Ad account audit: 15 points that show where your budget is leaking

An ad account audit is a breakdown of actual spend and settings over the last 30–90 days: which event the system was optimizing for, how many ad sets actually drove leads, how much budget was burned by ad sets with zero results. This isn't "take a look at the creatives." It's reconciling the money with the events, and the events with the CRM.

InSync does this audit for free before starting work. Over 4 years, 150+ projects, and 40+ niches, we've built up a stable list of the points where budget leaks most often — and in 9 out of 10 accounts, the problem isn't the creative, it's the structure and the events. Details on the service are on the Meta ads management page.

The audit needs three things: Analyst-level access to Ads Manager, access to Events Manager, and your CRM numbers for the same period. Without CRM data, the audit only shows cost per lead, not cost per customer — and budget most often leaks right between those two numbers.

The most common mistake is an account fragmented into a dozen ad sets with a $20-a-day budget. Each ad set gets $2–3, none of them gathers enough events, the system never gets to learn, and CPL swings wildly from day to day. The money goes not toward results, but toward an endless search.

The benchmark: one campaign per business goal, 2–4 active ad sets inside it, and a budget per ad set large enough to gather roughly 50 target events a week. If your CPL is $5, that's $250 a week per ad set. If you can't feed that many, don't multiply ad sets — cut their number instead.

We also separately check for duplicate "clone" campaigns, old disabled ad sets with an active budget, and campaigns someone launched "as a test" six months ago and forgot about. In accounts handed over to us, this consistently accounts for 10–20% of monthly spend.

• Number of active campaigns and ad sets versus daily budget
• Budget per ad set: is it enough for ~50 events/week
• Clone campaigns and forgotten tests with an active budget

Meta brings in the people who perform, at the lowest cost, whatever action you're optimizing for. Optimizing for "traffic" gets you clickers, for "content views" gets you viewers, for messages without qualification gets you meaningless conversations. Audience is secondary here: the wrong event burns budget even against a perfect target audience.

The benchmark: optimize for the deepest event the account can gather consistently — at least 50 a week. For services, that's "Leads" or a qualified message; for e-commerce, it's "Purchase." If purchases are too few, temporarily switch to "Add to Cart" and move back to Purchase once volume allows.

This approach delivers a predictable cost per lead: $2.06 in a fitness case in Kazakhstan, $2.80 in a construction B2B case in Ukraine, $1.95 in a medical devices case across Europe. What each metric means is covered in our piece on CPL, CPM, CTR, and ROAS.

A pixel being "installed" doesn't mean it's working. In Events Manager we check three things: whether the conversion event is arriving at all, whether it's being duplicated with the server-side one, and how many user parameters are passed. If the Lead event arrives late or doesn't arrive at all, the system is optimizing blind.

The benchmark: the event is sent through two channels — the browser pixel plus Conversions API — with a matching event_id, so deduplication works. Events Manager shows no red warnings, and event match quality holds at "good" or above thanks to email, phone, fbp, and fbc.

A separate leak point is conversions the account can't see: DMs, calls, offline visits. If 90% of inquiries come in through DMs, as in a beauty salon case with 3,100+ inquiries, and the account only counts forms, you're optimizing for 10% of reality and overpaying for the rest.

Auction overlap is when two of your ad sets compete for the same person and drive up each other's price. It happens when interests overlap, when similar 1% and 3% lookalike audiences run in parallel, or when retargeting isn't excluded from cold campaigns. You pay for this in a rising CPM.

The benchmark: overlap between active ad sets under 20–25%, buyers and already-captured leads excluded from cold traffic, and retargeting living in its own campaign with its own budget and its own creatives. One broad ad set is almost always cheaper than five narrow ones with the same interests.

This also covers audiences built two years ago and never updated, and customer lists uploaded once with no sync. Combined with the wrong optimization event, this makes up half the causes we cover in "why Facebook ads aren't working."

Frequency shows how many times one person has seen your ad. The benchmark for a cold audience is 1.5–2.5 over 7 days. For narrow retargeting, 4–5 is acceptable. Above that, you're paying for repeats to people who've already decided not to engage, and you're racking up ad hides, which then push the price up further.

Creative fatigue shows up in three numbers together: CTR down 30%+ from launch, CPM up, and CPL rising for a third week straight. One of the three is noise. All three means it's time to change the creative, not the budget.

The benchmark: 3–5 active creatives per ad set, with some refreshed every 2–4 weeks. What matters is changing the hook and the first three seconds, not the button color: in most accounts we take over, every ad is really just one idea in five outfits.

The easiest way to spot a leak is to sort ad sets by spend and see how much money went to the ones with zero or one lead. The benchmark is up to 15–20% of budget on such ad sets — that's the price of testing. If the figure is 40–50%, the account isn't being optimized, it's just being watched.

The second part is learning-phase cost. Any budget change of more than 20%, or a change to the optimization event or audience, restarts the learning phase, and the ad set pays for that experience all over again. The benchmark: no more than one or two significant edits a week per ad set, and no "just in case" edits on weekends.

The flip side is an underfed winner: the ad set with the best CPL sits on $5 a day while the weak ones eat up the rest. In a restaurant case, reallocating budget toward the working ad set delivered $7.5K in revenue from a $145 budget in 10 days — a 51.7× ROAS. More numbers are in our case studies.

A cheap lead and the right lead are two different things. We cross-check the account against the CRM: how many leads made it to a conversation, how many made it to a payment. Benchmarks from our projects: 45% qualified out of 4,263 leads in auto, 90% qualified at a $2.06 CPL in fitness, 1,240 bookings out of 3,100+ inquiries at a beauty salon.

Geography: ads showing in regions you don't deliver to or serve; the "people who traveled to this location" option enabled instead of "people who live here"; a whole country selected instead of a city plus radius. The benchmark: geo matches your actual service area, not the country's map outline.

Schedule: round-the-clock ad delivery when your sales team works 9 to 7 means some leads go cold before the first touchpoint. The benchmark: either a schedule matched to your managers' working hours, or an auto-reply and chatbot for overnight leads. What exactly to ask your contractor about these settings is covered in the checklist "how to check your Meta ads specialist."

This is the exact table we use to go through a client's account on Day 1. Go through it yourself: every row where you land in the right-hand column is either overpayment or leads you never got. Three or more red flags mean the problem isn't the budget.

If you'd like us to fill in this table for your account — submit a request for a free audit. We won't hand you a list of complaints — we'll hand you a plan of action: what to switch off today, what to reconfigure this week, and what CPL is realistically achievable in your niche with a budget from $600 a month.