Google Ads Audit Checklist: 12 Monthly Checks and the CPA Fixes Behind Each One

How to use this checklist

Run the twelve checks in order. The first six find waste and structural problems; the last six find efficiency. For each check, the paragraph describes what to look at and the bullets describe the fix that usually follows. Do the audit monthly, with a lighter weekly pass on checks one, four, and six for accounts spending more than a few thousand a month.

With Google Ads connected to Claude, each check is a prompt rather than a report to build. The prompts below are the ones that work; adjust the thresholds to the account size. Ask for the findings ranked by estimated spend impact, and you have the month plan.

1. Search terms: remove irrelevant traffic

Pull the last 30 days of search terms and sort by spend. Filter for terms with zero conversions and spend above your target CPA. These are the clearest negative-keyword candidates and usually the fastest CPA reduction in the account: the spend disappears, the conversions stay, and the average CPA drops within two weeks. Ask Claude: "Show me search terms from the last 30 days with more than $20 spend and zero conversions, sorted by spend, and group them by theme."

• Add terms with spend above target CPA and zero conversions as negatives at the campaign or ad-group level
• Exclude competitor brand terms unless you are bidding on them deliberately
• Exclude informational queries ("how to", "what is", "free", "jobs") from conversion campaigns
• Promote converting search terms that are not yet keywords to exact match in their own ad group

2. Match types: check where the waste is coming from

Break the search-term report down by the match type of the keyword that triggered it. Broad match pulls in the widest range of queries and, without Smart Bidding and a strong negative list, produces most of the waste found in check one. Phrase and exact match give you control. The fix is not to abandon broad match; it is to run it only where the bidding and the negatives can support it.

• Move broad-match keywords that are converting poorly to phrase or exact match
• Keep broad match only in campaigns on Target CPA or Target ROAS with a maintained negative list
• Do not run the same keyword in several match types in one ad group; it competes with itself
• See the keyword strategy guide for how to structure this from the start

3. Quality Score: find low-QS keywords dragging up CPC

Quality Score below 5 on keywords with significant spend is a tax on the whole account: you pay a higher CPC for the same position, which raises CPA before a single conversion-rate problem is considered. Pull keywords with Quality Score of 4 or below and more than $50 of spend in the last 30 days, and check which of the three components — expected CTR, ad relevance, landing page experience — is below average for each.

• Add the keyword to at least two RSA headlines where ad relevance is the weak component
• Split wide ad groups so each RSA can speak to its keywords directly
• Match the landing page headline to the ad where landing page experience is the weak component
• Pause low-QS keywords that are not converting; they raise costs without contributing results
• The Quality Score guide covers the mechanics in depth

4. Budget utilization: where is spend concentrated?

Pull campaign-level spend against budget. Any campaign consistently hitting 100% of budget with a CPA below the account average is leaving conversions on the table; any campaign spending under half its budget has a delivery problem — low bids, narrow targeting, disapproved ads, or a bid strategy that has throttled itself. Most accounts have one of each, and moving money from the second to the first is the cheapest CPA improvement available.

• Raise budgets on capped campaigns whose CPA is below target; lower or pause under-delivering ones
• Check "Limited by budget" status and the recommended budget, but decide on CPA, not on the recommendation
• Reallocate from campaigns whose CPA is above target to those below, rather than adding net new spend

5. Bid strategy health: is Smart Bidding working?

Target CPA and Target ROAS need enough conversions to learn from — roughly 30 to 50 a month per campaign is the common guidance. Below that, the algorithm is guessing, and it tends either to overpay to force volume or to throttle delivery hunting for conversions that do not exist at the target. Check each campaign bid strategy status: "Learning", "Limited", and "Eligible (Limited)" all need attention. Check the target too; a tCPA set far below the recent actual CPA is the most common cause of collapsing impression share.

• Campaigns below the conversion threshold: switch to Maximize Conversions without a target, Maximize Clicks, or manual CPC while volume builds
• Set tCPA targets 15–20% above recent actual CPA, then tighten by no more than 10% every two weeks
• Avoid changing targets and budgets in the same week; the learning phase restarts
• The Smart Bidding guide covers when to enter and when to revert

6. Conversion tracking: is the account optimizing toward the right thing?

Every check after this one depends on the conversion numbers being right. List the active conversion actions, note which are marked Primary, and compare the last 30 days of Google Ads conversions against the actual leads or orders in the CRM or order system. A gap of more than a few percent in either direction, duplicate actions marked Primary (a native tag and a GA4 import for the same event), or a thank-you page reachable by refresh are all signs the CPA you are auditing is not real.

• Keep one Primary conversion action per business outcome; set the rest to Secondary
• Fix duplicate firing before touching bids; Smart Bidding is optimizing toward the inflated number
• Add call conversions for businesses where most leads phone in
• The conversion tracking setup guide walks through verification

7. Ad strength and asset coverage

Responsive Search Ads need enough distinct headlines and descriptions to test — at least 8 to 10 unique headlines and 4 descriptions. Ads rated "Poor" tend to receive less impression share, which pushes the account toward weaker ads and higher CPC. Pull RSAs by ad strength and by asset performance, and look for ad groups with a single ad or with assets rated "Low" that have never been replaced.

• Add headlines that include the top keyword, a benefit, and a call to action; replace "Low" assets
• Run at least two RSAs per ad group so there is something to compare
• Pin sparingly; heavy pinning lowers ad strength and limits testing

8. Landing page conversion rate

This is the lever most audits skip because it lives outside the platform, and it is often the largest. A landing page that converts at 2% versus 4% halves the effective CPA without any change to bids or budget. Pull conversion rate by landing page for the last 30 days (GA4 makes this easy if it is connected), and compare pages receiving the same campaigns. Then load the worst page on a phone.

• Match the page headline to the ad headline and the keyword; a generic homepage is the usual failure
• Check mobile load time; anything over about three seconds is losing clicks you paid for
• Remove navigation that lets visitors leave without converting; keep one clear call to action above the fold
• Reduce form fields to the minimum; fewer fields almost always improves completion

9. Device and geo performance

Pull CPA by device and by top locations. Most accounts have at least one segment where CPA is double the average — mobile with a slow page, a city where the service is not really available, a country that was left in targeting by accident. The fix is a bid adjustment or an exclusion, not a rebuild. Ask Claude: "Compare my CPA by device and by top 20 locations for the last 30 days and flag any segment with CPA more than 50% above the account average and meaningful spend."

• Exclude locations you cannot serve; use Presence targeting rather than Presence or Interest for local businesses
• Apply negative device adjustments where a device converts poorly and the page cannot be fixed quickly
• Check the location report for spend in places you never intended to target

10. Audiences and bid adjustments

Check which audiences are layered on the search campaigns in observation mode and how they convert. Past visitors, Customer Match lists, and in-market audiences usually convert at a higher rate than the campaign average, and a positive bid adjustment on them lowers blended CPA without adding waste. Check the exclusions too: an acquisition campaign that is not excluding existing customers is paying to reach people who would have come back anyway.

• Add remarketing, Customer Match, and in-market audiences in observation mode where they are missing
• Apply positive bid adjustments to audiences with below-average CPA under manual bidding; under Smart Bidding, the signal alone is what matters
• Exclude existing customers from acquisition campaigns

11. Ad schedule and dayparting

Pull performance by hour of day and day of week. Some accounts are flat; many are not — a lead-generation account whose phone is not answered after six is paying for calls nobody takes, and a B2B account often converts far worse on weekends. Under Smart Bidding the algorithm already accounts for time, so the fix here is usually about the business, not the bid: when leads can actually be handled.

• Under manual bidding, reduce bids in hours with consistently poor CPA; under Smart Bidding, leave bids alone and fix the handling
• Schedule call ads to business hours where calls are the conversion
• Do not over-segment; a schedule with twenty adjustments is a sign of chasing noise

12. Change history and auto-applied recommendations

Finish with the change history for the last 30 days. Look for changes nobody on the team remembers making: auto-applied recommendations, keywords added by a broad-match suggestion, budgets raised by an automation. Then review the Recommendations tab with a sceptical eye — some are useful (adding negatives, fixing disapprovals), some quietly raise spend (budget increases, broad-match conversions) and should be declined. If auto-apply is on for anything that changes spend or match types, turn it off.

• Turn off auto-apply for budget, bidding, and keyword recommendations
• Apply recommendations that fix errors; decline those that expand spend without a CPA case
• Compare the change history against the week CPA moved; the cause is often in the log

Turning the audit into a CPA plan

The twelve checks produce a list of findings. Rank them by estimated spend impact, and the order almost always comes out the same: negatives and match types first, because they remove waste without any risk; tracking second, because everything else depends on it; bidding and budget third; landing pages and ads fourth; segments and schedules last. Fix in that order, one or two changes a week, and let each settle before judging the next — negative keywords show results in one to two weeks, bid strategy changes take two to four, landing page changes show as soon as traffic hits the new page.

With MCP Ads connected, the whole list is one prompt: "Run a Google Ads audit for the last 30 days covering search terms, match types, Quality Score, budget utilization, bid strategy status, conversion tracking, ad strength, landing pages, device and geo, audiences, ad schedule, and change history. Rank the findings by estimated wasted spend and give me the fix for each." Claude pulls the live data for each item and returns the ranked list; you decide what to change, and the negative keywords guide covers the first and biggest fix in detail.