15 Claude Prompts for GA4 Traffic, Funnels & Attribution (2027)

Connect Google Analytics 4 to Claude through Otto's Claude MCP connector (https://otto.islaintel.com/api/mcp) to run 15 live prompts across traffic acquisition, multi-step funnel drop-offs, and key-event attribution without exporting CSV files. Uploading a raw 3,000-row Ads Manager CSV into Claude can consume 40,000 to 80,000 tokens in a single turn, whereas querying aggregated metrics through Otto MCP cuts token usage by 80–90%. Querying live API endpoints lets you cross-check read-only GA4 data against Search Console, Meta Ads, Google Ads, Shopify, Google Sheets, and PageSpeed Insights across 227 tools and 15 integrations.

The 15 prompts fall into three operational groups: five runReport traffic and schema diagnostics, five runFunnelReport and landing-page drop-off checks, and five attribution-model and cross-channel revenue reconciliations. If you've only got two minutes on a Monday morning, run Prompt 1 first to get a clean breakdown of weekly channel movements and engagement changes.

Starter prompts give you immediate answers for core weekly analytics questions without manual data filtering. Paste these five starter prompts directly into Claude when you want a fast pulse check before diving into the full 15-prompt breakdown—they're quick-copy versions of Prompts 1, 2, 6, 11, and 12.

Starter prompt 1: Weekly channel traffic and engagement digest

Query GA4 via runReport for the last 7 days compared to the prior 7 days.
Group by sessionDefaultChannelGroup.
Return a table with:
- Sessions
- Engaged sessions
- Engagement rate
- Average engagement time per session
- Key events
Flag any channel where sessions dropped by more than 15% or key events declined while traffic remained flat.
Provide three operational takeaways explaining what caused the variance.

Starter prompt 2: Landing page key-event performance

Query GA4 via runReport for the last 30 days.
Break down performance by pagePathPlusQueryString.
Filter to landing pages with at least 100 sessions.
Return columns for:
- Sessions
- Bounce rate
- Key events
- Key event conversion rate
Identify the three landing pages with the lowest conversion rate despite high traffic.
Suggest specific above-the-fold copy adjustments for each.

Starter prompt 3: Multi-step checkout or lead funnel drop-off

Query GA4 via runFunnelReport for the last 30 days.
Map the steps:
1. session_start
2. view_item (or page_view on key landing page)
3. begin_checkout (or lead form open)
4. purchase (or generate_lead)
Show users at each step, step drop-off count, and step completion percentage.
Break the funnel down by deviceCategory to pinpoint mobile versus desktop leakage.

Starter prompt 4: Data-driven versus last-click attribution comparison

Query GA4 attribution data for the last 30 days across all key events.
Compare:
1. Data-driven attribution
2. Paid and organic last click
3. Google paid channels last click
Display key event credit side by side across sessionDefaultChannelGroup.
Highlight upper-funnel acquisition channels that receive significantly higher credit under data-driven attribution than under last-click models.

Starter prompt 5: Meta and Google Ads vs. GA4 cross-channel audit

Query the last 14 days of paid traffic.
Compare reported clicks and spend from connected Meta Ads and Google Ads accounts against GA4 sessions, engaged sessions, and key events mapped to utm_source=meta and utm_source=google.
Calculate cost per GA4 session and cost per GA4 key event for each platform.
Flag tracking discrepancies where ad platform reported clicks exceed GA4 sessions by over 20%.

You can query GA4 in Claude three ways: manual CSV exports, Google's local Python server, or Otto's hosted MCP connector. The Model Context Protocol standard lets AI applications like Claude connect directly to external data sources and tools. When you connect Google Analytics 4 (GA4) to Otto, you query GA4 alongside 14 other connected marketing and website integrations so you don't paste platform keys into chat.

According to Google for Developers, Google's official Google Analytics MCP server connects GA4 data to AI agents for natural-language queries, runs locally using Python and gcloud Application Default Credentials, and is strictly read-only. Otto Webmaster connects 227 live tools across 15 integrations to Claude using a single organization-scoped bearer token prefixed with otto_mcp__. Documented AI clients supported by Otto MCP include Claude, ChatGPT, Cursor, Claude Code, OpenClaw, and Hermes, while Otto core is also accessible via WhatsApp, web app, and email. Users don't paste private keys into Claude; credentials remain protected on Otto's server.

Connect Claude to Otto MCP

Diagnosing traffic channels requires inspecting volume, engagement quality, and referring platforms. You can pair these prompts with Otto Website Analytics and weekly briefings so you don't lose track of baseline performance. Under Google Analytics Data API runReport specifications, reports return customized tables of event data organized by requested dimensions and metrics.

Prompt 1: 7-day channel velocity and engagement audit

• What it pulls and why it matters: Compares weekly session volume, engagement rates, and key events across default channel groups to spot drop-offs.
• API method or tools used: GA4 Data API runReport comparing the last 7 days against the prior 7 days.
• The catch: GA4 is strictly read-only in Otto MCP, and runReport derives session and event statistics from the data collected by your Google Analytics tracking code.

Claude prompt:

Query GA4 via runReport for the last 7 days versus the previous 7 days.
Dimensions: sessionDefaultChannelGroup
Metrics: sessions, activeUsers, engagedSessions, engagementRate, averageSessionDuration, keyEvents
Output a Markdown table sorted by absolute session change.
Add a column calculating percentage change for sessions and key events.
Below the table, identify:
1. The channel with the strongest positive velocity.
2. Any channel experiencing traffic decay.
3. Recommended budget adjustments based on engagement quality.

Prompt 2: Organic landing page traffic paired with Search Console

Prompt 2 cross-references landing page sessions against organic search impressions and ranking queries. On June 24, 2026 (v1.2.0), Otto shipped Automated SEO Reports that generate weekly SEO reports highlighting areas for improvement and tracking keyword rankings over time.

• What it pulls and why it matters: Evaluates whether pages gaining organic search impressions generate engaged sessions and key events.
• API method or tools used: GA4 Data API runReport combined with read-only Google Search Console reporting tools.
• The catch: GA4 and Search Console are strictly read-only in Otto MCP, and Search Console tracks Google searches while GA4 captures all organic engines.

Claude prompt:

Query GA4 organic search landing pages for the past 28 days using runReport.
Dimensions: pagePathPlusQueryString
Metrics: sessions, engagedSessions, keyEvents, bounceRate
Cross-reference the top 10 pages against Google Search Console data for total impressions, clicks, and average position.
Identify pages that receive strong organic impressions but have bounce rates above average.
Provide three recommendations to improve page content and retention.

Prompt 3: AI assistant and emerging referral segmentation

• What it pulls and why it matters: Tracks session volume and engagement trends from AI assistants and conversational search engines.
• API method or tools used: GA4 Data API runReport with regex filters applied to sessionSource.
• The catch: Under GA4 runReport specifications, reports derive event statistics from tracking code data, and read-only GA4 queries reflect inbound referral visits rather than unclicked citations.

Claude prompt:

Query GA4 referral traffic for the last 90 days.
Dimensions: sessionSource, sessionMedium, landingPagePlusQueryString
Metrics: sessions, engagedSessions, engagementRate, keyEvents
Filter sessionSource using regex: '(chatgpt|openai|claude|anthropic|perplexity|copilot)'.
Calculate total sessions and conversion rate for AI referral sources.
Compare these metrics against your standard organic search benchmarks.

Prompt 4: Anomalous source and medium bounce detector

• What it pulls and why it matters: Detects source and medium combinations showing high session counts with near-zero engagement.
• API method or tools used: GA4 Data API runReport sorted by low average engagement duration.
• The catch: Under Google Analytics Data API runReport rules, data derives from site tracking code, and read-only GA4 access means referral exclusions or UTM tag corrections must be applied in site code or GA4 settings.

Claude prompt:

Query GA4 source and medium combinations for the last 14 days using runReport.
Dimensions: sessionSourceMedium
Metrics: sessions, engagedSessions, bounceRate, averageSessionDuration
Filter for sources with at least 50 sessions.
Flag any source where bounce rate is unusually high or average engagement duration is below 5 seconds.
Classify each flagged source as suspected referral spam, misconfigured UTM tagging, or poor targeting.

Prompt 5: Custom dimension and event schema integrity audit

• What it pulls and why it matters: Verifies event tracking parameters conform to your analytics schema and record live counts.
• API method or tools used: GA4 metadata endpoints and event-level runReport queries.
• The catch: Both Google's official Google Analytics MCP server and Otto MCP inspect GA4 schemas in read-only mode, so registering custom dimensions requires GA4 Admin access.

Claude prompt:

Query GA4 event data for the past 7 days.
Dimensions: eventName
Metrics: eventCount, totalUsers
Identify custom events collected across your property.
Check whether core e-commerce or lead generation events match standard conventions.
List any custom events with zero recorded counts over the past week and outline potential tracking fixes.

Funnel analysis highlights where prospects abandon your conversion paths, so you don't waste ad spend driving traffic into a broken flow. The Google Analytics Data API runFunnelReport method returns customized funnel data so you can visualize drop-offs at each step.

Prompt 6: Multi-step lead and checkout drop-off breakdown

• What it pulls and why it matters: Measures absolute user drop-off between viewing an offer and completing a conversion.
• API method or tools used: GA4 Data API runFunnelReport.
• The catch: Under Google Analytics Data API runFunnelReport specifications, funnels visualize sequential user progression in read-only mode, requiring every step name to match active tracking events.

Claude prompt:

Query GA4 via runFunnelReport for the last 30 days.
Define the conversion funnel steps:
- Step 1: session_start
- Step 2: view_item or view_service
- Step 3: begin_checkout or form_start
- Step 4: purchase or generate_lead
Generate a table displaying:
- Users at each step
- Drop-off count from previous step
- Step-to-step drop-off percentage
- Overall funnel completion rate
Pinpoint the specific step responsible for the largest volume loss.

Prompt 7: Mobile vs. desktop funnel completion disparity

• What it pulls and why it matters: Uncovers conversion friction that disproportionately affects mobile visitors during checkout or form submission.
• API method or tools used: GA4 runFunnelReport segmented by deviceCategory.
• The catch: Cross-device visitor journeys without persistent user identifiers record as separate journeys, and GA4 remains read-only in Otto MCP.

Claude prompt:

Query GA4 funnel completion metrics for the last 60 days segmented by deviceCategory (mobile, desktop, tablet).
Track progression from landing page view to completed key event.
Calculate step completion rates for mobile compared to desktop.
Highlight any step where mobile conversion rates fall substantially behind desktop performance.
Outline three technical adjustments to resolve mobile checkout friction.

Prompt 8: High-exit landing page Core Web Vitals check

Prompt 8 connects high landing page bounce rates to page performance and Core Web Vitals issues. Review how Core Web Vitals drift causes revenue leaks to understand technical performance impacts.

• What it pulls and why it matters: Pairs landing page bounce metrics with real-time Lighthouse scores to isolate technical speed bottlenecks.
• API method or tools used: GA4 runReport combined with Otto's built-in psi_analyze_url tool.
• The catch: The psi_analyze_url tool checks Lighthouse scores and Core Web Vitals on any URL in read-only mode using Otto's platform key without requiring a user Google API key, while site fixes require staging a WordPress draft or GitHub pull request preview.

Claude prompt:

Query GA4 for the top five landing pages with the highest bounce rate over the last 30 days (minimum 200 sessions).
Dimensions: landingPagePlusQueryString
Metrics: sessions, bounceRate, averageSessionDuration
For each identified URL, run psi_analyze_url using Otto MCP to pull mobile Core Web Vitals:
- Largest Contentful Paint (LCP)
- Cumulative Layout Shift (CLS)
- Total Blocking Time (TBT)
- Performance score
Present a consolidated table comparing bounce rates against performance scores.

Prompt 9: New versus returning user funnel velocity

• What it pulls and why it matters: Shows whether marketing campaigns close prospects on the first visit or require multi-touch nurturing.
• API method or tools used: GA4 Data API runFunnelReport segmented by visitor type.
• The catch: Under GA4 runFunnelReport documentation, step metrics reflect data collected by tracking code in read-only mode.

Claude prompt:

Query GA4 via runFunnelReport for the last 30 days comparing new versus returning visitors.
Define the conversion funnel steps:
- Step 1: session_start
- Step 2: view_item
- Step 3: purchase
Segment the funnel by newVsReturning.
Calculate step completion rates and drop-offs for both segments.
Explain how the business should adapt retargeting and email capture workflows based on user conversion timing.

Prompt 10: Staged funnel fixes for WordPress or GitHub

Prompt 10 moves from funnel drop-off analysis to generating actual website fixes. Otto Webmaster fixes landing pages after diagnosing performance leaks by opening GitHub pull requests or WordPress drafts in the same workflow. Site changes on GitHub open as pull requests with previews, and WordPress content remains a draft until you approve it.

• What it pulls and why it matters: Connects GA4 funnel drop-off diagnostics to staged landing page edits.
• API method or tools used: GA4 runFunnelReport alongside Otto WordPress or GitHub tools.
• The catch: Website changes remain drafts or pull requests that require human review and publication before going live.

Claude prompt:

Query GA4 via runFunnelReport for the last 30 days across these steps:
1. session_start
2. view_item (or view_service)
3. begin_checkout (or form_start)
4. purchase (or generate_lead)
Identify the conversion step and landing page with the steepest user drop-off.
Draft revised landing page copy to address that drop-off, including:
- A clearer value proposition headline
- Bulleted benefit proof points
- A simplified form call to action
Stage the output as a WordPress page draft or GitHub pull request with a preview for team review.

Attribution modeling determines how conversion credit is distributed across marketing touchpoints, helping you see where last-click reporting doesn't tell the full story. According to Google Analytics Help (Get started with attribution), GA4 Attribution reports provide three attribution models: Data-driven attribution, Paid and organic last click, and Google paid channels last click. All three attribution models exclude direct visits from receiving attribution credit unless the path to the key event consists entirely of direct visits, and under data-driven attribution, conversions can be reattributed for up to seven days after the conversion.

Prompt 11: Data-driven versus last-click key-event comparison

• What it pulls and why it matters: Identifies early-stage discovery channels that last-touch models undervalue.
• API method or tools used: GA4 key-event reporting across GA4's three Attribution report models documented in Google Analytics Help (Get started with attribution).
• The catch: Direct visits don't receive attribution credit unless the path to the key event consists entirely of direct visits, data-driven attribution can reattribute conversions for up to 7 days after the conversion, and GA4 remains read-only in Otto MCP.

Claude prompt:

Query GA4 attribution data for the last 30 days across all key events.
Compare key-event credit across sessionDefaultChannelGroup under three models:
1. Data-driven attribution
2. Paid and organic last click
3. Google paid channels last click
Note that direct visits are excluded unless the path to key event consists entirely of direct visits, and under data-driven attribution, conversions can be reattributed for up to 7 days after the conversion.
Build a comparison table showing absolute key events and share of total under each model.
Highlight top-of-funnel channels that gain significant credit under data-driven attribution.

Prompt 12: Meta and Google Ads vs. GA4 conversion reconciliation

Ad platforms and GA4 don't count conversions the same way, so you can pair Prompt 12 with GA4 and paid media landing page dashboards and 10 Claude workflows for Meta Ads when auditing paid media performance. As Head of Growth & Marketing at IslaIntel, I lead paid ads, SEO, and AI marketing workflows across Meta, Google, and Claude MCP at Otto.

• What it pulls and why it matters: Compares Meta Ads Manager and Google Ads reported conversions against GA4 key events.
• API method or tools used: GA4 runReport and Google Ads read-only endpoints alongside Meta Ads reporting tools in Otto MCP.
• The catch: Under Google Analytics Help (Get started with attribution), GA4 attribution reports exclude direct visits from credit unless the path to the key event consists entirely of direct visits, and both GA4 and Google Ads remain strictly read-only in Otto MCP.

Claude prompt:

Query ad performance data for the last 14 days from our connected Google Ads and Meta Ads accounts.
Pull spend, impressions, clicks, and reported conversions.
Next, query GA4 via runReport for the same 14-day window for sessions, engaged sessions, and key events where sessionSource matches google or meta.
Build a reconciliation table comparing:
- Platform reported conversions vs. GA4 key events
- Platform reported clicks vs. GA4 sessions
- Effective cost per GA4 key event
Summarize the primary causes of reporting discrepancies.

Prompt 13: LinkedIn and X Ads B2B pipeline assist check

• What it pulls and why it matters: Evaluates assisted conversion impacts from LinkedIn and X campaigns across extended sales cycles.
• API method or tools used: GA4 attribution path data and paid social reporting tools.
• The catch: Otto manages your website, SEO, and paid ads on Meta, Google, LinkedIn, X, and ChatGPT with your approval on every change that goes live or spends money; in Otto MCP, GA4 is read-only, LinkedIn Ads campaigns default to DRAFT with $0 budgets rejected, and paid ad write access requires the Master plan.

Claude prompt:

Query GA4 attribution path data for the last 60 days.
Filter paths containing touches from linkedin or x (twitter) paid campaigns.
Calculate the percentage of total key events where paid social appeared as an early touchpoint versus the final touchpoint.
Determine whether paid social spend is functioning primarily as a prospecting engine or a closing channel.
Provide recommendations for campaign budget allocation.

Prompt 14: GA4 purchase revenue vs. live Shopify orders

GA4 purchase events don't always match completed store checkouts, so Prompt 14 audits analytics revenue against live store orders. For deeper store and spreadsheet workflows, our COO at IslaIntel, Enrique Vásquez, writes operations-focused guides covering Shopify orders, inventory, and Google Sheets live KPI reporting.

• What it pulls and why it matters: Audits whether GA4 e-commerce tracking matches recorded store transactions.
• API method or tools used: GA4 runReport e-commerce metrics matched against Shopify read-only order endpoints.
• The catch: Under Google Analytics Data API specifications, GA4 runReport derives purchase statistics from tracking code events while Shopify records store transactions, and both integrations are read-only in Otto MCP.

Claude prompt:

Query GA4 purchase revenue and transaction count for each day over the last 14 days using runReport.
Metrics: purchaseRevenue, transactions
Next, pull actual completed gross sales and order counts for the same 14 days from our connected Shopify store.
Generate a daily comparison table showing:
- Shopify actual revenue vs. GA4 reported revenue
- Shopify order count vs. GA4 transactions
- Tracking capture percentage (GA4 revenue divided by Shopify revenue)
Flag any days where revenue capture dropped below normal thresholds and outline probable technical causes.

Prompt 15: Monday morning executive rollup checked against Google Sheets KPIs

Monday reporting shouldn't mean copying numbers across four tabs. Otto performs the weekly execution that marketing retainers typically cover—site updates, SEO maintenance, ad checks, and reporting—while strategy and final approval stay with you.

• What it pulls and why it matters: Compiles cross-channel marketing KPIs into an executive summary checked against team targets.
• API method or tools used: GA4 runReport, Search Console, Google Ads, and Google Sheets read-only tools alongside Meta Ads reporting tools.
• The catch: Google Sheets, GA4, Search Console, and Google Ads are all strictly read-only in Otto MCP, so Claude reads sheet targets rather than modifying spreadsheet cells.

Claude prompt:

Run an executive marketing review for the prior complete week (Monday through Sunday).
Pull:
1. GA4: Total sessions, engagement rate, total key events, and top three landing pages.
2. Google Search Console: Total organic clicks, impressions, and top three queries by click growth.
3. Meta Ads and Google Ads: Total combined spend, average CPC, and blended cost per acquisition.
Cross-check these metrics against our live KPI target rows in connected Google Sheets.
Format the summary as a clean executive briefing with bulleted highlights.

How to choose which GA4 prompt to run first

Select your starting prompt based on the challenge currently in front of you:

• If total conversions dropped this week, start with Prompt 1 to locate the declining channel, followed by Prompt 6 to see where users left the checkout flow.
• If paid advertising costs are increasing, run Prompt 12 to check for conversion attribution overlap between Meta and Google, followed by Prompt 8 to inspect landing page load speeds.
• If organic search traffic looks unaligned with revenue, run Prompt 2 to compare Search Console rankings against GA4 engaged sessions.

Google Analytics 4, Google Search Console, Google Ads, Google Business Profile, YouTube, Google Sheets, Shopify, and PageSpeed Insights are strictly read-only in Otto MCP, with no spend or configuration change capabilities. Under documentation from Google for Developers, Google's official Google Analytics MCP server is also read-only and cannot modify your Google Analytics settings.

Otto operates with strict human approval gates across website, SEO, and paid ad updates. Otto is the AI marketer on your team that manages your website, SEO, and paid ads on Meta, Google, LinkedIn, X, and ChatGPT with your approval on every change that goes live or spends money. Otto reads connected accounts, proposes what to change, and ships updates only after the owner approves: new ads stay paused, WordPress posts remain drafts, and site edits wait on a preview. For website updates, changes on GitHub open as pull requests with previews, and WordPress content stays a draft until published. Unlike WordPress (where new posts and pages default to draft), publishing to Facebook Pages and Instagram through Otto MCP goes live immediately and requires reviewer or admin permissions.

In Otto MCP, LinkedIn Ads campaigns and ad sets are created as DRAFT by default, activating or pausing requires confirm=true plus a reviewer or admin, and $0 budgets are rejected. Meta Ads write operations also require reviewer or admin access.

Pricing options reflect these capabilities across Otto Trial, Pro, and Master plans. Otto offers a $1 trial for 3 days (a trial of Pro: website updates and SEO, plus read-only access to ads and socials), a Pro plan at $199/mo or $99/mo billed annually, and a Master plan at $3,500/mo or $3,000/mo billed annually. Pro includes website updates, SEO, and read-only access to ads and socials, whereas Master adds write access to ads and socials with human approval on every change, and users can cancel anytime. Otto MCP uses the same Trial, Pro, and Master plans as the core Otto website product rather than a separate MCP subscription price. Write access to ads and socials requires the Master plan, and white-label agency portals are limited.