Google Ads Account Audit: 54-Point Checklist

Skipping the Google Ads account audit is the most expensive mistake you can make when inheriting an account: instead of checking it, the team starts “optimising” on day one. Two weeks later you discover that half the conversions were double-counted, that a forgotten campaign was quietly burning budget, and that the traffic drop came from a feed that broke a month ago rather than from anything you did. And you will not be able to prove it, because you never captured a baseline.

What follows is a sequential 54-point checklist across eight blocks. The order matters: access and data first, structure second, creative last. On a mid-sized account it takes three to five hours and saves weeks of arguing about who caused the drop.

When a Google Ads account audit is mandatory

A formal review is required in four situations: you are taking over from a previous agency; performance dropped and the cause is not obvious; nobody has touched the account in six months; you are about to multiply the budget. In all four, the cost of a missed problem exceeds the cost of a few hours of checking.

The governing rule: change nothing until you have worked through the whole list. One exception — a campaign spending money with obviously zero return can be paused immediately.

Block 1. Access and ownership (1–6)

Your access level. Admin, standard or read-only. Without admin you will not see billing or parts of the configuration.
Who else has access. Export the user list for both the account and the manager account. Former agencies holding admin rights are risk number one.
Which entity owns the account. If the account belongs to the agency rather than the business, the client loses all history at separation. Establish this before work starts, not after.
Manager account links. Check which MCC the account sits under and whether any stale links remain — the mechanics are in manager account operations.
Billing setup. Payment method, billing threshold, outstanding balance, account type. A billing suspension looks exactly like “the ads stopped working”.
Login security. Two-factor authentication, no shared passwords, current recovery options. The baseline hygiene is in account access security and passkeys.

Block 2. Conversion tracking (7–18)

This is the block that matters most. An error here invalidates every other conclusion: you will be optimising toward numbers that do not exist.

Inventory the conversion actions. Open the full list and for each one answer: what exactly fires, who created it, when did it last record.
Primary versus secondary. Check which actions feed optimisation. Phone-number clicks and contact-page views set as primary are the classic source of junk leads — see primary versus secondary conversions.
Duplicates. One action configured both through the tag and through an analytics import doubles your numbers. How to catch it: conversion deduplication.
Counting setting. “Every” versus “One”. Lead generation almost always needs “One”; e-commerce needs “Every”.
Conversion window. Does it match your real sales cycle — see conversion windows and lag.
Conversion value. Is real revenue passed, or does every conversion carry a value of 1? Without value, ROAS strategies are theatre.
Enhanced conversions. Enabled, and is hashed data transmitting correctly — see enhanced conversions.
Consent configuration. Is consent mode in place, and is it suppressing data where it does not need to.
Last recorded date per action. An action with nothing for 30+ days is probably broken, not merely quiet.
Reconcile with the back end. Compare CRM leads or orders with Google Ads over the same period. Anything beyond 20% variance needs an explanation — the logic is in Google Ads vs GA4 conversion discrepancies.
Lead quality. Ask the client what share of leads is valid. If 12 of 100 leads qualify, the problem is not CPA.
Tag diagnostics. Check tag status and errors. If conversions “disappeared”, start here — the sequence is in diagnosing conversion tracking.

Block 3. Account structure (19–26)

Campaign inventory. Export every campaign with status, spend and conversions for the last 90 days.
Ghost campaigns. Enabled campaigns with spend and zero conversions over a quarter.
Data density. Calculate 30-day conversions per bid strategy. Below 15 means the structure is over-engineered — see account structure for Smart Bidding.
Brand separation. If brand is blended in, every CPA report in the account is misleading.
Campaign overlap. Check the search terms report: identical queries across campaigns means self-competition.
PMax over Shopping. One catalogue in two campaign types — see PMax and Search cannibalisation.
Ad group count. More than 20 groups in a campaign is almost always SKAG residue.
Proximity to account limits. Check against Google Ads account limits.

Block 4. Bidding and budgets (27–33)

Strategy per campaign. Write them into one table. Manual CPC in 2026 is a question to ask, not an automatic fault.
Campaigns in learning. How many strategies are learning right now and why — see the Smart Bidding learning period.
Target realism. Compare target CPA/ROAS with the 90-day actual. A target twice out of reach simply suppresses delivery.
Budget-limited campaigns. “Limited by budget” on a profitable campaign is money left on the table.
Pacing and spend spikes. Review daily spend patterns — see budget pacing.
Shared budgets and portfolio strategies. Do they exist, and do they group campaigns with conflicting goals — see portfolio strategies and shared budgets.
Bid adjustments. Legacy device and schedule adjustments are often dead weight under automated bidding — see device bid adjustments.

Block 5. Targeting, keywords and audiences (34–39)

Negative keywords. Do account-level lists exist, and when were they last updated — see account-level exclusions.
Negative conflicts. A negative blocking an active keyword is a common and invisible source of lost traffic.
Search terms, last 30 days. What share of spend went to irrelevant queries. Above 15% is a working problem.
Match type balance. Is broad balanced against phrase — see keyword match types and close variants.
Location targeting. Presence versus interest, and whether you are serving in countries you do not sell to.
Audiences. Are remarketing lists attached, populated and still within membership duration.

Block 6. Ads and assets (40–45)

Ads per group. A group with a single ad and no variant is a missed opportunity.
Ad Strength. Widespread “Poor” means a thin headline set, not a cosmetic issue.
Pinning. Excessive pinning kills combinations — see pinning RSA headlines.
Assets and extensions. Are sitelinks, callouts and structured snippets present and current — see ad assets.
Disapprovals. Review disapproved ads and their reasons; each one is lost reach.
Offer freshness. A promotion that ended last quarter sits in live ad copy more often than anyone expects.

Block 7. Landing pages (46–50)

Broken and redirecting URLs. Crawl every final URL and check response codes.
Ad-to-page match. The promise in the headline should be visible without scrolling.
Speed and Core Web Vitals. See landing page speed and Core Web Vitals.
Mobile experience. Test on an actual phone, not only in an emulator.
Forms and checkout. Excess fields and an unclear checkout — see checkout optimisation for paid traffic.

Block 8. Policy, risk and history (51–54)

Account status and restrictions. Warnings, limited serving, unresolved violations — see Policy Manager and appeals.
Advertiser verification. Completed and not expiring; in several categories serving is restricted without it.
Change history, 90 days. Overlay major edits against spend and conversion charts — see change history and drop diagnosis.
Automated rules and scripts. Check what runs in the background: a forgotten “pause when CPA exceeds X” rule explains half of all unexplained pauses.

Five mistakes that ruin a Google Ads account audit

A bad audit is worse than no audit: it creates the false comfort that the account has been checked. Five reliable ways to ruin your own work:

Starting with the ads. Rewriting copy is quick and satisfying, but if conversions are counted wrong, the new ads will optimise toward the same false signal. Creative is always the last block.
Treating the interface as truth. The numbers in the account are an output of the very settings you are auditing. Every performance conclusion should be reconciled against the client’s back end at least once.
Judging on too short a window. Thirty days hides seasonality and misses delayed conversions. Ninety days is the minimum, and six months for long sales cycles.
Benchmarking against “best practice” instead of the account’s own data. Broad match is not a mistake by itself. Manual bidding is not a mistake by itself. The mistake is a setting that neither follows from the goals nor is supported by the numbers.
Delivering a hundred findings. A report nobody can execute does not get executed. Three critical items with concrete actions beat a hundred observations.

A note on tone: the audit often lands in front of the person who built the account. “Setting X conflicts with goal Y, here is what I propose instead” moves the work forward; “this is all wrong” reliably turns implementation into a fight.

How to present the result

The audit report should fit on one page and carry four sections:

Critical — anything distorting data or burning budget right now (broken conversions, duplicates, ghost campaigns). Fixed in week one.
Important — structural issues that need a plan (consolidation, goal resets). Fixed through experiments.
Cosmetic — assets, extensions, copy. Background work.
Baseline — a table of spend, conversions, CPA and value for the 90 days before you arrived. Without it you cannot prove impact a month from now.

Record the start date of your engagement in the same document. Any drop before that date is outside your responsibility, and that fact should be written down rather than argued later.

Capturing the baseline: four exports

A baseline is not a dashboard screenshot — it is four specific tables saved before you change anything. Without them, any conversation about results a month later becomes an argument about impressions.

Campaigns, 90 days. Columns: spend, clicks, conversions, conversion value, CPA, ROAS, search impression share. This is your reference table.
Weekly trend, 26 weeks. Spend and conversions by week. This chart reveals both seasonality and the exact week something broke — usually long before you arrived.
Search terms, 30 days. With spend and conversions. This export simultaneously supplies negative keyword material and shows how controlled the account really is.
Conversions by action, 90 days. Broken out per action with last-recorded dates. This is where most measurement failures surface.

Save the exports as dated files, not just as saved views: historical data stays available in the account, but once settings change, the exact same cut will not reproduce. On large accounts it is easier to pull these through the API — the trade-offs are covered in Google Ads API versus scripts.

The 30-minute monthly Google Ads account audit

The full 54-point list is for handovers. A live account only needs a short recurring check across nine items:

last recorded date for every primary conversion action;
variance against CRM for the previous month;
campaigns flagged “limited by budget”;
strategies that re-entered learning this month, and why;
share of spend on irrelevant search terms;
new ad disapprovals and policy warnings;
broken final URLs (a quick crawl of the list);
major entries in change history for the month;
campaigns with spend and zero conversions in 30 days.

Those nine cover nearly every scenario in which an account quietly degrades for months without an obvious signal. The easiest place to anchor the check is your client reporting day — it then never gets skipped and feeds straight into the report instead of becoming extra work.

Three findings that show up almost every time

One — conversions are counted twice. The tag fires on site and the same action is imported from analytics. Real CPA is double the reported figure, and every scaling decision was made on false numbers.

Two — junk sits in primary conversions. Phone-link clicks, contact-page views, scroll depth. The algorithm faithfully optimises toward those actions and delivers people who perform them instead of people who buy.

Three — the account is over-engineered. Twenty campaigns against sixty monthly conversions, all permanently learning. Consolidation lifts performance without extra budget, simply because the algorithm can finally see the data.

An audit is not a hunt for blame — it is the construction of a baseline. Until you know which numbers in the account are lying, every optimisation is a blind one.

After the audit

The post-audit order is stable: fix measurement, then structure, then bidding, and creative last. Reversing it is the single most common reason teams report “we rebuilt everything and nothing changed”.

A sensible rollout looks like this. Week one — the critical block only: repair conversions, remove duplicates, pause ghost campaigns, lift blocking negatives. Leave the structure alone so the effect of fixing measurement is isolated from everything else. Weeks two and three — observation: strategies relearn on the corrected signal and a share of the “problems” resolve themselves. From week four — structural change, and only through experiments rather than direct edits.

Warn the client in advance that once measurement is fixed, the reported numbers will probably get worse. If conversions were double-counted, the count halves and CPA doubles while actual business results have not moved a cent. Have that conversation before the fix and put it in writing — otherwise an honest repair of the analytics reads as a first-month failure by the new agency.

If the audit shows the constraint is not configuration but the account itself — unresolved restrictions, missing verification, a category that needs a different launch path — agency Google Ads accounts from PPC Rebels are one option worth evaluating. For accounts that promote mobile apps, add the event checks from our guide to App campaigns — measurement breaks there more often than on web.

FAQ: Google Ads account audit

How long does a full Google Ads account audit take?

Three to five focused hours for an account with 5–15 campaigns; up to two working days for a large multi-market account. Most of that time goes into conversion tracking, and that is as it should be.

Where do I start if time is short?

The conversion block (7–18) and the campaign inventory. Those two produce roughly 80% of the findings. Everything else can follow later.

Can I make changes during the audit?

Preferably not, with one exception: a campaign obviously spending into nothing can be paused immediately. Everything else waits until the baseline is captured, otherwise you cannot separate your effects from the previous owner’s.

How often should a live account be audited?

Full audit twice a year. A short version — conversions, search terms, budgets, policy — monthly. After any major website change, run an off-cycle measurement audit.

What if the previous agency will not hand over access?

If the business owns the account, the owner can reassign admin rights directly in access settings. If the account is registered to the agency, the history cannot be recovered — you will need a new account and a manual rebuild.

How do I tell a badly configured account from a data-corrupted one?

The tell is that interface numbers diverge from business results by multiples, in both directions in different months. That means optimisation ran on an unreliable signal, and strategies will need to relearn once measurement is fixed.

Should I audit an account that is performing well?

Yes — before scaling. Measurement problems surface precisely when budget grows: what was tolerable at $2,000 a month becomes the same percentage error ten times more expensive at $20,000.

What matters more, structure or tracking?

Tracking. A perfect structure optimising toward the wrong signal produces reliably bad results. A correct signal still works through a mediocre structure.

How do I assess lead quality inside an audit?

Request a 60–90 day CRM export and match it by date and campaign. If attribution does not reach the CRM at all, that is itself a critical finding.

How many ghost campaigns usually turn up?

In an account run for over a year without review, typically two to five, consuming 5–20% of budget between them. It is the fastest finding with measurable impact.

Should the client see the audit report?

Always. The report records the state of the account at handover and protects both sides: you from blame for inherited problems, the client from repeating the situation with the next agency.

Can a Google Ads account audit be automated?

Partly. Inventory, ghost-campaign detection, URL checks and negative conflicts automate well through scripts and API exports. Judging signal quality, relevance and ad-to-page match still needs a human.

Google Ads Account Audit: A 54-Point Checklist to Run Before You Take Over