and offboarding obligations.
• Decide who will monitor the account during the transition.

Do not let a change in partners erase the knowledge your ad spend paid to create.

If you are deciding what type of support should replace the agency, compare a Meta Ads agency vs. freelancer. If cost and scope are the main concern, review the Meta ads management pricing guide.

Do not ask whether your agency had a bad month. Ask whether it is operating like an owner.

Does the team communicate promptly? Connect ad performance to the business? Keep the creative system moving? Diagnose the real constraint? Challenge weak decisions? Preserve learnings? Recommend the next move?

If the answer is mostly yes, fix the process before replacing the partner.

If the answer is consistently no, staying is not patience. It is accepting the coordination tax and hoping the same operating model produces a different result.