Supermetrics lets you plug your live marketing data into AI agents like Claude, ChatGPT, Google Gemini, and Microsoft Copilot. You can also use the Supermetrics Insights Agent on the Supermetrics Hub.
AI agents are very good at answering questions about your data. They can also compare different types of data, and based on the data, generate summaries and explanations.
In this article, we highlight some practical examples of what you can do in different AI agents
Note
To ask questions from your data, first connect Supermetrics to the AI agent you're using.
Best practices in using AI agents to work with your data
Define what you need: When querying data, ensure that you specify the following in your prompt to guarantee the accuracy and relevance of the responses:
Date range, such as 'past 7 days' or 'last month'
Metrics, such as 'Impressions', 'Clicks', 'Cost', 'CTR', or 'CPC'
Dimensions for breakdown, such as 'breakdown by country' or 'use age group for rows'
Specify new tasks or new tasks to switch to: Start each task with a specific prompt that informs the AI agent about the task at hand. For example, "Let's gather advertising data for an educational app and then analyze it to refine marketing strategies", or "Next, help me with analyzing marketing data for my fashion webshop".
Break series of tasks to steps: If you need the AI agent to perform a series of tasks, break them down into smaller, more manageable steps. This will help the agent understand your instructions and execute the tasks accurately.
Ask the AI agent to explain its reasoning: Often, questions seem obvious and simple to a human but actually involve many more steps of reasoning for an AI. "Let's think step by step" or "Explain your chain of reasoning before answering" would be good instructions to be included in your prompt. This helps understand the logic behind the insights, aiding better decision-making.
Types of questions to ask your AI about data
Natural‑language KPI questions
“How much did we spend on the Acme Corp Google Ads account last month?”
“Which Facebook campaigns had the best ROI this quarter?”
“Show me our website traffic trends from Google Analytics.”
“Compare Instagram vs. TikTok ad performance.”
“What are our top converting keywords?”
Cross‑platform comparisons
“Compare cost per acquisition across my Facebook Ads campaigns.”
“Create a summary comparing all my clients’ Facebook ad performance this month.”
“Compare Meta vs LinkedIn cost per lead and conversion volume over the last 90 days.”
Automated reporting and text summaries
“Generate a short management summary of campaign performance across Google Ads, Meta, and LinkedIn for last month. Include top KPIs and key changes vs. the previous month.”
“Explain any anomalies you find in last week’s traffic from GA4 and propose 3 hypotheses for why they happened.”
Use case examples for ChatGPT
Executive 30–90 day performance deck
Goal: Turn cross‑channel performance data into a slide‑ready executive review.
Example flow:
Pull the data
“Pull the last 30 days of performance data from my Google Ads and Facebook Ads accounts. Clicks, Cost, and Impressions by data source, date, and campaign name.”
Build the deck outline
“Create a marketing performance overview as a slide‑ready deck, including in different slides a summary intro, individual channel‑level deep dives, cross‑channel comparisons, strategic recommendations, a next‑period forecast, and actionable next steps. Structure it into a clean, concise slide outline with titles, subtitles, bullet points, and insight callouts.”
Generate the exec email
“Write a concise email summarizing the deck to share with my CMO.”
Outcome:
Ready‑to‑use deck structure for quarterly or Month-over-Month reviews
Clear talking points and recommendations, grounded in real data
Saved time vs. building slides manually from spreadsheets
Week‑over‑week campaign optimization
Goal: Compare a single campaign across channels week‑over‑week and get an optimization plan.
Example flow:
Pull Week-over-Week data
“Pull performance data for my ‘Brand Testimonial’ campaign running on LinkedIn Ads and Facebook Ads. Retrieve data for the last 7 days and the previous 7‑day period.”
Visual WoW comparison
“Provide a cross‑channel visual performance analysis of the campaign, comparing this week vs. the previous week.”
Optimization plan
“Provide a document that includes recommended optimizations for my campaign.”
Outcome:
Simple Week-over-Week charts/tables (CPA, spend, clicks, conversions)
Channel‑specific actions (pause, scale, test) derived from the data
Reusable “performance review” prompt pattern that can be run weekly
Use case examples for Claude
90‑day cross‑channel performance review
Goal: Produce a 90‑day multi‑channel performance analysis and executive summary.
Example flow:
Pull 90‑day data
“Pull the last 90 days of performance data from my LinkedIn Ads and Facebook Ads accounts.”
Build the deck
“Create a marketing performance overview as a slide‑ready deck, including in different slides a summary intro, individual channel‑level deep dives, cross‑channel comparisons, strategic recommendations, a next‑period forecast, and actionable next steps. Structure it into a clean, concise slide outline with titles, subtitles, bullet points, and insight callouts.”
Leadership email
“Write a concise email summarizing the deck to share with my CMO.”
Outcome:
Consistent, repeatable analysis pattern for quarterly reviews
Strong “reasoning‑heavy” narrative leveraging Claude’s strengths
Weekly paid media budget pacing report
Goal: Generate a weekly snapshot of spend pacing vs. budget across all paid channels, flag under/over-delivery, and recommend reallocation.
Example flow:
Pull week-to-date and month-to-date spend data
"Pull this week's and month-to-date spend, impressions, clicks, and conversions from my Google Ads, Facebook Ads, and LinkedIn Ads accounts."
Build a pacing dashboard document
"Create a visual report showing each channel's actual spend vs. planned budget, percent paced, projected end-of-month spend at current run rate, and a RAG status for each. Include a reallocation recommendation section based on which channels are outperforming on CPA."
Slack summary for the media team
"Write a short Slack message for my paid media team summarizing this week's pacing status, flagging any channels that need immediate attention, and listing the top 3 recommended budget shifts."
Outcome:
Eliminates manual spreadsheet pacing calculations every Monday
Proactive budget reallocation recommendations grounded in live data
Clear team communication artifact ready to post
Creative fatigue and ad refresh analysis
Goal: Identify which ad creatives are fatiguing across channels and produce a brief recommending refreshes, pauses, or scale-ups.
Example flow:
Pull ad-level performance over time
"Pull the last 60 days of ad-level data from my Facebook Ads and LinkedIn Ads accounts — I need ad name, spend, impressions, CTR, CPC, and conversion rate broken out by week."
Analyze creative performance trends
"Identify which ads show declining CTR or rising CPC over the past 4 weeks while still receiving significant spend. Flag these as 'fatiguing.' Also highlight any newer ads with improving metrics that could absorb more budget. Present this as a creative health report with a table of fatiguing ads, rising ads, and stable performers, plus a recommended action for each."
Brief for the creative team
"Write a short creative brief document summarizing which themes and formats are fatiguing, what's working, and 3–5 specific creative directions to test next based on the patterns in the data."
Outcome:
Data-driven creative refresh cycle instead of gut-feel rotation
Bridges the gap between media buying data and creative strategy
Repeatable monthly or bi-weekly cadence
Competitor benchmarking and share-of-voice snapshot
Goal: Combine your own paid performance data with publicly available benchmarks to assess competitive positioning and identify gaps.
Example flow:
Pull your own channel metrics
"Pull the last 90 days of performance data from my Google Ads and Facebook Ads accounts — focus on CPM, CPC, CTR, and conversion rate by campaign type."
Benchmark and contextualize
"Compare my metrics against industry benchmarks for B2B SaaS paid media. For each metric and channel, show where I'm above, at, or below the benchmark, and calculate the gap. Present this as a competitive positioning scorecard with a summary of my strongest and weakest areas relative to the market."
Strategic recommendations memo
"Write a one-page strategy memo for my VP of Marketing outlining where we're outperforming the market and should double down, where we're underperforming and need to investigate, and 3 specific initiatives to close the biggest gaps over the next quarter."
Outcome:
Turns internal data into a market-relative story that leadership actually cares about
Combines Supermetrics data pull with Claude's reasoning to contextualize raw numbers
Natural quarterly or campaign-launch cadence
Use case examples for Google Gemini
Monthly cross‑channel performance review in Gemini
Goal: Let leaders and marketers ask “How are we doing?” directly in Gemini Enterprise
Example prompts:
“Give me a marketing performance summary for all channels this month.”
“Compare Meta Ads and Google Ads ROAS for Q1.”
“Which campaigns are pacing ahead of budget this month?”
What happens:
Gemini Enterprise agent uses Supermetrics MCP tools to:
Discover connected sources (such as Google Ads, Meta, LinkedIn, GA4)
Identify the relevant accounts to use
Pull the relevant metrics in real time
Return a narrative summary and optional charts
Outcome:
Self‑serve, natural‑language access to live performance data in the Google Workspace that Supermetrics users already use.
Deep‑dive analysis on a specific channel or KPI
Goal: Let analysts and managers explore a particular area without dashboards.
Example prompts:
“Show me our email open and click‑through rates over the past 6 months.”
“Identify campaigns with the highest cost per lead in LinkedIn Ads last quarter.”
“Create a bar chart of ad spend by channel for the last quarter.”
Outcome:
Quick exploratory answers and visuals without needing a BI tool
Great for ad‑hoc questions, pre‑meeting prep, or hypothesis testing
Use case examples for Microsoft Copilot
Performance snapshots directly in Teams/Copilot
Goal: Give Teams users quick access to marketing KPIs via Copilot.
Example prompts:
“Using Supermetrics Marketing Analyst, summarize last month’s Google Ads performance. Highlight the top 3 campaigns by ROAS.”
“Show Meta + LinkedIn spend and conversions by country for Q1.”
Outcome:
Instant summaries inside the Microsoft 365 environment
Easy for marketers and stakeholders who live in Teams to stay on top of performance
Auto‑draft stakeholder updates in Outlook / Teams
Goal: Turn data into comms (emails, Teams posts) without manual copying.
Example prompts:
“Based on our latest campaign performance from Supermetrics, draft a Teams update summarizing key wins and risks for the growth team.”
“Write an email to our VP Marketing summarizing where we should reallocate budget next month, using our Supermetrics data.”
Outcome:
Faster reporting and stakeholder communication
Consistent format for weekly/monthly updates, built on live data