Semrush vs SERP Split

Semrush has no write-up on Specifiction yet; its figures are above.

SERP Split is a free web tool for SEO holdout experiments on similar pages. Users upload a CSV containing URLs and clicks to create test and control groups through stratified sampling. They then upload daily traffic data with date, test_traffic, and control_traffic columns for analysis. The site says uplift and confidence are measured with Bootstrap Casual Inference. Results preview uplift, absolute clicks, a p-value, significance, and an exportable graph. Test examples include meta titles, H1s, internal links, templated copy, AI content, FAQs, schema, and performance changes. The recommended test period is about 21–30 days, with one main variable isolated to the test group. Guidance suggests 100,000 or more clicks across pages over roughly the previous 100 days, while lower volume may reduce confidence. The listed search data source is Search Console, and organic traffic impact is the focus. Tool data is processed locally and temporarily, not stored or transmitted for storage. Personal data is not collected, stored, or processed, though limited non-personal cookies measure aggregated usage.