Landing page optimization is usually either a redesign project that takes three months or a gut-feel copy swap that doesn't move the needle. Both miss the real problem: without a framework, you're testing randomly instead of systematically, and test results accumulate without building toward anything. This stack gives you a structured middle path: pull actual GA4 conversion data and Hotjar behavioral signals, feed them into Claude alongside your current page copy, and get a ranked hypothesis backlog with rewritten variants ready to test — without a full redesign cycle and without guessing what to test next.
The Stack
The Prompt
This stack is built around the Ad Creative Audit Prompt. Here's the abbreviated version — the full prompt with all variables and usage notes is on its own page.
You are a B2B conversion rate specialist auditing landing page performance. Review the GA4 conversion data, Hotjar behavioral signals, and current page copy below. For each page: identify the primary conversion bottleneck (headline, CTA, social proof, page length, or load/form friction). Generate a rewritten headline, subheadline, and CTA for the lowest-performing element. Rank your 5 highest-impact test hypotheses by expected conversion lift, with test structure and success metric for each.[ ... continued — see full prompt ]
The Workflow
-
Export GA4 landing page performance data
Pull conversion rate, bounce rate, scroll depth, and device breakdown for your top 5 landing pages. Include traffic source breakdown — a page converting poorly on mobile from paid traffic needs a different fix than one failing on desktop.
-
Export Hotjar behavioral data
Pull heatmap summaries and session recording insights for the same pages. Look specifically for where attention drops off, which CTAs are being ignored, and where form friction is causing abandonment.
-
Copy the current page copy for each page
Capture the headline, subheadline, body copy, social proof elements, and CTA text. This is what Claude will rewrite — paste it as the 'current state' alongside the performance data.
-
Run through the Ad Creative Audit prompt
Paste GA4 data, Hotjar insights, and current copy together with landing page optimization framing. Claude produces a ranked hypothesis backlog with specific rewritten variants for the highest-impact tests.
-
Build the A/B test calendar in Notion
Export Claude's output to Notion as a structured test backlog. Assign each test an owner, a start date, and a success metric. Run one test per page at a time so results stay clean.
What This Replaces
- Landing page redesigns that take three months and reset your data baseline when they launch
- Gut-feel copy changes that produce inconclusive results because there was no clear hypothesis
- Test backlogs that drain to empty because nobody has a systematic way to generate new hypotheses
Related Stacks
New stacks drop weekly.
Each one includes the tools, the Claude prompt, and the workflow logic. Free — built for in-house B2B demand gen managers.