Google Ads — Audit Workflows with Claude
Step-by-step: export data, analyse with Claude, make changes, log everything
Last updated: March 2026
How this works
Each workflow below follows the same loop: download a specific report from Google Ads, paste it into Claude with a ready-made prompt, review the recommendations, make changes, then log what you did. Focus on quick wins — the goal is frequent, efficient optimization, not hours of deep-diving.
Client Changelog
Every audit session should end with a changelog entry. See the Changelog System page for the format and where to store them.
Search Campaign Audit ~20 min/client
Run weekly. This is the highest-impact audit — search terms alone can save significant wasted spend.
Step 1
Export Search Terms Report
In Google Ads: Insights & Reports > Search terms. Set date range to last 7 days. Click the download icon, export as CSV. For a deeper audit (monthly), use last 30 days.
Columns needed: Search term, Campaign, Ad group, Impressions, Clicks, Cost, Conversions, Conv. rate.
Step 2
Analyse Search Terms with Claude
Paste the CSV data into Claude with this prompt:
Claude Prompt — Search Terms
You are auditing a Google Ads Search Terms report for an Adfinity client.
Here is the Search Terms data (last 7 days):
[PASTE CSV DATA HERE]
Please analyse and return:
1. **Negative keyword candidates** — Search terms with spend but zero conversions, or clearly irrelevant to the business. Group by theme where possible.
2. **High-performer keywords** — Terms with conversions at or below target CPA. Flag any that aren't already added as exact match keywords.
3. **Wasted spend** — Total spend on zero-conversion terms. Calculate as % of total spend.
4. **New keyword opportunities** — High-intent terms appearing that we're not explicitly targeting. Suggest match type.
5. **Quick wins summary** — Top 3 actions to take right now, ranked by impact.
Format the output as a clear list I can action immediately. Use the term exactly as shown (don't rephrase) so I can copy it directly into Google Ads.Step 3
Take Action
Based on Claude's output:
- Add negatives — Add irrelevant terms to the appropriate negative list (campaign-level or account-level master). Use phrase match for negatives unless very specific.
- Add new keywords — Add high-intent opportunities as exact match or phrase match to the relevant ad group.
- Adjust bids — Increase bids on high-converting terms, decrease on high-CPC/low-conversion terms.
- Pause underperformers — If an ad group consistently shows irrelevant terms, review the theme and keywords.
Step 4
Export Keyword Performance Report
In Google Ads: Keywords > Search keywords. Set date range to last 14 days. Export as CSV.
Columns needed: Keyword, Match type, Campaign, Ad group, Impressions, Clicks, Cost, Conversions, Conv. rate, CPC, Quality Score.
Step 5
Analyse Keyword Performance with Claude
Claude Prompt — Keyword Performance
You are auditing Google Ads keyword performance for an Adfinity client.
Here is the Keyword Performance data (last 14 days):
[PASTE CSV DATA HERE]
Client target CPA: ฿[INSERT TARGET] per conversion
Monthly budget: ฿[INSERT BUDGET]
Please analyse and return:
1. **Bid adjustment recommendations** — Keywords where CPC is too high relative to conversion rate. Suggest specific bid changes (increase/decrease by %).
2. **Quality Score issues** — Any keywords with QS below 5. For each, note the likely cause (ad relevance, landing page, expected CTR) and suggest a fix.
3. **Paused keyword candidates** — Keywords with significant spend (>5% of budget) but zero or very expensive conversions.
4. **Match type opportunities** — Broad match keywords that should be tightened to phrase/exact based on performance, or phrase match winners worth adding as exact.
5. **Budget allocation** — Are high-converting keywords limited by budget? Should spend shift between campaigns?
Give me a prioritised action list — top 5 changes ranked by estimated impact.Step 6
Audit Ad Copy
In Google Ads: Ads & assets > Ads. Filter by campaign. Export as CSV.
Claude Prompt — Ad Copy Audit
You are auditing Google Ads RSA (Responsive Search Ad) performance for an Adfinity client.
Here is the Ad Performance data:
[PASTE CSV DATA HERE]
The client's business: [BRIEF DESCRIPTION — e.g. "B2B marketing agency in Bangkok"]
Target audience: [e.g. "Business owners looking for advertising services"]
Please analyse and return:
1. **Winning vs losing ads** — Which ads have the best CTR and conversion rate? Which should be paused?
2. **Headline analysis** — Which pinned headlines are working? Any patterns in high-performing headlines (numbers, questions, specific benefits)?
3. **New ad suggestions** — Based on what's working, suggest 3 new headline variations and 2 new description variations I can add.
4. **Extension check** — Based on the data, are there any obvious extension types missing or underperforming?
Keep suggestions specific to the client's business and audience. I want copy I can paste directly into Google Ads.Step 7
Log Changes
Paste this into Claude after making changes:
Claude Prompt — Generate Changelog
I just completed a Search campaign audit for [CLIENT NAME] on Google Ads.
Here's what I did:
[DESCRIBE CHANGES IN YOUR OWN WORDS — e.g. "Added 12 negative keywords, paused 3 keywords with high spend/no conversions, increased bids on 'marketing agency bangkok' by 15%"]
Previous changelog context:
[PASTE PREVIOUS CHANGELOG OR "First audit"]
Please generate an updated changelog entry in this exact format:
## [TODAY'S DATE] — Search Audit
**Analyst:** [NAME]
**Account:** [CLIENT] — Google Ads
**Campaign(s):** [CAMPAIGNS TOUCHED]
**Date range reviewed:** [e.g. Last 7 days]
### Changes Made
- [Bulleted list of specific changes]
### Key Findings
- [What the data showed — top 2-3 insights]
### Next Session
- [What to check next time — e.g. "Monitor new exact match keywords after 7 days", "Review QS on 'agency' ad group"]
---
Append this to the existing changelog (don't replace it).Performance Max Audit ~15 min/client
Run every 2 weeks. PMax is more of a black box — focus on asset performance, audience signals, and conversion quality.
Step 1
Export Asset Group Performance
In Google Ads: select PMax campaign > Asset groups. Review the asset performance ratings (Best, Good, Low). Screenshot or note down the ratings.
Then go to Insights tab for the PMax campaign — copy the search categories and audience segments shown.
Step 2
Export Conversions by Time & Placement
Go to Insights & Reports > Predefined reports > Basic > Campaign. Filter to PMax campaigns. Add: Campaign, Conversions, Conv. value, Cost, Network (with search partners). Export as CSV.
Also check: Campaigns > [PMax campaign] > Listing groups for product-level data (if e-commerce).
Step 3
Analyse with Claude
Claude Prompt — PMax Audit
You are auditing a Google Ads Performance Max campaign for an Adfinity client.
Campaign context:
- Client: [CLIENT NAME]
- Objective: [e.g. Lead gen / E-commerce]
- Monthly budget: ฿[BUDGET]
- Target ROAS or CPA: [TARGET]
- Running since: [START DATE]
Asset performance ratings:
[LIST ASSETS AND THEIR RATINGS — e.g. "Headline 'Get a Free Quote' = Best, Image 'office-team.jpg' = Low"]
Search categories from Insights tab:
[PASTE THE SEARCH CATEGORIES SHOWN]
Audience segments from Insights tab:
[PASTE THE AUDIENCE SEGMENTS SHOWN]
Campaign performance data:
[PASTE CSV DATA]
Please analyse and return:
1. **Asset cleanup** — Which "Low" rated assets should be replaced? Suggest specific replacement ideas based on what's rated "Best".
2. **Search category review** — Are the search categories relevant to the business? Flag any that look like wasted spend (irrelevant queries PMax is matching to).
3. **Audience signal check** — Are the audience segments making sense? Suggest signal adjustments if Google is finding the wrong people.
4. **Conversion quality** — Based on the data, does it look like PMax is cannibalising branded Search (cheap conversions that would have happened anyway)?
5. **Quick wins** — Top 3 actions to improve this campaign right now.
If there isn't enough data yet (less than 50 conversions), say so and recommend waiting.Step 4
Take Action & Log
- Replace "Low" rated assets with new variations
- Add negative brand keywords to PMax via account-level negatives (if brand cannibalisation suspected)
- Refine audience signals based on who's actually converting
- For e-commerce: exclude low-ROAS product groups from asset groups
Use the changelog prompt from the Search section above to log changes.
Display Campaign Audit ~10 min/client
Run every 2 weeks. Display is typically retargeting only — focus on placement quality and frequency.
Step 1
Export Placement Report
In Google Ads: select Display campaign > Content > Where ads showed > Placements. Export as CSV.
Also export: Audiences > Segments data for the campaign.
Step 2
Analyse with Claude
Claude Prompt — Display Audit
You are auditing a Google Ads Display campaign for an Adfinity client. This campaign is used for [retargeting / brand awareness].
Placement report (where ads showed):
[PASTE CSV DATA]
Audience segment performance:
[PASTE AUDIENCE DATA]
Campaign stats (last 14 days):
- Spend: ฿[AMOUNT]
- Impressions: [NUMBER]
- Clicks: [NUMBER]
- Conversions: [NUMBER]
- Frequency: [IF AVAILABLE]
Please analyse and return:
1. **Placement exclusions** — Flag any placements that look like junk (mobile apps, parked domains, irrelevant sites). List them so I can add as placement exclusions.
2. **Audience performance** — Which retargeting segments are converting? Which have high impressions but zero conversions?
3. **Frequency check** — Is the estimated frequency too high? Recommend a cap if needed.
4. **Creative performance** — If ad-level data is included, which creatives are working and which should be swapped?
5. **Quick wins** — Top 3 actions. Focus on cutting waste first.Step 3
Take Action & Log
- Add junk placements as exclusions (campaign or account level)
- Add the standard mobile app exclusion:
adsenseformobileapps.com - Adjust frequency cap if impressions per user are too high
- Pause low-performing audience segments
YouTube / Video Campaign Audit ~10 min/client
Run every 2 weeks. Focus on view rate, audience relevance, and placement quality.
Step 1
Export Video Performance & Placements
In Google Ads: select Video campaign > Videos for creative performance. Export as CSV.
Also export: Content > Where ads showed > Placements for YouTube channel/video placements.
Step 2
Analyse with Claude
Claude Prompt — YouTube Audit
You are auditing a Google Ads YouTube/Video campaign for an Adfinity client.
Campaign objective: [Awareness / Consideration / Action]
Video performance data:
[PASTE CSV DATA]
Placement data (YouTube channels/videos where ads showed):
[PASTE PLACEMENT DATA]
Please analyse and return:
1. **View rate analysis** — Are view rates healthy? (Benchmark: >15% for in-stream, >50% for bumpers). Flag underperforming videos.
2. **Audience engagement** — Which audiences or demographics have the highest view-through rate and watch time?
3. **Placement quality** — Any YouTube channels or videos that are clearly irrelevant to the target audience? List for exclusion.
4. **Creative insights** — If multiple videos are running, which hook/format is winning? Recommend what type of video to produce next.
5. **Quick wins** — Top 3 actions.Step 3
Take Action & Log
- Exclude irrelevant YouTube channels and kids' content placements
- Adjust audience targeting based on engagement data
- Pause low-view-rate videos, brief new variations of winners
Shopping Campaign Audit ~15 min/client
Run weekly for active e-commerce clients. Focus on product performance, ROAS, and feed quality.
Step 1
Export Product Performance
In Google Ads: Products > Products (or Listing groups). Export as CSV.
Also check: Merchant Center > Diagnostics for feed issues.
Step 2
Analyse with Claude
Claude Prompt — Shopping Audit
You are auditing a Google Ads Shopping campaign for an Adfinity e-commerce client.
Product performance data:
[PASTE CSV DATA — include Product title, Product type/category, Impressions, Clicks, Cost, Conversions, Conv. value, ROAS]
Target ROAS: [e.g. 3:1]
Monthly budget: ฿[BUDGET]
Please analyse and return:
1. **Product winners & losers** — Top 10 products by ROAS and bottom 10. Which products are eating budget with no return?
2. **Category performance** — ROAS by product category. Which categories should get more/less budget?
3. **Bid recommendations** — Products where ROAS is strong but impression share is low (underbidding). Products where ROAS is weak and bids should decrease.
4. **Feed quality flags** — Any products with very low CTR that might indicate poor titles, images, or pricing?
5. **Quick wins** — Top 3 actions to improve ROAS this week.
Give specific product names and recommended bid changes where possible.Step 3
Take Action & Log
- Exclude or reduce bids on consistently negative-ROAS products
- Increase bids on high-ROAS products with low impression share
- Flag feed quality issues to the client (missing images, bad titles, wrong prices)
- Structure product groups by category for cleaner bid management
Audit Frequency Summary
| Campaign Type | Frequency | Time | Priority Focus |
|---|---|---|---|
| Search | Weekly | ~20 min | Search terms, negatives, keyword bids |
| Performance Max | Every 2 weeks | ~15 min | Asset ratings, search categories, audience signals |
| Display | Every 2 weeks | ~10 min | Placement exclusions, frequency, audience segments |
| YouTube / Video | Every 2 weeks | ~10 min | View rates, placements, creative performance |
| Shopping | Weekly | ~15 min | Product ROAS, category performance, feed quality |