CRO With Full Data Context in Claude

The prompts in this post assume you have your MCP stack running — GA4 for conversion and behavioural data, Google Search Console for the demand side, FreeCrawl for technical context, and Microsoft Clarity for heatmaps and session recordings. If you haven’t set up Clarity yet, Part 5 covers the setup including the multiple-project workaround. Some prompts also use Google Tag Manager (Part 6), Google Business Profile (Part 7) and an A/B testing platform such as Convert or Unbounce (Part 8).

What makes these prompts different from typical CRO advice is the data depth. Rather than asking Claude to reason from general principles, you’re asking it to reason from your actual user behaviour — the heatmaps on your specific pages, the session recordings of your actual visitors, the conversion funnels built on your real event data.

Funnel Analysis Prompts

Start here to understand where you’re losing people.

“Using GA4, pull my conversion funnel from [landing page event] through to [conversion event] for the past 30 days. Show me the drop-off rate at each step. Then cross-reference the step with the highest drop-off with Clarity heatmap data — what are users clicking on, where is their attention going, and what does the scroll depth look like on that page? Based on all of this, give me three specific hypotheses for why users are dropping off and what to test.”

“From GA4, compare conversion rates for users who arrive via organic search versus paid search versus direct traffic on [target page]. Is there a meaningful difference? If organic converts significantly lower, pull the top organic queries driving traffic to this page from Search Console — are users arriving with a different intent than what the page is optimised for?”

Heatmap & Behaviour Analysis Prompts

“From Clarity, what are the most-clicked elements on [page URL]? Now tell me: which of those elements are CTAs that lead toward conversion, and which are links that take users away from the conversion path? If users are clicking on exit paths more than conversion CTAs, that’s a page layout problem — what specific changes would you recommend?”

“Find pages in Clarity with a high dead-click rate — places where users click but nothing happens. Cross-reference these with FreeCrawl to understand the page structure and GA4 to see the conversion rate on those pages. Dead clicks on non-interactive elements usually signal user confusion about what they’re supposed to do. What are users trying to click on, and how should the page be redesigned to meet that expectation?”

“Look at the quick-back rate from Clarity for my top five organic landing pages. A high quick-back rate means users are returning to Google immediately after landing — classic expectation mismatch. Cross-reference with the Search Console queries bringing people to each page. Is there a disconnect between the search intent of those queries and what the page delivers? Give me specific recommendations for each page.”

Tracking Checks Before You Test

An experiment is only as reliable as its tracking. With Google Tag Manager connected, Claude can confirm the measurement is in place before you spend any traffic on a test.

“I’m planning a test on [page] with [goal, e.g. quote form submissions] as the primary metric. Check Google Tag Manager for the tag and trigger that track this goal, and check GA4 to confirm the event has been recording on this page for the last 14 days. If anything is missing or misfiring, tell me exactly what to add or fix in GTM before the test goes live.”

A/B Test Hypothesis Generation Prompts

“Based on GA4 conversion data, Clarity heatmaps, and FreeCrawl page analysis for [page URL], generate five A/B test hypotheses ranked by potential impact and ease of implementation. For each hypothesis, state: the observation that prompted it, what you’d change, what you’d measure, and what result would confirm the hypothesis. Format this as a testing brief I can hand to a developer.”

“I want to run a CTA test on [page]. From Clarity, tell me where users’ attention is currently focused on the page (based on heatmap click density). From GA4, tell me what percentage of visitors click the existing CTA. Based on both, suggest three alternative CTA placements and copy variants that are likely to outperform the current version — and explain your reasoning for each.”

A/B Testing Platform Prompts: Convert and Unbounce

With your testing platform connected, Claude can read experiment results directly and combine them with everything else in your stack. These examples use Convert and Unbounce, but similar prompts work with other platforms that have an MCP server.

“Pull the results for my Convert experiment on [page]. Break the results down by device and traffic channel. Then use GA4 to check what users in each variant did after converting, and Clarity to compare behaviour between control and variant sessions. Explain why the variant won or lost, and whether the result holds across segments or is driven by one group.”

“List all active experiments in Convert, with their target pages, audiences and start dates. Flag any that run on the same pages or overlapping audiences and could interfere with each other’s results, and suggest which to pause or reschedule.”

“Review the Clarity heatmaps and scroll depth for my Unbounce landing page [page name]. Based on what you find, create a new variant in Unbounce that addresses the biggest problem, for example moving the form higher or adding trust signals near the CTA. Leave the page unpublished and summarise every change so I can review it.”

Mobile Optimisation Prompts

“Compare mobile versus desktop conversion rates from GA4 for my top 10 pages by organic traffic. For any page where mobile converts at less than 60% of the desktop rate, pull the FreeCrawl mobile usability data and Clarity mobile scroll depth. What are the common patterns across underperforming mobile pages, and what are the three highest-impact fixes?”

Local Conversion Prompts

For local businesses, many conversions never touch the website: people call or ask for directions straight from the Business Profile listing.

“For each Google Business Profile location, show calls, direction requests and website clicks over the last 90 days. Compare them with GA4 conversions on the matching location page. Which locations convert well from the listing but poorly on the website, and what does Clarity show on those location pages?”

Post-Purchase & Retention Prompts

“From GA4, what percentage of converting users return for a second session within 30 days? Segment by acquisition channel — which channels bring the highest-value repeat visitors? Cross-reference with the pages those returning users visit most — are there content patterns on my site that correlate with retention? What should I be doing more of?”

Combining SEO and CRO Insights

The most powerful analyses come from treating SEO and CRO as one connected system rather than two separate disciplines. A page that ranks well but converts poorly is a missed opportunity on both fronts — and fixing the CRO issue often also improves the engagement signals that matter to Google.

“Give me a combined SEO/CRO priority list for my site. Use Search Console to find pages with strong organic impressions, GA4 to find the conversion rate on those pages, and Clarity to identify any obvious UX issues. Rank the opportunities by: (impressions × gap between actual and benchmark conversion rate). These are my highest-leverage pages to work on.”

The automations to run these analyses on a schedule — so you’re always working from the latest data — are covered in Part 13.