Windsor MCP for Agencies: Better Client Reports, KPI Dashboards, and Strategy

Agency reporting usually starts with exports. An account manager opens Meta Ads Manager, Google Ads, TikTok Ads and GA4, downloads a file from each, matches the date ranges in a spreadsheet and then writes the summary for every client.

Windsor MCP connects those platforms, plus LinkedIn Ads, Shopify, CRMs and 350+ other data sources, to Claude, ChatGPT and other AI assistants. You ask for the report in plain English. The assistant pulls live data for the client’s accounts and writes the summary, the channel comparison or the KPI dashboard.

This guide covers seven agency use cases. Each one lists the problem it solves, ready-to-copy prompts and the output you get.

🚀 Start for free with Windsor MCP. The free forever plan needs no credit card. Cross-channel reports use more than one data source: the first 30 days include up to 10 sources, and after that paid plans start at $19/month for 3 sources or $99/month for 7.

How agencies use Windsor.ai with AI tools: use cases and prompts

Replace the values in square brackets with your client’s details and account IDs. The prompts work in Claude, ChatGPT and any other AI client connected to Windsor MCP.

1. Weekly cross-channel KPI digest

The challenge:

Five clients need a performance summary before the Monday team meeting. Without a connection to the data, that means opening every platform the client runs campaigns on, downloading exports, comparing the numbers and writing each report by hand.

With Windsor MCP, one prompt returns the summary for all of those channels.

Best for:

• Account managers
• Paid media leads
• Client success teams

Prompt: Weekly paid media digest

You are an analyst for [Agency Name]. Use windsor.ai to generate a weekly performance digest for [Client Name] covering the last 7 days vs. the previous 7 days.

Pull data from [Your Data Sources]:
- Meta Ads (account ID: XXXX)
- Google Ads (account ID: XXXX)
- TikTok Ads (account ID: XXXX)

For each channel, include:
- Total spend
- Impressions
- Clicks
- CTR
- Conversions
- CPA

Calculate blended CPA across all three channels.

Highlight:
1. The channel with the biggest CPA improvement
2. Any channel where spend increased but conversions dropped
3. 5 recommendations for next week

Format the output as a concise summary followed by a short narrative overview.

Prompt: Weekly client email summary

Use windsor.ai data to create a brief, client-facing performance update for [Client Name] covering Meta Ads for the week [dates].

Write in plain language suitable for a marketing director who is not a paid media specialist.

Include:
- Total spend
- Leads generated
- Cost per lead
- Top-performing campaign

Suggest next steps.
Keep the response under 150 words.

What you get:

A client-ready KPI summary with trends and recommendations that you can paste into Slack, an email, a slide deck or a reporting dashboard.

Give the AI enough context to write a complete digest: the client name, the reporting period and the KPIs that matter most to that client. In Claude, you can also ask for a visual dashboard built from the same data.

Example from our own accounts: we ran the weekly digest prompt in Claude on windsor.ai’s Google Ads and Meta Ads accounts, comparing 30 Sep to 6 Oct 2026 with the week before. The answer, with our absolute spend left out:

Week over week	Google Ads	Meta Ads
Spend	+27%	+6%
Clicks	+16%	+12%
Conversions	+40%	13 leads, down from 16
CPA	-9%	not compared

• Google Ads conversions grew faster than spend, so the extra budget in the second week converted at a lower cost per conversion than the first week.
• Meta Ads was 0.1% of combined spend, a small test budget. The assistant kept it out of the blended CPA and flagged that 13 leads are too few to compare week over week.

The second point is the kind of check that matters in client reports: a blended CPA that mixes a large channel with a small test can hide what each channel does.

2. Monthly executive client report

The challenge:

At the end of the month, clients expect context, explanations and a plan for the next month, not only the numbers. Combining spend, channel performance, creative results and month-over-month trends into a report or deck takes hours when it is done by hand.

With Windsor MCP, one prompt produces the full executive summary.

Best for:

• Agency and team leads
• Client strategists
• Analysts

Prompt: Full monthly report narrative

Using Windsor.ai data, act as the lead strategist for [Client Name]'s account at [Agency Name].

Generate a monthly performance report for [Month Year].

Use data from [Data Sources]:
- Meta Ads (account ID: XXXX)
- Google Ads (account ID: XXXX)
- GA4 (property ID: XXXX)

Structure the report as follows:

1. Executive summary
- Summarize overall performance in 3-4 sentences.

2. Channel performance
- Include spend, conversions, CPA, and ROAS for each channel.

3. Month-over-month comparison
- Show the key metric changes compared to the previous month, including percentage changes.

4. What worked
- Highlight the top two campaigns, channels, or tactics with supporting data.

5. What needs attention
- Flag the top two issues or areas of concern with supporting data.

6. Recommended actions
- Suggest three specific, data-backed actions for each channel for next month.

Write in a professional but direct tone. The audience is a CEO.

Prompt: Google Ads competition summary

Use the Windsor.ai connector. Based on [Client Name]'s Google Ads performance over the last 30 days:

- Analyze impression share trends
- Analyze average CPC trends
- Explain what this suggests about competitive pressure in the category
- Determine whether CPCs are rising faster than conversions
- Determine whether efficiency is improving or declining

Write a two-paragraph market context summary suitable for a client report.

What you get:

A monthly report with channel performance, trends and next-step recommendations that you can add to a client deck or executive summary.

3. Cross-channel budget reallocation

The challenge:

Clients ask every month whether the budget is in the right places. Answering properly means comparing Meta Ads, Google Ads, TikTok Ads and LinkedIn Ads side by side, not one channel at a time.

With Windsor MCP, the assistant reads all of these accounts in one request. It shows which channels deliver the best CPA, where budget is wasted and where more spend could bring more conversions.

Best for:

• Paid media strategists
• Performance marketing leads
• Agency owners presenting budget recommendations

Prompt: Cross-channel budget efficiency analysis

Use Windsor.ai. Compare performance across all paid channels for [Client Name] over the last 30 days.

Use data from [Data Sources]:
- Meta Ads (account ID: XXXX)
- Google Ads (account ID: XXXX)
- TikTok Ads (account ID: XXXX)
- LinkedIn Ads (account ID: XXXX)

For each channel, include:
- Total spend
- Total conversions
- CPA
- ROAS
- Percentage of total budget

Then answer:
1. Which channel has the best CPA?
2. Which channel has the worst CPA?
3. If we moved 15% of budget from the worst-performing channel to the best-performing channel, what would the projected increase in conversions be at current efficiency rates?

Format the output as a summary with key findings and recommend how we should optimize the budget.

Prompt: Identify wasted spend

Based on the Windsor.ai data, for [Client Name]'s paid media accounts, identify campaigns and ad sets that are producing low or no return over the last 30 days.

Look for:
- Campaigns with more than $500 in spend and zero or near-zero conversions
- Campaigns where CPA is more than 2x the account average
- Ad sets with high frequency and declining CTR

For each finding, include:
- Campaign or ad set name
- Spend amount
- Conversions
- CPA
- A short explanation of the likely issue

Finish with a summary of the biggest areas of wasted spend and the recommended next actions.

What you get:

One view of where each client’s money is spent efficiently, where it is wasted and how the budget could move to improve results.

Windsor can also make the change. On Meta Ads, Google Ads, TikTok Ads, LinkedIn Ads and Microsoft Ads, you can ask the assistant to pause a campaign or change its budget, and it applies the change after you confirm it. See Windsor.ai write actions for the full list.

4. Creative performance audit across platforms

The challenge:

Most brands run different creative formats on Meta, TikTok, Instagram and LinkedIn, and each platform reports them in its own dashboard. That makes these questions slow to answer:

• Which creative formats convert best?
• Which hooks and messages work on more than one channel?
• Which creatives are starting to fatigue?
• Where is too much budget going to low-performing formats?

With Windsor MCP, you can compare creative performance across all channels in one conversation.

Best for:

• Creative strategists
• Social media managers
• Content teams

Prompt: Creative format performance comparison

Use Windsor.ai data. For [Client Name], compare creative format performance across [Meta Ads and TikTok Ads] over the last 60 days.

On Meta Ads, compare:
- Single image ads
- Carousel ads
- Video ads

For each format, include:
- CTR
- CPA
- Conversion rate

On TikTok Ads, compare in-feed video performance by:
- 2-second view rate
- Conversion rate
- CPA

Then answer:
- Which creative format is generating the most cost-efficient conversions on each platform?
- Are there any formats receiving too much budget despite underperforming against the account average?

Finish with a short creative strategy recommendation.

Prompt: Detect creative fatigue

Use Windsor.ai. Scan all active creatives for [Client Name] across [Meta Ads and TikTok Ads] over the last 30 days.

For each creative, include:
- Platform
- Spend
- CTR
- Frequency
- Conversion rate

Flag any creative where:
- Frequency has risen above 3.5 in the last 7 days
- CTR has dropped by more than 25% week over week
- Conversion rate is declining while spend remains stable or increases

For each flagged creative, recommend one of the following actions:
- Refresh creative
- Pause creative
- Test a new variation (suggest how to change it)

Prompt: Identify winning creative patterns

Using the Windsor.ai connector, look at the top 5 highest-converting posts for [Client Name] on [Data Source] over the last 30 days.

Analyze what they have in common, including:
- Hook style
- CTA type
- Post format
- Product angle
- Messaging style
- Use of social proof

Then summarize the creative patterns that appear most often in top-performing posts.
Use these findings to generate a content plan for the next 10 posts.

What you get:

A list of the creative formats, messages and hooks that perform best, early warnings on creative fatigue and suggestions for what to test next.

5. Anomaly detection and early warnings

The challenge:

A CPA spike, a falling conversion rate, an unexpected rise in spend or a paused campaign can go unnoticed until the weekly report, and the budget keeps running in the meantime.

With Windsor MCP, you can run a daily health check across all client accounts and catch these issues on the day they start.

Best for:

• Account managers
• Paid media leads

Prompt: Daily account health check

Use Windsor.ai and run a health check on [Client Name]'s paid media accounts for the last 48 hours compared to the previous 7-day average.

Analyze:
- Spend
- CPA
- CTR
- Conversion volume
- ROAS

Flag any metric that changed by more than 25% in either direction.

For each anomaly, include:
- The metric name
- The percentage change
- The likely cause, if it can be determined from the data
- The urgency level (Watch, Investigate, Act now)

Format the output as a short bullet list that can be reviewed in a few minutes. Suggest how to fix each issue.

Prompt: Cross-client anomaly scan

Use Windsor.ai and scan all connected Meta Ads accounts for the last 7 days.

For each client account, flag campaigns where:
- Spend increased by more than 30% compared to the previous week without a corresponding increase in conversions
- CPA is more than 50% above the account's 30-day average
- A campaign paused unexpectedly
- A campaign exhausted its budget before month-end

For each issue, include:
- Client name
- Campaign name
- Metric affected
- Severity level
- Recommended next action

Ignore accounts with less than $100 in spend during the period.

What you get:

A daily check that flags wasted spend, performance drops and pacing problems before the client sees them in a report.

6. B2B lead quality analysis: from click to pipeline

The challenge:

A B2B campaign can show a low cost per lead in LinkedIn Ads or Google Ads while most of its leads never become sales opportunities. To see which channels and campaigns bring qualified pipeline, you need the ad data and the CRM data in the same analysis.

With Windsor MCP, you can follow leads from the ad click to MQL, SQL, opportunity and closed revenue.

Best for:

• B2B agency leads
• Demand generation specialists
• Client success teams managing CRM-connected accounts

Prompt: Lead quality by channel

Use Windsor.ai to combine [Client Name]'s LinkedIn Ads data with their HubSpot CRM data.

For leads generated from LinkedIn Ads over the last 30 days, analyze:
- Percentage of leads that became MQLs
- Percentage of leads that became SQLs
- Average deal size for closed-won deals
- Cost per SQL

Then compare this to Google Ads leads from the same period.

Answer:
- Which channel produces the highest-quality leads?
- Which channel has the strongest MQL-to-SQL conversion rate?
- Which channel has the lowest cost per SQL?
- Which channel generates the highest-value deals?

Finish with a short recommendation on where budget should be increased or reduced.

Prompt: Lead quality by campaign

For [Client Name]'s HubSpot CRM, analyze all contacts created in the last 90 days with a lead source of [LinkedIn Ads or Google Ads].

Break down performance by campaign using the UTM source, medium, and campaign fields.

For each campaign, include:
- Number of leads
- Number of MQLs
- Number of SQLs
- MQL conversion rate
- SQL conversion rate
- Average days to SQL

Then identify:
- Campaigns producing the most qualified leads
- Campaigns producing high lead volume but low quality
- Campaigns where leads never progressed beyond the subscriber or MQL stage

Flag any campaigns that may need a landing page, targeting, or offer update, and suggest those changes.

What you get:

A view of which channels and campaigns bring real pipeline, so you can judge them on SQLs and revenue as well as cost per lead.

7. E-commerce performance: from ad spend to revenue

The challenge:

For e-commerce clients, the number of orders is only part of the answer. You also need to know which channels bring customers who place larger orders, buy again and keep what they bought.

That means putting ad spend next to store revenue, repeat purchases and customer value over time. With Windsor MCP, the ad platforms and the store are in the same conversation.

Best for:

• E-commerce specialists
• Performance agencies with DTC clients
• Teams combining ad and store data

Prompt: True blended ROAS analysis

Using Windsor.ai data, combine [Client Name]'s Meta Ads, Google Ads, TikTok Ads, and Shopify data.

Calculate the true blended ROAS for the last 30 days using:
- Total Shopify net revenue after refunds
- Total ad spend across all paid channels

Then compare this to the self-reported ROAS inside each ad platform.

Answer:
- Which platform's reported ROAS is closest to the blended figure?
- Which platform appears to overstate performance the most?
- By how much does each platform differ from the true blended ROAS?

Finish with a short recommendation on which number should be used in client reporting.

Prompt: Customer quality by acquisition channel

Using Windsor.ai, for [Client Name]'s Shopify store, analyze customers acquired over the last 30 days.

Segment customers by first purchase source:
- Meta Ads
- Google Ads
- TikTok Ads

For each segment, include:
- Average order value
- Average orders per customer
- Repeat purchase rate
- Total customer spend after 90 days

Then answer:
- Which channel acquires customers with the highest lifetime value?
- Which channel drives the strongest repeat purchase behavior?
- Does this change how budget should be allocated across channels?

Prompt: Product-level ad efficiency

Use Windsor.ai. For [Client Name]'s top 10 products by Shopify revenue last month:

Identify:
- Which products are being actively promoted in [Meta Ads or Google Ads]
- Which products are generating strong revenue without paid promotion
- Which products have the highest product-level ROAS
- Which products may deserve more paid support because of strong organic demand

For each product, include:
- Revenue
- Ad spend
- ROAS
- Main traffic source

Finish with a short recommendation on which products should receive more or less advertising budget.

What you get:

A view of which channels, products and customer segments bring profitable growth, so budget decisions can use repeat purchases and lifetime value as well as short-term ROAS.

Tips for agencies using Windsor MCP

These tips help you use the prompts above in the same way across all of your clients.

1. Keep each client’s data separate

One Windsor account can hold many data sources, which works well for a small agency. As you add clients, consider a separate Windsor account per client so each client’s data stays isolated.

If you manage every client from one account, include the account ID in every prompt. A prompt that asks about “all connected Meta accounts” without a filter blends data across clients. That gives wrong numbers and can expose one client’s data in another client’s report.

Filter by account in your prompts:

For [Client Name], Meta Ads account ID XXXX, analyze the last 30 days...

2. Build a prompt library for your most common reports

Treat the prompts in this guide as starting points. When you find the versions that work for your clients, save them as an internal library with variants for the weekly digest, the monthly executive summary, the budget recommendation, the anomaly alert and the creative audit. New account managers can then run them without writing prompts from scratch. The Windsor.ai prompt library has more prompts by data source.

3. Ask for the output format you need

Add formatting instructions to the end of any prompt:

...Output this as:
- A summary table (channel | spend | CPA | trend)
- Followed by a 4-bullet executive summary
- Followed by 3 recommended actions, each with the data point that supports it

Or:

...Write this as a slide narrative: 3 short paragraphs that a presenter can read aloud during a monthly client call. 200 words maximum.

4. Choose the assistant by the deliverable

The prompts work the same way in Claude and ChatGPT. Claude can turn the answer into an interactive dashboard or chart in the same chat, which is useful for visual client reports. ChatGPT can export the answer as a document or spreadsheet file. Use the assistant your team already works in.

5. Add organic sources next to paid

The use cases above focus on paid media, where most agency reporting time goes. Windsor also connects organic sources such as Instagram, Facebook Pages, LinkedIn Pages, Google Search Console and YouTube.

With paid and organic data connected, you can check how paid campaigns affect branded search, which channels send the best traffic and which content topics convert. You can also see whether paid media creates new demand or makes up for low organic visibility.

How to connect Windsor MCP

1. Claude: open Customize, then Connectors, search for “Windsor” and click Add. Sign in to Windsor.ai and connect the client accounts you want to analyze.
2. ChatGPT: open Plugins in the sidebar, search for “Windsor” and click Install plugin. Sign in to Windsor.ai and connect your accounts.

Step-by-step guides: Windsor MCP in Claude and Windsor MCP in ChatGPT.

Conclusion

Clients pay an agency to explain what changed, find opportunities and recommend the next step. Windsor MCP with Claude, ChatGPT or another AI assistant takes over the data collection and first draft of each report, so your team spends its time on the analysis and the recommendations.

Start with one client and one prompt from this guide, and compare the output with your current report.

🚀 Start for free at Windsor.ai and connect your first client’s data in about a minute. No credit card needed.

Explore more

• Windsor MCP use cases and prompts
• Windsor MCP for PPC
• Windsor MCP for e-commerce
• AI Prompt Library by Windsor.ai
• AI analytics learning hub