Blog

What Does a Paid Social Audit Actually Check? (And What It Doesn't)

A paid social audit is a diagnosis, not a sales pitch. Here's exactly what a real audit examines — spend forensics, account structure, vendor claims, quick wins — and what it leaves out.

Most “free audits” are sales calls with a spreadsheet attached. A real audit is something narrower and more useful: an independent read of your ad account, based on the account’s own data, that tells you what’s working, what’s leaking, and what your current reporting isn’t showing you.

We run these as a fixed-scope diagnostic — the $1K Q4 Growth Audit — and the scope is the same whether the account spends $5K a month or $50K. Here’s what it actually covers, drawn from a recent audit of a local service brand with a six-figure annual Meta budget. The brand and its agency aren’t named; the findings are real, and they’re typical.

1. The money: where the spend actually went

What this means: Before opinions, arithmetic. An audit starts by re-computing the numbers from raw account exports — total spend, attributed revenue, blended ROAS — instead of trusting the summary slides in a monthly report. It also separates the ROAS your agency reports (ad spend only) from the ROAS you actually experience (ad spend plus management fees). In the recent audit, the reported blended ROAS was about 3.3x. Count the monthly management fee and the real figure was closer to 2x. Both numbers are “true.” Only one of them is the one you’re banking.

Why it matters: Every budget decision downstream — scale, hold, cut — is made off these numbers. If the headline metric silently excludes a third of your true cost, you’re steering with a miscalibrated compass.

What to do: Ask whoever reports to you to show ROAS both ways, and to reconcile their report against a raw Ads Manager export at least once. If the two don’t match, the export wins.

2. The structure: what’s running, and what’s gone stale

What this means: Accounts decay quietly. In the audit above, one ad set — a single broad campaign — carried more than half of total spend, and it hadn’t had a significant edit in months. Account-wide frequency sat above 6, meaning the average person had seen the ads six-plus times. Five campaigns with ROAS below 1x had burned through five figures before anyone paused them.

Why it matters: None of this shows up as a red alert. The account still “works” — revenue comes in, reports go out. But a hero ad set left untouched for a season is creative fatigue compounding in slow motion, and dead campaigns left running are a tax you pay for nobody checking.

What to do: Three numbers tell you most of it: the share of spend in your top ad set, its last significant edit date, and account frequency. All three are visible in Ads Manager in under five minutes. If nobody on your team can tell you them offhand, that’s the finding.

3. The claims: what you were told vs. what the account shows

What this means: Agencies market themselves with numbers — case study ROAS, growth percentages, “proprietary frameworks.” An audit checks the verifiable ones against your account. In this case, the eye-catching multiplier on the agency’s own website didn’t appear anywhere in the account data, and the “exclusive structure” in their reporting turned out to be a standard broad-targeting CBO setup — a reasonable choice, competently labeled, but not proprietary. Their monthly update emphasized organic reach and engagement; the same month produced a handful of link clicks on tens of thousands of views — a click-through of about 0.02%. Not fabricated. Curated.

Why it matters: The gap between marketing claims and account reality is where trust should be calibrated. You don’t need your agency to be magicians. You need their reports to survive contact with your own data.

What to do: Pick one claim from your agency’s last report and trace it to the account. One. If it holds, your confidence is earned. If it doesn’t, you have a specific, factual conversation to have — which is the entire point.

4. The quick wins: what can be fixed without firing anyone

What this means: A good audit ends with a short list of changes that don’t require a new agency, a new tool, or a bigger budget: pause the proven dead spend, refresh the hero creative before the seasonal peak (this account’s holiday peaks ran at more than double its yearly average — the calendar was the strategy), set a reporting rhythm that includes the honest numbers from section 1, and sort out account ownership so the business — not the vendor — holds admin control of its own ad account, pixel, and data.

Why it matters: Diagnosis that ends in “hire us to fix it” is a pitch. Diagnosis that ends in fixes you can hand to your current team is a deliverable.

What to do: Sequence the quick wins by cost of delay. Seasonal fixes go first; structural ones can wait a month; ownership paperwork should happen while everyone is still friendly.

5. What an audit doesn’t do

Honesty about scope is part of the product. Our audit does not include a creative teardown, an audience analysis, or a landing-page/funnel review — those are separate engagements, and bundling them into a “$1K audit” would make all four shallower. An audit doesn’t manage your account, fix what it finds, or guarantee a ROAS. It tells you, with evidence, where you stand. What you do with that is your call — including handing the report to your current agency and asking them to respond to it.

That’s also why it costs money. A free audit is paid for by the engagement it’s designed to sell you. A paid one is paid for by you, which makes you the client instead of the target. The full scope, deliverables, and current availability are on the audit page.

Key takeaways

  • Recompute the headline numbers from raw exports — and count management fees in your real ROAS.
  • Staleness is silent: check your top ad set’s share of spend, last edit date, and account frequency.
  • Trace one claim from your agency’s report back to the account. One is enough to calibrate trust.
  • The best audits end in quick wins your current team can execute, not a pitch for a retainer.
  • A paid, fixed-scope diagnostic answers to you. A free one answers to whoever’s selling the follow-up.

Wondering what your account would show?

The Q4 Growth Audit is a fixed CA$1,000 diagnostic: a forensic report on your ad accounts, a data appendix, and a 30-minute walkthrough, delivered 3–5 business days after we receive your data exports. The fee is credited toward any follow-on engagement booked within 30 days. We take a maximum of two audits per week, and the report is yours to keep — share it with your current agency if you like. We would.

Want to check it yourself first? The Paid Social Self-Audit is a 90-minute DIY screening toolkit (CAD 69): step-by-step data exports, six checks with red-flag thresholds, break-even math, and the exact email to send your agency. Use the score as a meeting agenda, not a verdict — if it raises questions you can’t answer, that’s what the full audit is for.


FAQ

What do you need from us to start? Standard exports from your ad accounts — Meta, Google, TikTok, whichever you run (campaign, ad set, ad level) — or read-only access. We work from exports, under NDA, and we never need the ability to change anything in your account.

How long does it take? Three to five business days from the day your data arrives complete. If exports come in pieces, the clock starts when the set is complete.

Will this turn into a sales pitch for management services? The report stands on its own, and the quick wins are written so your current team can run them. Follow-on work exists — creative, landing pages, management — and the audit fee is credited if you book it within 30 days. If you don’t, you still have the report.

Who owns the report and the data? You do. The report is addressed to your business only, and your exports are used for your audit and nothing else.

Our agency says audits are a waste of money. Are they? Ask what specifically they’d object to: an independent recompute of numbers they reported. An agency confident in its work has nothing to lose from a second read of the same data — and the good ones know it.