What does this tool do?
It splits URLs into two statistically similar groups for SEO A/B testing: a variant group and a control group.
SEO A/B Page Splitter
Create statistically comparable variant and control groups for SEO A/B testing, with automatic outlier removal, quality scoring, and visual trend validation.
- Upload data
- Run split
- Review groups
Input data
Upload a CSV where the first column is URL and all following columns are daily traffic values.
Q&A
What input format is required?
Upload a CSV where the first column is URL and all following columns are day-by-day traffic values for each URL.
How much history is recommended?
At least three months of daily history gives more stable splits and stronger comparability between groups.
How are outliers handled?
The tool calculates outlier scores on total clicks and average traffic per URL, then excludes anomalies before splitting.
Can I limit the number of URLs?
Yes. Use “URLs to include” to sample from the cleaned URL pool. If empty, all available cleaned URLs are used.
How is the split generated?
Primary method is stratified splitting on average traffic bins. If data variation is too low, it falls back to deterministic alternating split.
What should I check in the metrics?
Check variant/control counts, excluded URLs, and traffic balance. A smaller imbalance generally means cleaner test design.
What is the Split Quality score?
Split Quality combines level balance, trend alignment, volatility/seasonality similarity, and robustness checks into a readiness score for test setup.
How should I use Ready / Caution / Not ready?
Use it as a launch signal: Ready means strong baseline comparability, Caution means usable but weaker sensitivity, and Not ready means improve split quality first.
How do I read the visual representation?
The two lines should follow similar trend shapes over time. Better overlap suggests a stronger baseline comparability.
What does Normalize do?
It converts both lines to index scale (base 100) so you can compare trend shape without absolute volume bias.
What if URL formatting is invalid?
The tool warns but continues. Fix invalid URLs for cleaner exports and fewer downstream mistakes in implementation.
What if traffic columns contain non-numeric values?
The tool attempts conversion and sets invalid cells to 0, then reports a notice so you can review data quality.
What should I do after splitting?
Implement the SEO change on variant pages only, run the test, then analyze impact in the integrated SEO A/B Test Analyzer.