I've audited over 300 paid media accounts in the last decade. Every single one had at least three of these leaks. Most had all six. The numbers below aren't theoretical — they're the median I see.
Paid media is the easiest line item in your P&L to waste money on, because the waste hides inside dashboards that look healthy. Your ROAS is "okay." Your CPM is "trending up but it's the platform." Your CAC is "in line with industry."
None of those answers are the real one. The real one is a series of specific leaks — most of which your agency knows about, and none of which they're motivated to fix. Their incentive is your monthly retainer. Yours is your monthly P&L. Those incentives don't align.
Here's where the money is actually going.
The six leaks.
— 01Branded search you'd win anyway.
Open your Google Ads account. Look at your branded keyword spend — clicks from people who typed your literal company name into the search bar.
Most B2B and DTC brands spend somewhere between 5% and 25% of their search budget on branded terms. The rationalization is "we have to defend the SERP from competitors bidding on our name."
Maybe. But more often, you'd have ranked #1 organically anyway. The buyer was already coming to you. You paid Google to be the first link instead of the second.
— 02Lookalikes built off bad data.
Lookalike audiences only work if the seed list is clean. The seed list is almost never clean.
Most brands feed their lookalike audiences off email signups, free trial users, or "all website visitors" — populations that include enormous amounts of noise. The algorithm dutifully finds more people who look like that noise.
The fix is to rebuild lookalikes off your top 10% LTV cohort — actual customers who paid you actual money over an actual period of time. Smaller seed list, better signal, dramatically more efficient acquisition.
I've seen this single change cut blended CAC by 22-35% on accounts where everything else stayed identical. Same creative. Same offer. Same daily spend. Just feeding the machine better data.
— 03Creative fatigue you're not measuring.
Every creative has a half-life. Once an audience has seen an ad more than 4-6 times, performance starts to degrade — frequency rises, engagement falls, CPMs climb to compensate.
Most agencies don't proactively measure this. They notice when ROAS dips, then react with a "creative refresh" — usually two weeks later than they should have, after you've already paid premium rates for fatigued ads.
If your top-performing creative is older than 60 days and your CPM is climbing, you're funding the platform's revenue, not your customer's awareness. Set a creative expiry policy in writing — "no creative runs past 90 days at scale without a refresh test in flight." Most agencies will resist this. That's the tell.
If your top-performing creative is older than 60 days and your CPM is climbing, you're funding the platform's revenue.
— 04Attribution windows that lie.
Default Meta attribution is 7-day click + 1-day view. That window will report a number for "Meta's contribution to revenue" that is — pretty consistently — 30 to 50% higher than reality.
The lie isn't malicious. It's just how attribution windows work. They count any conversion that touched a Meta ad, even when the buyer would have converted from organic, email, or direct anyway.
The fix is to compare two numbers every month:
- Platform-reported ROAS (what Meta tells you it delivered).
- Blended ROAS (total revenue ÷ total marketing spend across all channels).
The gap between those two numbers is the lie. If Meta reports 4.5× and your blended is 1.8×, the truth is somewhere closer to your blended number. Most agencies will keep showing you Meta's number. Insist on the blended one. It's the only one your CFO is going to care about.
— 05Audiences too narrow to scale.
Tight targeting feels safe. It also caps your spend ceiling.
I see this constantly: agencies launch with a tightly defined audience — say, "women 28-45, interested in skincare, household income $100K+, parents." It works at $5K/day. ROAS is 4.2×. Everyone's happy.
Then you try to scale to $20K/day. ROAS collapses. The agency blames "fatigue" or "creative." It's neither — the audience was too small to absorb four times the spend without burning frequency through the roof.
— 06Ad spend on a broken funnel.
This is the most expensive leak and the one no media team wants to admit, because it implicates work that isn't theirs.
If your landing page converts at 1.5%, doubling your traffic doubles your spend and doubles your conversions. ROAS stays flat. You spent more. You earned more. Nothing got better.
If your landing page converts at 4.0%, doubling your traffic still doubles spend — but conversions more than double, because you've built a funnel that compounds. ROAS climbs as you scale.
Before you spend another dollar on traffic, run the math: if I doubled my traffic tomorrow, would my unit economics get better, or just bigger? If the answer is bigger, you don't have a paid media problem. You have a conversion problem. Fix the funnel before you scale the spend.
The conversation to have with your agency.
Run the audit. Then take three questions to your media team:
- What percentage of our search spend is on branded terms, and have you ever run an incremental holdout test on it?
- What's our blended ROAS this month versus platform-reported, and why is there a gap?
- If we 3×'d our daily budget tomorrow, where would performance break first and why?
Their answers will tell you whether you have an operator running your account or a vendor. Operators have answered these questions before you asked. Vendors will scramble.
The agencies that fight you on this audit are the ones you should fire. The ones that volunteer it before you ask are the ones worth keeping.
Want to run this audit on your account? The full version — plus five more growth frameworks — is in The Persuasian Growth Playbook. Free, ungated, no email wall. Or have us audit your spend ourselves: apply for a strategy call.