Google Ads + ChatGPT: 40 Questions That Change How You Manage Paid Media

Managing Google Ads campaigns requires answering dozens of questions every day: which campaign converts best? What keywords are bleeding budget? Where is ROAS declining? The Google Ads interface shows data, but cross-referencing metrics, comparing periods, and identifying opportunities requires exporting CSVs, building spreadsheets, and losing hours on mechanical tasks. Now there is another path: ask these questions directly in Claude or ChatGPT, in English, and receive answers based on your real data in seconds.

Kondado replicates your Google Ads data (campaigns, keywords, search terms, geographic and demographic) to a Via Kondado destination through an automated pipeline. The Kondado MCP connects that Via to Claude and ChatGPT at the endpoint https://mcp.kondado.io/mcp, and you ask in English (for example: "which keywords spend over $500 and do not convert?") to receive analysis of your real data in seconds.

How it works: your data in ChatGPT in 3 steps

The Kondado MCP (Model Context Protocol) is the channel that connects Claude, ChatGPT, and other AI tools to your Via Kondado where your Google Ads data is already replicated. Configuration is simple and requires no technical knowledge:

1. Replicate your Google Ads data: create a pipeline in Kondado connecting your Google Ads account (including MCC) to a Via Kondado destination (BigQuery, PostgreSQL, MySQL, or another supported destination). Kondado automatically replicates campaigns, ad groups, keywords, search terms, geographic, demographic, and device data.
2. Connect your Via to Claude or ChatGPT: install the Claude Desktop or ChatGPT Desktop application, access MCP settings, and add the endpoint https://mcp.kondado.io/mcp. Authentication is automatic via OAuth.
3. Start asking questions: open a conversation in Claude or ChatGPT and ask questions in English about your data. The assistant translates your question into queries on your Via Kondado and returns analysis based on your real numbers.

40 questions that transform your paid media management

Below you will find 40 questions organized by business area. Each one can be asked directly to Claude or ChatGPT via the Kondado MCP, replacing hours of manual spreadsheet work.

Block 1: Performance and ROAS (7 questions)

Media managers need to quickly identify where investment is returning and where it is being wasted. These questions reveal optimization opportunities that the Google Ads interface does not show in a consolidated way:

1. What are the 10 campaigns with the best CPA (cost per conversion) in the last 30 days?
2. Compare ROAS (conversion value divided by cost) by campaign between the last 30 days and the previous 30 - which ones dropped more than 20%?
3. Rank my campaigns by spend versus conversions in the last 14 days and tell me where to concentrate an additional $2,000 of budget.
4. Which campaigns have the largest difference between conversions and all_conversions? This means a lot is coming from cross-device or view-through that is not entering my main attribution.
5. Show the weekly CTR (click-through rate) trend by campaign over the last 8 weeks and highlight those with persistent decline.
6. Compare average CPC this week versus the average of the last 4 weeks by campaign - which ones rose more than 15%?
7. Show conversions and CPA by day of week over the last 90 days to identify the best investment window.

Block 2: Keywords and Search Terms (7 questions)

Keywords are the heart of search campaigns. These questions help identify waste, expansion opportunities, and match type problems:

1. List keywords with spend over $500 in the last 30 days and zero conversions.
2. Which search terms had conversions but do not exist as registered keywords? Rank by conversion value.
3. Which search terms with more than 50 clicks in the last 30 days have zero conversions? These are candidates to be added as negative keywords.
4. List keywords with search_impression_share below 40% and CPA below account average - where should I increase bids?
5. Compare average CPA by match_type (EXACT versus PHRASE versus BROAD) by campaign - where is the degradation greater?
6. Show conversions by number of words in the keyword (short versus long-tail) - where is my best return?
7. List ad groups with more than 50 active keywords and fewer than 5 conversions in the last month.

Block 3: Ads and Creatives (5 questions)

Creative management requires identifying saturated ads, Performance Max campaigns that need attention, and opportunities to relaunch old pieces:

1. List ads with campaign_status equal to PAUSED that had conversion_rate above 5% in the 30 days before pausing - should I relaunch?
2. Show campaigns with campaign_advertising_channel_type equal to DEMAND_GEN that had increasing spend and increasing CPA in the last 14 days.
3. List Performance Max campaigns ordered by spend and CPA in the last 30 days - which ones are above account average?
4. Compare CTR and conversion_rate of ads created in the last 14 days versus older ads in the same campaign.
5. Which ad groups have ads with increasing impressions but decreasing conversion_rate in the last 30 days?

Block 4: Geo, Demographic and Device (5 questions)

Geographic and demographic data allow adjusting investment by region, age, gender, and device. These analyses are essential for campaigns with limited budget:

1. Rank states by conversions_value divided by cost over the last 90 days using the replicated geographic data.
2. Show the percentage of my spend by city - does any single city represent more than 30% of total?
3. Compare CPA by age_range over the last 60 days using the available demographic data.
4. Breakdown of conversions by gender and by campaign - any notable difference?
5. Compare CPA by device (mobile versus desktop versus tablet) by campaign in the last 30 days.

Block 5: Budget and Pacing (4 questions)

Controlling budget is critical to avoid overspending or underutilization. These questions help monitor pacing and identify imbalances:

1. Which campaigns had daily spend above the configured budget in the last 7 days?
2. List campaigns with spend below 70% of budget and search_impression_share below 30%.
3. Project my spend until end of month based on average spend of the last 7 days - will I be above or below monthly budget?
4. For my shared budgets, show the percentage consumed by each linked campaign in the last 14 days.

Block 6: Cross-Source (Google Ads + Meta + GA4 + E-commerce) (6 questions)

This is where the Kondado MCP shines more than any Google Ads report alone: the assistant performs analyses crossing different sources that live in the same Via Kondado.

1. Compare ROAS between Google Ads (campaigns) and Meta Ads (Facebook/Instagram) by week over the last 90 days - which channel gives better return with the same budget?
2. For each Google Ads campaign, compare conversions with sessions/conversions from GA4 for the same campaign (via UTM) - where is the divergence greater?
3. Sum Shopify revenue by campaign_utm versus Google Ads conversion value - what is the gap and in which campaigns?
4. Cross Google Ads search_term with Search Console query - which terms do I rank organically in positions 1-3 and still pay for?
5. For conversions from 7 days ago, which source (Google Ads, Meta, organic) was the first touch in the GA4 journey?
6. Of customers acquired via Google Ads in the last 6 months, how many returned to buy on Shopify or VTEX?

Block 7: Account Portfolio Management for Agencies (4 questions)

Agency managers with multiple MCC accounts need to quickly identify which clients need attention and consolidate reports without building dozens of spreadsheets:

1. Rank my MCC sub-accounts (customer_ids) by ROAS evolution week-over-week over the last 30 days.
2. List sub-accounts with conversion_rate decline greater than 30% week-over-week - which clients need urgent attention?
3. For each MCC sub-account, give me spend, conversions, CPA, and ROAS this week versus last, in a single consolidated summary.
4. For each sub-account, list the 3 search terms with the most conversions that are not yet registered as keywords.

Bonus: 2 additional strategic questions

1. Which campaigns have average impression_share above 80% but CPA below target - opportunities to scale investment?
2. Compare estimated lifetime value of customers acquired via Google Ads versus Meta Ads in the last 6 months, cross-referenced with CRM or e-commerce data.

What changes in your media management routine

Replacing navigation between dozens of Google Ads screens with a conversation in natural language saves hours every week. Here is how to incorporate this practice into your operation:

Monday morning - quick diagnostic ritual Start the week by asking Claude: "Which campaigns had ROAS decline greater than 20% last week?" and "Which search terms converted and are not yet keywords?" In 2 minutes you have the week's optimization playbook, without opening a spreadsheet.

Bi-weekly client meeting - data on demand During a call, the client asks: "How is my Performance Max compared to last month?" Instead of "I will check and send you," you query the assistant live and respond with real numbers.

Budget increase decision - argument with data Need to justify more investment? Ask the assistant: "Compare the ROAS of my top 20% of campaigns with the bottom 80% and show how much I would gain if I reallocated budget." Decisions based on analysis, not intuition.

Weekly report for multiple clients - automation For agencies, the assistant generates consolidated summaries of multiple MCC accounts in a single text block, ready to copy and paste into the client report or team Slack.

Who benefits most from this capability

SMB media manager operating alone: you do not have a BI analyst building reports, but need to make quick data-driven decisions. The MCP delivers enterprise-level analysis without requiring technical knowledge.

Traffic agency with portfolio of 10+ accounts: drastically reduce weekly report preparation time and identify problems in clients before they notice.

E-commerce internal marketer: cross Google Ads with Shopify, GA4, and Meta Ads in a single question - something that would require complex reports or manual export between multiple tools.

Marketing director at SMBs with limited budget: identify precisely where each dollar of investment is returning and where it is being wasted, without relying on generic platform analyses.

Start a conversation with your data

Kondado replicates data from over 100 data sources (Google Ads, Meta Ads, Shopify, VTEX, RD Station, Salesforce, and more) to destinations like BigQuery, PostgreSQL, MySQL, and Via Kondado. With the MCP, this data also answers questions in Claude and ChatGPT.

Start your 14-day free trial and connect your Google Ads today. No credit card required, with 30 pipelines and 10 million records to experiment. Configure your first pipeline in minutes and start asking questions that transform your paid media management.

Frequently Asked Questions

What is Kondado MCP?

Kondado MCP (Model Context Protocol) is the official channel that connects AI assistants like Claude and ChatGPT to your data replicated through Kondado. The endpoint is https://mcp.kondado.io/mcp and allows the assistant to query your data on demand, with secure authentication tied to your Kondado account.

Which Google Ads data does Kondado replicate?

Kondado replicates campaigns, ad groups, keywords, search terms, geographic, demographic and device data. The replication supports MCC (manager) accounts, at the refresh frequency you choose.

Do I need SQL or programming knowledge to use MCP?

No. You ask questions in English or Portuguese in natural language inside Claude or ChatGPT, and the assistant queries the data through Kondado MCP. No SQL or code is required.

Can I connect MCP to data sources beyond Google Ads?

Yes. Kondado replicates data from over 100 data sources (Meta Ads, Shopify, VTEX, RD Station, Salesforce, GA4 and more). Once your data lives in a Via Kondado, MCP answers questions that cross multiple data sources in the same conversation.

How does the free trial work?

The Kondado free trial offers 14 days with 30 active pipelines and up to 10 million records replicated, with no credit card required. You create your account in minutes and set up your first pipeline the same day.

Where is my data stored?

The data replicated by Kondado lives in the Via Kondado you choose: BigQuery, PostgreSQL, MySQL, Redshift, Amazon S3, SQL Server, Google Sheets or Excel. Kondado MCP queries that data on demand when you ask Claude or ChatGPT.