GEO Testing

GEO Testing measures how SEO and content changes affect AI visibility. Run a structured experiment, then compare citations, cited URLs, prompts, models, and AI traffic. Available on Scale and Enterprise plans.

What it does

• Runs time-based tests that compare the same URL group's performance before and after a change date
• Runs split tests that compare a test group against a control group for more reliable results
• Tracks citations (how often your URLs appear as AI sources) and unique cited URLs per group
• Provides 6 analysis tabs: Overview, By AI Model, By Prompt, Top Sources, AI Traffic, and AI Analysis
• Generates AI-powered analysis that interprets your test results and provides actionable insights
• Automatically creates annotations on change dates for visual correlation on charts

How to use it

1. Navigate to GEO Testing in the sidebar (under Optimization)
2. Create URL Groups first, named collections of URLs you want to track as a cohort
3. Add URLs to each group (paste one per line, up to 100 at a time)
4. Click New Test and choose your test type:
   • Time-based: select one URL group and a change date. LLM Pulse compares the period before the change with the period after it. Use periods of similar length
   • Split test: select a test group and a similar control group. Summary results compare both groups from the change date onward. The earlier observation period remains visible on the chart for context
5. Monitor results across the 6 tabs as data accumulates
6. Use the AI Analysis tab to get an AI-generated interpretation of your results

URL Groups

URL Groups are the foundation of GEO Testing. A URL group is simply a named collection of URLs that you want to track together.

• Give each group a name, optional description, and optional color for chart identification
• LLM Pulse automatically normalizes URLs (removes tracking parameters, trailing slashes) and deduplicates
• For split tests, make your test and control groups as similar as possible (same template, similar traffic levels)
• You can create as many groups as you need

Tips & notes

• Available on Scale and Enterprise plans
• Time-based tests are best for: title tag changes, meta description updates, content rewrites, schema markup additions
• Split tests are best for: large-scale template changes, new content strategies, structural site changes
• Annotations are created automatically when you set a change date, making it easy to correlate changes with performance shifts
• The AI Analysis feature uses your actual test data to generate insights, it's not generic advice
• For weekly prompt tracking, aim for four complete weekly runs before and four after a time-based change. If that is not possible, use at least two complete runs on each side
• Keep time-based periods the same length. Avoid a short after period against months of before data
• For split tests, judge the control and test groups over the same post-change dates. Use the observation period to check that the groups behaved similarly before the change