Windsor MCP for PPC: Audit Campaigns, Cut Waste & Improve ROAS

PPC teams usually run campaigns on Google Ads, Meta Ads, LinkedIn Ads and TikTok Ads at the same time, and each platform reports in its own interface with its own metric names. Answering a simple question, such as which platform brought conversions at the lowest cost last month, means exporting from each one and matching the columns by hand.

Windsor MCP connects those ad accounts to ChatGPT, Claude and other AI assistants. You ask for all platforms in one view and get one table back, with spend, clicks and impressions under the same names. On Google Ads and Meta Ads the assistant can also make the change you decide on, such as pausing a campaign or moving budget, after you confirm it.

🚀 Try Windsor.ai for free and connect your ad data to AI in minutes: https://onboard.windsor.ai/.

PPC analysis with Windsor MCP and AI: Use cases and prompts

1. Multi-platform campaign performance analysis

The challenge:

When campaigns run on Google Ads, Meta Ads and LinkedIn Ads at the same time, each platform reports its own ROAS, conversions and engagement, counted in its own way. None of them shows the other two, so comparing cost per conversion across all three means building the table yourself.

With the three platforms connected through Windsor MCP, the assistant puts them in one table and ranks them by CPA or ROAS. The same setup answers a question agencies ask us often: how did one campaign do on each platform it ran on?

Prompt: Cross-platform performance digest

Use Windsor.ai to pull campaign performance data for [Account Name] over the last 30 days from the following sources:

From Google Ads include:
- campaign, clicks, impressions, ctr, conversions, spend, cpc, roas

From Meta Ads include:
- campaign, clicks, impressions, ctr, conversions, spend, cpc

From LinkedIn Ads include:
- campaign, clicks, impressions, ctr, conversions, spend, cpc

For each platform calculate:
- Total spend
- Total conversions
- CPA (spend divided by conversions)
- ROAS where available

Format the output as a comparison table with one row per platform, followed by a short summary of which platform is delivering the strongest efficiency.

Prompt: Platform efficiency comparison

Use Windsor.ai to pull Google Ads, Meta Ads, and LinkedIn Ads data for [Account Name] over the last 30 days.

For each platform include:
- Total spend
- Total conversions
- CPA
- ROAS
- CTR

Then answer:
1. Which platform has the lowest CPA?
2. Which platform has the highest ROAS?
3. Which platform has the strongest CTR?
4. If budget were shifted from the lowest performing platform to the highest, what would the projected impact be at current efficiency rates?

Format as a ranked table followed by a budget allocation recommendation.

Prompt: Same campaign on every platform

Use Windsor.ai to pull data for [Account Name] over the last 30 days for every campaign whose name contains [campaign keyword] on Google Ads, Meta Ads and LinkedIn Ads.

For each platform include:
- campaign, spend, clicks, conversions, daily budget

Then show:
1. One row per platform with total spend, conversions and CPA
2. Which platform delivered the campaign at the lowest CPA
3. Whether the budget split matches where the conversions came from

Format as a comparison table followed by a two-sentence recommendation.

What you’ll get:

Spend, conversions, CPA and ROAS for each platform in one table, and a recommendation on where the next part of the budget should go.

2. Budget allocation and waste identification

The challenge:

Wasted spend is spread across many small items: campaigns with no conversions, ad groups with a CPA far above the account average, and placements that spend without results. Each item is small, so it rarely stands out in a platform’s own reports.

Asking the assistant to apply spend and CPA thresholds across Google Ads and Meta Ads gives a weekly list of what to pause or cut, sorted by cost.

Prompt: Wasted spend audit

Use Windsor.ai to pull Google Ads and Meta Ads data for [Account Name] over the last 30 days.

From Google Ads identify campaigns and ad groups where:
- cost is above $300 and conversions = 0
- CPA is more than 2x the account average
- roas is below 1.0

From Meta Ads identify campaigns and ad sets where:
- spend is above $300 and conversions = 0
- CPC is more than 2x the account average
- CTR has dropped more than 30% week over week

For each flagged item include:
- Campaign or ad group name
- Total spend
- Conversions
- CPA or CPC
- Recommended action: pause, reduce budget, or restructure

Sort by spend from highest to lowest.
Suggest an action plan for the next week to optimize results.

Prompt: Budget reallocation recommendation

Use Windsor.ai to pull Google Ads and Meta Ads data for [Account Name] over the last 30 days.

For each platform calculate:
- Total spend per campaign
- Total conversions per campaign
- CPA per campaign
- ROAS per campaign

Then identify:
1. The top 3 campaigns with the lowest CPA and strongest ROAS
2. The bottom 3 campaigns with the highest CPA and weakest ROAS

If 20% of budget were moved from the bottom 3 campaigns to the top 3, calculate the projected  impact on total conversions at current efficiency rates.

Format as a reallocation table with a short strategic recommendation.

What you’ll get:

A list of campaigns and ad groups to pause, cut or restructure, and a reallocation table that shows where the same budget would bring more conversions at current CPA.

3. Keyword performance and search term analysis

The challenge:

Most Google Ads accounts have keywords that spend without converting, and search terms that trigger ads they have nothing to do with. Finding them by hand means sorting thousands of rows across campaigns and ad groups.

The assistant can sort those rows by cost and conversions, flag the irrelevant search terms and propose negative keywords, which it adds to Google Ads after you confirm.

Prompt: Keyword performance audit

Use Windsor.ai to pull Google Ads data for [Account Name] over the last 30 days.

For all active keywords include:
- campaign
- ad_group
- keyword_text
- clicks
- impressions
- ctr
- conversions
- cost
- cpc
- roas

Identify:
1. Top 10 keywords by conversions with the lowest CPA
2. Bottom 10 keywords by conversions with the highest CPA
3. Keywords with more than $200 in cost and zero conversions
4. Keywords with a roas below 1.0

For each flagged keyword recommend one action: scale, pause, reduce bid, or restructure match type.

Format as four separate tables with a short summary underneath each one.

Prompt: Search term waste finder

Use Windsor.ai to pull Google Ads search term data for [Account Name] over the last 30 days.

For all search terms include:
- search_term
- campaign
- clicks
- impressions
- cost
- conversions

Identify:
1. Search terms with more than $100 in spend and zero conversions
2. Search terms that are clearly irrelevant to the business based on the keyword context
3. Search terms with high click volume but a conversion rate below 1%

For each flagged search term recommend:
- Add as exact match negative at campaign level
- Add as broad match negative at account level
- Monitor before adding to negatives

Summarize the total wasted spend across all flagged search terms.

What you’ll get:

A prioritized keyword list: what to scale, what to pause, and which search terms to add as negative keywords.

4. Ad performance analysis (top and underperforming ads)

The challenge:

Most PPC teams know some ads are working better than others, but identifying exactly which ones to scale or cut requires digging through multiple campaigns and ad sets.

Without a clear view, the budget gets spread across both high- and low-performing creatives.

An ad-level prompt ranks every ad by CPA in one list, so the best ads and the ads spending without conversions appear side by side.

Prompt: Creative performance audit

Use Windsor.ai to pull [Meta Ads] data for [Account Name] over the last 30 days at the ad level.

Include:
- campaign, adset_name, ad_name, impressions, clicks, ctr, spend, cpc, frequency, conversions

Then identify:
1. Top 5 ads by conversions with the lowest CPA
2. Bottom 5 ads by conversions with the highest CPA
3. Ads with more than $200 in spend and zero conversions

For each top performing ad, summarize what the creative has in common with other top performers.

Format as a table with a short creative insight summary at the end.

What you’ll get:

A clear view of which creatives to scale, which to pause, and where budget is being wasted.

5. Creative fatigue detection

The challenge:

Even strong creatives stop working over time. Frequency rises, CTR drops, and performance declines, but this often goes unnoticed while campaigns continue spending.

A weekly prompt that checks frequency and CTR at the ad level shows which creatives are losing performance while they still convert.

Prompt: Detect creative fatigue

Use Windsor.ai to pull [Meta Ads] data for [Account Name] over the last 30 days at the ad level.

For all active ads include:
- campaign, adset_name, ad_name, impressions, frequency, ctr, spend

Flag any ad where:
- frequency has risen above 3.5 in the last 7 days
- ctr has dropped more than 25% week over week
- spend is above $150 but conversions are declining

Summarize total spend running on fatigued creatives.

For each flagged ad recommend one of the following:
- Refresh creative
- Pause and replace
- Test a new variation with a suggested change

What you’ll get:

A list of creatives showing early fatigue signals, with clear actions to refresh, pause, or replace them before performance drops further.

6. Audience and conversion analysis

The challenge:

Audience performance is fragmented across platforms. Meta, LinkedIn, and other ad platforms each show their own version of what is working, but none of them show which audiences generate revenue.

At the same time, the most valuable signals live in your CRM: which leads convert, which segments close fastest, and which customers are worth paying more to reach again.

Without the CRM data, teams optimize for clicks and platform conversions, which say nothing about deal value.

With Meta Ads, LinkedIn Ads and HubSpot connected, one prompt compares how each audience performs in the ad platform with the deals it produced in the CRM.

Prompt: Audience performance across platforms

Use Windsor.ai to pull [Meta Ads and LinkedIn Ads] data for [Account Name] over the last 30 days.

From Meta Ads, break down performance by audience segment and include:
- campaign, adset_name, age, gender
- impressions, clicks, ctr, spend, cpc, conversions

From LinkedIn Ads, break down performance by audience segment and include:
- campaign, member_job_title, member_seniority, member_industry, member_company_size
- impressions, clicks, ctr, conversions, cpc

For each platform, identify:
- Top 3 audience segments by conversion rate
- Bottom 3 audience segments by conversion rate
- Segments with high spend but low conversion rate

Finish with a short recommendation on which audiences to scale and which to reduce or exclude.

Prompt: Identify high-value audiences using CRM data

Use Windsor.ai to pull [Meta Ads] and [CRM (HubSpot)] data for [Account Name] over the last 90 days.

From Meta Ads include:
- campaign, adset_name, clicks, spend, conversions

From HubSpot, filter contacts where contact_hs_analytics_source = "PAID_SOCIAL" and include:
- contact_lifecyclestage
- contact_jobtitle
- contact_industry
- deal_amount
- deal_dealstage

Then identify:
- Which audience segments generate the highest value deals
- Which segments produce leads that do not progress beyond MQL
- Which job titles and industries convert fastest from lead to closed deal

Finish with a remarketing recommendation based on the highest-value customer profiles.

What you’ll get:

A clear view of which audiences perform well in-platform and which generate high-value customers, so you can scale the right segments, cut wasted spend, and build remarketing strategies based on real revenue.

7. Automated PPC reporting

The challenge:

Clients and leadership do not want to deeply analyze dashboards. They want clear answers: how much we spent, what we got, and what to do next.

Building that report manually means pulling data from multiple platforms, formatting it, and writing the summary. It takes hours.

The two prompts below build that report from live data, with a comparison against the previous period and a short written summary.

Prompt: Monthly (weekly) PPC performance report

Use Windsor.ai to pull [Google Ads, Meta Ads, and LinkedIn Ads] data for [Account Name] for the last 30 days compared to the previous 30 days.

From each platform, include:
- campaign, clicks, impressions, ctr, conversions, spend/cost, cpc, roas (if available)

For each platform calculate:
- Total spend
- Total conversions
- CPA
- ROAS (if available)
- Month-over-month change for each metric

Structure the report as:
- Executive summary (max 3 sentences)
- Platform comparison table
- Top 3 wins
- Top 2 issues or opportunities
- 3 recommended actions for the next period

Write in plain language for a CMO or CEO.

Prompt: Weekly client update

Use Windsor.ai to pull [Google Ads and Meta Ads data] for [Account Name] for the last 7 days.

Include:
- campaign, clicks, conversions, spend/cost, cpc, roas

Show these numbers in a simple dashboard with one chart per metric.

Add a short summary (under 150 words) covering:
- Total spend
- Total conversions and CPA
- One key win
- One issue
- One recommended action

Use a clear, direct tone for a non-specialist.

What you’ll get:

A clear, ready-to-share report covering performance, key changes, and next steps without manual work or formatting.

8. Campaign anomaly detection and early warning

The challenge:

CPA spikes, campaigns that stop delivering and budgets that spend too fast often go unnoticed until someone opens the platform. With several accounts on several platforms, that check rarely happens every day.

A daily health-check prompt compares each campaign with its previous week, and a pacing prompt projects month-end spend for each account, across Google Ads and Meta Ads at once.

Prompt: Daily account health check

Use Windsor.ai to pull [Google Ads] data for [Account Name] and compare the last 7 days against the previous 7 days.

Include:
- campaign, clicks, conversions, cost, ctr, roas

Flag any campaign where:
- Cost increased more than 30% with no corresponding increase in conversions
- CPA increased more than 25% week over week
- CTR dropped more than 25% week over week
- ROAS dropped below 1.0
- Conversions dropped to zero despite continued spend

For each flagged campaign include:
- Campaign name
- Metric affected
- Percentage change
- Severity level: Critical, High, or Watch
- Recommended action

Format as a prioritized alert list that can be reviewed in under 5 minutes.

Prompt: Budget pacing and overspend alert

Use Windsor.ai to pull [Meta Ads] data for [Account Name] over the last 7 days.

Include:
- campaign, spend, conversions

For each campaign identify:
1. Campaigns that have spent more than 80% of their monthly budget before the 20th of the month
2. Campaigns where daily spend has increased by more than 40% compared to the previous 7-day average
3. Campaigns that paused unexpectedly with remaining budget unspent

For each finding include:
- Campaign name
- Total spend to date
- Projected end of month spend at current rate
- Recommended action: reduce daily budget, pause, or investigate

Format as a budget pacing table with a short summary of total projected overspend.

Prompt: Pacing across many ad accounts

Use Windsor.ai to pull this month's spend and campaign budgets for all Google Ads and Meta Ads accounts of [Client or Account IDs].

For each account include:
- account_name, campaign, spend, daily budget, currency

Calculate:
- Spend to date and projected month-end spend at the current daily rate
- Accounts projected to finish more than 10% over or under their monthly budget

Format as one pacing table sorted by projected overspend, with each account's currency in its own column.

What you’ll get:

A short alert list and a pacing table to check each morning, with the campaigns that need action at the top.

9. Landing page and conversion rate analysis

The challenge:

A campaign can have a strong CTR and still convert poorly if the landing page is slow or does not match the ad.

Most PPC teams optimize inside the ad platforms, but what happens after the click is recorded in GA4. Looking at one without the other hides where paid visitors leave.

With Google Ads and GA4 connected, one prompt matches campaigns to landing pages and shows which pages lose the paid traffic.

Prompt: Landing page conversion audit

Use Windsor.ai to pull Google Ads and GA4 data for [Account Name] over the last 30 days.

From Google Ads include:
- campaign, clicks, conversions, cost, ctr, roas

From GA4 (medium = "cpc") include:
- page_path, sessions, bounce_rate, average_session_duration, conversions

Match campaigns to landing pages and identify:
- Pages with 100+ paid sessions and bounce rate above 60%
- Pages with session duration below 30 seconds and spend above $200
- Pages where conversion rate is 50% below account average

For each page include:
- URL
- Paid sessions
- Bounce rate
- Session duration
- Conversion rate
- Spend
- Recommended action (copy, design, offer, or speed)

Sort by spend and end with an action plan for the next week.

Prompt: Campaign vs landing page efficiency

Use Windsor.ai to pull Google Ads and GA4 data for [Account Name] over the last 30 days.

Calculate for each campaign and landing page:
- Click → session rate
- Session → conversion rate
- Overall conversion rate
- CPA (based on GA4 conversions)

Identify:
- Campaigns with strong CTR but weak conversion
- Landing pages that convert well across campaigns
- Where a 20% conversion lift would reduce CPA the most

What you’ll get:

A full-funnel view showing where paid traffic drops after the click, which pages hurt performance, and where to focus to improve conversion rate and lower CPA.

10. Bidding strategy and bid optimization

The challenge:

Most teams set a bidding strategy and leave it running. But performance changes constantly; what worked last month may now be overpaying for the same results.

Without reviewing bid efficiency, campaigns either miss scale opportunities or spend too much for each conversion.

A monthly bid review compares each campaign’s CPA and ROAS with the account average and flags the campaigns limited by budget.

Prompt: Analyze bid efficiency and optimization opportunities

Use Windsor.ai to pull Google Ads data for [Account Name] over the last 30 days.

Include:
- campaign, bidding_strategy_type, clicks, impressions, cost, conversions, cpc, cpa, roas

Identify:
- Campaigns with high CPA vs account average
- Campaigns with strong ROAS but low spend (potential underbidding)
- Campaigns limited by budget or impression share (if available)
- Campaigns where CPC increased but conversions did not

For each flagged campaign recommend:
- Increase bid or target
- Decrease bid or tighten targeting
- Change bidding strategy (e.g. manual → tCPA, tROAS)

What you’ll get:

A clear view of where bids are hurting performance or limiting scale, so you can adjust strategy and improve efficiency.

11. Scaling opportunities

The challenge:

Most PPC workflows focus on fixing what’s broken. But growth comes from identifying what is already working and scaling it.

High-performing campaigns, audiences, or creatives often stay underfunded simply because no one surfaces them clearly.

A scaling prompt lists campaigns with a below-average CPA and steady conversions, and estimates what a 20% to 30% budget increase would add.

Prompt: Identify scaling opportunities

Use Windsor.ai to pull [Google Ads and Meta Ads] data for [Account Name] over the last 30 days.

Include:
- campaign, clicks, conversions, cost, cpa, roas

Identify:
- Campaigns with CPA below account average and consistent conversions
- Campaigns with strong ROAS and stable CTR
- Campaigns with low spend but strong performance

Then:
- Estimate potential conversion lift if budget increases by 20% to 30%
- Flag campaigns that may saturate (high frequency or declining CTR)

Recommend:
- Which campaigns to scale
- How much budget to increase
- Where to shift budget from

What you’ll get:

A prioritized list of campaigns to scale, with clear budget recommendations based on real performance.

12. Ad copy and creative insights

The challenge:

Teams know which ads perform, but not always why. Without understanding patterns, new creatives are based on guesswork instead of data.

This leads to inconsistent performance and repeated testing of ideas that have already failed.

Comparing the top 10 and bottom 10 ads by CPA in one prompt shows which headlines, offers and formats the best ads share.

Prompt: Extract winning creative patterns

Use Windsor.ai to pull Meta Ads or Google Ads data for [Account Name] over the last 30 days at the ad level.

Include:
- ad name, impressions, clicks, ctr, conversions, cpa

For top 10 performing ads:
- Analyze headlines, messaging themes, offers, and formats

Compare with bottom 10 ads and identify:
- Common elements in high-performing ads
- Common issues in low-performing ads

Then summarize:
- Top 3 messaging angles that drive conversions
- Creative formats that perform best
- Specific ideas for new ad variations

What you’ll get:

The messaging angles and formats your best ads share, and specific ideas for new variations to test.

13. Audience targeting optimization

The challenge:

Audience targeting often expands over time without clear evaluation. Budgets get spread across segments that may not convert or deliver value.

Without structured analysis, teams keep targeting audiences that look active but do not drive results.

An audience breakdown by conversion rate and CPA shows which segments to scale, narrow or exclude, and where audiences overlap across campaigns.

Prompt: Optimize audience targeting

Use Windsor.ai to pull [Meta Ads] data for [Account Name] over the last 30 days.

Break down performance by:
- audience segment, campaign, impressions, clicks, ctr, conversions, cpa

Identify:
- Top-performing audience segments by conversion rate
- Segments with high spend and low conversion
- Overlapping audiences across campaigns

Recommend:
- Which audiences to scale
- Which to exclude or narrow
- Where to test new targeting

What you’ll get:

A clear targeting strategy showing which audiences to focus on and which to remove to improve efficiency.

14. Quality Score analysis

The challenge:

A low Quality Score raises CPC and reduces ad visibility. Many teams notice rising costs but do not trace the issue back to ad relevance, expected CTR, or landing page experience.

Without addressing this, campaigns become more expensive over time.

Pulling Quality Score with its three components for every keyword shows which one to fix first: ad relevance, expected CTR or landing page experience.

Prompt: Analyze Quality Score drivers

Use Windsor.ai to pull Google Ads keyword data for [Account Name] over the last 30 days.

Include:
- keyword, quality_score, expected_ctr, ad_relevance, landing_page_experience, impressions, clicks, cpc, conversions

Identify:
- Keywords with quality_score below 5
- Keywords with high CPC and low Quality Score
- Keywords where ad relevance or landing page experience is below average

For each keyword recommend:
- Improve ad copy relevance
- Adjust keyword targeting
- Optimize landing page

What you’ll get:

A prioritized list of Quality Score issues with clear actions to reduce CPC and improve ad performance.

Pro tips to get more value from Windsor MCP for PPC teams

These tips answer the questions PPC teams ask us most after their first prompts.

Include account IDs in prompts

Windsor supports multiple connected ad accounts, which means a prompt that queries without specifying an account can pull data from the wrong property.

Always include the exact account ID for every platform you are querying to keep your data clean and your analysis accurate.

For Google Ads account [Account ID] and Meta Ads account [Account ID], analyze the last 30 days...

Say which currency you want totals in

Each ad account reports in its own currency. When a client runs accounts in more than one currency, ask for the currency column and tell the assistant which exchange rate to use before it adds anything up, or create a custom field in Windsor.ai with a fixed rate so every report uses the same one.

Set a spend threshold in audit prompts

When running budget waste or keyword audits, always define a minimum spend threshold in your prompt. Without it, the AI will flag hundreds of low-spend items that are not worth the attention.

Setting a threshold like $200 minimum spend keeps the output focused on the decisions that drive real results.

Only include campaigns and ad groups where  cost exceeds $200 over the last 30 days...

Run weekly anomaly checks

A monthly report shows problems only at the end of the month. Set up a weekly prompt that checks for sudden CPA spikes, CTR drops, budget pacing issues and campaigns that paused unexpectedly.

Compare the last 7 days against the previous 7-day average and flag any metric that changed by more than 25% in either direction...

Platform-level for reporting, cross-channel for strategy

Use cross-channel prompts when you need a high-level comparison across platforms. For reporting and audits, run separate prompts for each platform (Google Ads, Meta, LinkedIn) to get more accurate and detailed insights. Avoid combining too much into one query; splitting analysis into smaller, focused prompts improves clarity, reduces noise, and makes recommendations more actionable.

Use shared field names when you compare platforms

Spend, clicks, impressions and campaign have the same names on every ad platform in Windsor.ai, and spend works for Google Ads as well as Meta Ads. Ask for those when you put platforms side by side, and add the datasource field to see which platform each row came from. Switch to platform-only fields, such as member_job_title on LinkedIn Ads or quality_score on Google Ads, when the question is about one platform.

Cross-reference platform ROAS with GA4

Platform-reported ROAS uses each platform’s own attribution window and counting rules, so Google Ads, Meta Ads and LinkedIn Ads can each claim the same sale. Add GA4 or your CRM to monthly reporting prompts and compare the revenue the platforms report with the revenue those systems record.

Standardize your prompts for repeatable workflows

Turn your most useful prompts into templates (weekly checks, audits, reports). This makes analysis faster, more consistent, and easier to scale across accounts or clients.

Conclusion

Windsor MCP puts Google Ads, Meta Ads, LinkedIn Ads and your other ad platforms in one place for ChatGPT, Claude and other AI assistants, with shared names for spend, clicks and impressions. Start with the cross-platform digest in use case 1, then add the wasted spend audit as a weekly check.

For the setup steps, see how to use Windsor.ai for paid media. For dashboards built on the same accounts, see the paid media dashboard templates.

🚀 Get started with Windsor.ai for free and connect your first paid media data source to AI in under a minute: https://onboard.windsor.ai/