TL;DR
Our Meta ads audit follows a fixed order: tracking and measurement first (because broken data poisons every decision), then unit economics (because great ads can't fix bad math), then creative (volume, variety, and fatigue signals), then account structure (consolidation, learning status, budget logic), and finally the red-flag sweep (the settings and habits that silently kill performance). Most audits die at step one or two. If your tracking is broken or your economics don't work, nothing downstream matters.
Why the order matters
Every agency has an audit checklist. Most of them are alphabetical, which is to say, useless. Ours is ordered by a single principle: check the things that invalidate everything else first.
There's no point analyzing creative performance if the pixel is double-firing. There's no point restructuring campaigns if the business loses money on every sale. An audit that starts with "let's look at your ad sets" is an audit performed by someone who likes the feeling of auditing more than the outcome. We start where the bodies are buried.
1. Tracking and measurement (if this is broken, stop here)
This is where most audits end, because most accounts have something wrong here and everything downstream is suspect until it's fixed.
- Pixel and CAPI health. Is the pixel firing on the right events? Is CAPI set up, and are events deduplicated correctly? Check Events Manager for volume trends and sudden drops. A 40% drop in reported purchases with flat revenue isn't a performance problem, it's a plumbing problem.
- Event configuration. Is the account optimizing toward the right event? We have seen accounts optimizing for InitiateCheckout, PageView, and once, memorably, for a button click that fired on every page load. Verify the conversion event matches an actual business outcome.
- Attribution settings. What window is the account using, and does it match the sales cycle? A 1-day click window on a product with a 10-day consideration cycle will make everything look terrible and cause constant false panics.
- Domain verification and consent. Is the domain verified? Did a consent banner update quietly nuke match rates? Post-ATT, this stuff drifts, and nobody notices until the numbers look weird.
- Meta vs reality. Compare Meta-reported revenue against Shopify, Stripe, or whatever the source of truth is. Small discrepancies are normal. Sudden divergence means stop and fix measurement before touching anything else.
If tracking fails, the audit pauses here. There is no step 2 until step 1 is green. Optimizing against broken data is how you scale the wrong ads for a quarter.
2. Unit economics (great ads can't fix bad math)
Before looking at a single ad, we check whether the business can afford to advertise at all.
- Target CPA or ROAS. Does one exist? Is it written down? Is it based on actual margins or vibes? An alarming number of accounts are optimizing toward targets nobody calculated.
- The margin math. Take the average order value, subtract cost of goods, shipping, fees, and overhead. What's left is what you can spend to acquire the customer. If the target CPA is higher than that number, no amount of media buying fixes it. This is a business problem wearing an ads costume.
- Repeat purchase behavior. A $50 CPA is suicide for a one-time $60 purchase and perfectly fine for a $60 purchase with a 40% repeat rate. If the business has retention data, the allowable CPA changes completely. Most audits ignore this and judge everything on first-order economics.
- Break-even ROAS. Simple formula: 1 divided by gross margin as a decimal. At 50% margin, break-even ROAS is 2.0. If the account target is below break-even, someone set it by copying a competitor's slide deck.
If the economics don't work, we say so. We'd rather lose the audit than take on an account that's mathematically doomed. Agencies that skip this step aren't being nice, they're being cowards.
3. Creative (volume, variety, fatigue)
Now, and only now, do we look at ads. Creative is the biggest performance lever in the account, which is why amateurs start here and professionals check the foundation first.
- Volume. How many active creatives per ad set? How many new concepts launched in the last 30 days? Most underperforming accounts are running 3-5 creatives total and haven't launched anything new in two months. That's not a strategy, that's a museum.
- Variety. Are the creatives actually different from each other, or is it the same ad with 12 background colors? We look for distinct hooks, angles, and formats. Ten variations of one idea is one idea.
- Fatigue signals. Frequency trends, CTR decay, first-time impression ratio. If frequency is climbing past 3 in prospecting while CTR slides, the account isn't underperforming, it's exhausted.
- Hook quality. Watch the first 3 seconds of every video. If the hook doesn't earn the thumbstop, nothing downstream matters. Most creative "testing" we see is actually just polishing ads nobody stops for.
- UGC vs polished. Is there a mix? Overproduced brand creative and raw UGC behave differently, and accounts running only one type are leaving money on the table.
4. Account structure (boring, but it compounds)
Structure doesn't win accounts, but bad structure quietly taxes everything.
- Consolidation. Count the ad sets. Then count how many are in learning limited. If half your ad sets are starving, merge them. Every fragmented budget is a tax on learning.
- Learning status. Anything stuck in learning or learning limited gets flagged. The fix is almost always fewer ad sets with more budget each, not cleverer settings.
- Budget logic. Are budgets set by strategy or by superstition? We look for the classic mistakes: 47 ad sets at $10/day, daily budget changes over 20%, and "testing" budgets so small they can't produce signal.
- Campaign objectives. Is each campaign using the right objective for its job? (More on this in our objectives guide, but the short version: most misused objectives are someone optimizing for traffic when they want sales.)
- Audience overlap. Are multiple ad sets fishing the same pond? With broad targeting as the default, overlap is more common than people think, and it means you're bidding against yourself.
5. The red-flag sweep
The things that don't fit neatly above but kill accounts quietly:
- The account has 200 dead campaigns. Clutter isn't just ugly, it slows down analysis and hides what's actually running. Archive ruthlessly.
- Someone "tests" something new every week. Constant structural changes mean nothing ever exits learning. Innovation is great. Innovation on a weekly schedule is sabotage.
- No naming convention. If you can't tell what an ad set is by its name, your reporting is guesswork. This takes 10 minutes to fix and nobody does it.
- Boosted posts running alongside real campaigns. Every audit finds at least one. Boosted posts: not a strategy.
- Agency reports on ROAS, client cares about profit. If the reporting doesn't connect to business outcomes, the agency is grading its own homework.
What a good audit actually delivers
A good audit doesn't deliver a 90-slide deck. It delivers a ranked list: the three things killing this account, in order, with the expected impact of fixing each. Everything else is appendix material.
Our audits almost always come down to some version of: fix tracking, fix the economics or the offer, refresh creative, consolidate structure. The order varies. The list doesn't. If your last audit gave you 47 "optimizations" with no ranking, you got a checklist, not a diagnosis.
FAQ
How often should I audit my Meta ad account? A light audit monthly (tracking health, fatigue signals, budget logic), a deep audit quarterly (full economics review, creative pipeline, structure). Plus an immediate audit any time performance shifts suddenly and the cause isn't obvious.
Can I audit my own account objectively? Partially. You'll catch the mechanical stuff (tracking, structure, fatigue). The hard part is judging your own creative honestly and questioning your own economics. That's what outside eyes are for.
What's the single most common issue you find in audits? Broken or misconfigured tracking, followed closely by creative volume so low it barely qualifies as testing. The glamorous problems (attribution models, bid strategies) are rarely the actual problem.
How long does a proper audit take? A few hours for a competent analyst on a typical account. Anyone selling you a 15-minute automated audit is selling you a PDF generator, not an audit.
Should I pause everything during an audit? No. Audits are diagnostic, not destructive. The only exception: if tracking is fundamentally broken, pause scaling decisions until measurement is fixed, because every optimization decision made on bad data is suspect.
What's the difference between an audit and account management? An audit is a point-in-time diagnosis. Management is the ongoing work of acting on it. Beware agencies whose "audit" is just a sales document for their retainer. A real audit has findings that are useful even if you never hire them.
Written by the MetaMaxd team. We run Meta ads for brands that are tired of agency theater: no vanity metrics, no "hacks," just accounts engineered to make money. If you want this checklist applied to your account by people who do it daily, that's what our audits are for.
