Account audits

Updated November 12, 2026

What the audit checks

The audit answers one question in five parts: where is this account losing money right now? The five areas:

• Tracking: are conversion actions and pixels firing, deduplicated, and consistent with what the platforms report?
• Waste: which search terms, ads and placements have spent meaningfully with nothing to show for it, in dollars?
• Structure: do campaigns overlap, fragment data, or cannibalize each other, like PMax eating brand Search?
• Settings: which quiet configuration traps are active: auto-applied recommendations, display expansion, surprise broad match?
• Budget: are winners capped while losers spend freely, and is pacing on plan?

Each finding names the problem, the evidence, and the dollar impact where it can be computed. Findings that can be fixed with an account change become proposals; findings that need you, like a landing page or store-side tracking fix, are flagged with specifics.

Tracking checks, in detail

Tracking is checked first because every other number depends on it. An account optimizing toward wrong conversion data is wrong everywhere, confidently.

On Google, the audit verifies that conversion actions are receiving conversions, that primary versus secondary designations make sense (bidding optimizes toward primary actions only), that no two actions double-count the same event, and that counting settings fit the action, 'one per click' for leads, 'every' for purchases.

On Meta, it verifies pixel and Conversions API events arrive, deduplicate through matching event IDs, and that ad sets optimize toward the event that matches the business goal rather than a cheaper proxy. Cross-platform, it sanity-checks the platforms against each other: when one platform claims wildly more conversions than the business could have had, the configuration, usually double counting or attribution settings, explains it. Details of the Meta checks live on the Meta connection page, and the store-side view on AdsGuru and Shopify.

Wasted spend checks

Waste is defined narrowly so the number is defensible: spend on things with enough history to judge and nothing to show. The main buckets:

Search terms with meaningful spend, zero conversions, and wrong or hopeless intent, the raw material of the search terms cleanup. Ads and ad sets that have spent well past the account's evidence threshold with results far from target while better siblings existed. Placements and networks that quietly absorb budget, with Display expansion on Search campaigns the classic case. And duplicated effort: two campaigns bidding on the same intent against each other.

Waste findings come with dollar figures over the audit window, which makes prioritization mechanical: fix the biggest number first. These are also the findings that convert fastest into proposals, since pausing a proven loser or adding negatives is low-risk surgery.

Waste is also reported with its trend: a junk query family that cost a little last month and double that this month is a growing leak, and growth changes the priority even when the absolute number is still small. Static snapshots miss exactly those.

Structure checks

Structure is audited for one underlying problem in several costumes: data spread too thin to learn from, or campaigns competing against each other. The specific checks:

Fragmentation: many tiny ad sets or campaigns, each starved below the volume that lets bidding algorithms calibrate, where consolidation would pool the signal. Overlap: audiences or keywords duplicated across campaigns, so you bid against yourself. Cannibalization: Performance Max serving on brand queries that brand Search would have caught cheaper, booking credit for demand that already existed. Orphans: ad groups with no active ads, campaigns optimizing toward the wrong conversion action, experiments nobody ended.

Structural findings are flagged with evidence, and consolidation moves are always proposed with the trade-offs stated, never auto-applied; see account structure for how those decisions get made.

Settings traps

Settings traps are defaults and auto-applied changes that alter account behavior without anyone deciding. The audit watches the classics:

Each active trap is reported with its current state and what changing it would do. Reverting a trap is a normal proposal: reason, evidence, approval, log.

Budget allocation checks

The budget pass asks whether money sits where results are. Typical findings: a winner limited by budget (on Google, visibly losing impression share to budget) while a loser spends unconstrained; pacing projecting an over- or under-spend against the monthly plan; and spend concentrated in a campaign whose marginal results have decayed.

These findings feed budget automation: pacing corrections and stepwise reallocations arrive as proposals with caps. The audit's job is the diagnosis; the routine's job is the ongoing correction.

How findings are prioritized

Findings are ranked by a simple rule: trust the data first, stop the bleeding second, improve the engine third.

Run your own audit

A manual version of the first pass, doable in an afternoon:

1. Tracking: in Google Ads, open Goals, Conversions; confirm each primary action is real, recent and counted once. On Meta, open Events Manager; confirm purchase or lead events arrive from pixel and CAPI and show as deduplicated.
2. Waste: pull 90 days of search terms sorted by cost; list queries with meaningful spend and zero conversions. Pull ads by spend; list those far from target with no recent wins.
3. Settings: check auto-applied recommendations (Google: Recommendations, Auto-apply), Display expansion and search partners per Search campaign, and match types on your highest-spend keywords.
4. Budget: compare month-to-date spend to plan; check impression share lost to budget on your best campaigns.
5. Structure: map campaigns to intents on paper; mark overlaps, brand coverage, and anything too small to ever exit learning.
6. Total the dollar findings, fix in priority order, and diarize the re-check.

The honest pitch for automation is simply that this afternoon of work compounds when it happens nightly instead of annually. The free audit answer page covers what a connect-time audit shows during the trial.

Nightly re-audits

After the first audit, the same checks re-run nightly, and only new findings are reported: a fresh settings trap, a new disapproval cluster, a tracking break, a new waste pattern. No news means the nightly note stays short, by design; a quiet audit is the system confirming health, and manufactured findings would train you to ignore real ones.

The first audit lands within the hour of connecting, inside the free trial. One honesty note: audit findings need no baseline, so they arrive at full strength immediately, while the trend-based proposals that follow mature over the first 7 to 14 days as AdsGuru learns your account's normal volatility.

Common questions

Related