How I Audit a Google Ads Account in One Pass Without Forgetting Half the Checks

Google keeps adding things to Google Ads.

Auto-applied recommendations. Asset automation. Smart Bidding tweaks the system makes for you. AI-generated headlines that show up in your ads without you knowing. Audience expansion that quietly broadens your targeting. Performance Max optimizations that turn themselves on after every product update.

Every few weeks there is something new to remember to check. And once you are managing more than a handful of accounts, the mental checklist gets long enough that you start missing things.

I was running roughly thirteen different checks on every account I touched. Some I would remember. Some I would skip because I was in a hurry. The ones I skipped were usually fine, until the one time they were not.

The hard question that forced me to build this:

How do I scan an entire Google Ads account against every surface area Google has quietly expanded into — performance, conversions, keywords, assets, settings, compliance — without relying on my own memory to hold the checklist?

That question is what the account-audit skill exists to answer. It is open-sourced as part of my PPC AI Skills repo, and it is the first thing I run before every monthly optimization, every pre-sales call, and every new account I take over.

Here is what it does, why it works, and why every PPC manager needs some version of it.

The Core Problem: Google’s Surface Area Keeps Growing

A few years ago, auditing a Google Ads account was straightforward. Check spend pacing. Look at top campaigns. Skim search terms. Glance at ad copy. Done in fifteen minutes.

That checklist has not been complete in a long time.

Today the surface area includes asset automation settings inside every PMax campaign, Google AI auto-created assets at the ad level, auto-applied recommendations that turned themselves on without your consent, conversion actions that may have stopped firing weeks ago, Smart Bidding signals you did not configure, and audience expansion settings buried three menus deep. None of these are exotic. They are all standard surface area on a typical account today.

The problem is not that any single check is hard. The problem is that there are now so many things to check that no human running through them by memory catches all of them every time.

The fix is not “be more disciplined.” Discipline does not scale across a hundred accounts. The fix is codifying the audit into a skill that runs the same way every time, in the same order, and produces the same structured report at the end.

That is what account-audit is.

Rule 1: Same Audit, Same Way, Every Time

The skill runs thirteen sections in a fixed order. It does not skip anything. It does not get lazy in section eleven because section one was clean. It does not forget about negative keywords because the search terms looked fine.

Every section produces the same structured output every time:

• Account performance and budget pacing • Campaign and keyword performance • Search term waste • Conversion tracking health • Business-standard conversion checks • Impression share with diagnosis • Asset labels (BEST, GOOD, LOW) • Creative compliance (DKI, disapprovals, AI auto-assets) • PMax detection • Quality Score distribution • Negative keyword coverage

The order is the order. The report format is the format. There is no “I will check that next time” mode.

That sounds boring. The boredom is the feature. A boring, repeatable audit is one I can trust to surface the same problems whether I run it on my best account or my worst one.

Rule 2: Period Comparison, Not Just a Snapshot

A snapshot tells you the state of the account. A comparison tells you the direction.

Every metric in the report — spend, conversions, conversion value, CPA — is shown alongside the previous period with delta arrows and percentage change. Up green for improvements. Down red for declines. Flipped logic for CPA, where down is good.

The reason this matters: an account can look fine in absolute terms while quietly trending the wrong way. A CPA that is “still under target” can be 30% higher than last month and still pass a snapshot review. The comparison forces the trend into the report whether the absolute number triggers an alarm or not.

This is the section I look at first on any account I have not touched in a few weeks. It tells me whether I am reading a healthy account or a deteriorating one before I dig into the specifics.

Rule 3: Severity Coding So I Know What to Fix First

Every section flags issues with green, yellow, or red:

• Green: Healthy. No action needed. • Yellow: Worth reviewing when I have time. • Red: Needs attention now.

The colors are not decorative. They are how I triage a 13-section report in under five minutes.

If everything is green, I close the report and move on. If something is yellow, I add it to the optimization queue for that account. If something is red — a stale conversion, a 70% Rank Lost IS, a campaign with three negative keywords on it — I stop and fix it before I do anything else.

The severity grades come from rules baked into each section. Conversion tracking health uses last-fired date: under 14 days is green, 15–30 days is yellow, 30+ days is red. Impression share uses the standard diagnosis matrix: Budget Lost IS over 30% signals a budget constraint, Rank Lost IS over 60% with low Budget Lost signals a quality issue. Negative keyword coverage uses a per-campaign threshold: under 10 negatives flags red.

You can argue with the thresholds. I picked numbers that work for the kinds of accounts I manage and validated them across dozens of clients. The point is not that my numbers are universal. The point is that the audit applies some threshold consistently so I do not have to make a judgment call on every metric for every account.

Rule 4: It Catches What Google Quietly Does to Your Accounts

This is the section that drove me to build the skill in the first place.

The audit specifically checks for the kinds of changes Google makes to your account without you doing anything:

• Auto-applied recommendations that turned themselves on • Google AI auto-created assets at the ad level — headlines and descriptions Google wrote and inserted into your responsive search ads • Asset automation settings inside Performance Max campaigns • DKI (dynamic keyword insertion) usage that may have been added by an automated tool • Disapproved ads that may have stopped serving without you noticing

These are not mistakes you made. They are surfaces Google has expanded into. The audit does not fix them — that is a downstream decision. It just makes sure they are flagged in the report so you have a chance to make the call instead of finding out three months later.

Every time Google ships a new auto-applied feature, the audit gets one more check. That is the whole reason it is a skill instead of a one-time script.

Rule 5: The Pre-Sales Use Case

This is where the skill earns its keep beyond ongoing optimization.

PPC sales conversations almost always include an account audit. The prospect wants to know what is wrong with their current setup. The agency wants to demonstrate value before the contract is signed. Manually doing a thorough audit on every prospect — especially when most of them will not sign — burns time you do not have.

The skill turns a pre-sales audit into a five-minute job. Run it on the prospect’s account. Generate the HTML report. Use the red items as the proof points in your sales call. Done.

The output is professional enough to share directly. Color-coded summary cards, period comparison charts, severity flags by section, actionable insights. It looks like a deliverable because it is one.

I use the same audit format for both internal monthly reviews and pre-sales calls. The audit does not know the difference. It just runs the thirteen sections, produces the report, and lets me decide what to do with it.

What This Has Actually Surfaced

Things the audit has flagged that I would have missed in a manual review:

• Stale conversion actions that had not fired in three weeks because a tag broke after a site redesign • Missing standard conversions — accounts running for months without a Phone Call conversion configured for a service business that runs on phone leads • Budget Lost IS over 30% that I had been mentally writing off as “the account is fine” without realizing how much demand was leaking • Google AI auto-created assets in RSAs where I had not approved any of the headlines that were running • Campaigns with under 10 negative keywords on accounts that had been running for over a year — under-built negative lists are the most common quality drain I find on inherited accounts • LOW-labeled assets that had been sitting there for months because nobody had looked at the asset labels in a while • Disapproved ads that had stopped serving and dragged the impression share down before anyone noticed

None of these are exotic. They are the texture of normal Google Ads accounts today. The audit just makes sure I see them on a schedule instead of catching them by accident.

Get the Account Audit Skill

mkdir -p .claude/skills/account-audit
curl -o .claude/skills/account-audit/SKILL.md \
  https://raw.githubusercontent.com/fourteenwm/ppc-ai-skills/main/account-audit/SKILL.md

The full repo has thirteen other PPC AI skills I use in production every day — mutation safety, RSA refresh, SQR classification, neg conflict detection, and more. All at github.com/fourteenwm/ppc-ai-skills.

The Bigger Point

Most of the AI-in-PPC conversation right now is about clever optimization. Better bidding. Better creative. Smarter targeting. That is the fun part.

The boring part is the part most agencies never automate: the audit. The same thirteen checks on every account, every month, every pre-sales call, regardless of how the account is doing or how busy you are. It does not sound exciting. It is the foundation of everything else.

If you do not have a structured audit you trust to catch what Google keeps quietly adding to your accounts, the optimization work happening on top of it is built on guesswork. You are optimizing what you can see. The audit is what makes sure you can see all of it.

Codify the audit first. Optimize on top of it. Not the other way around.