# Run one useful SEO experiment

1. Choose one existing page and the actual search questions it answers. Save the original text and a dated Search Console export.
2. Write a hypothesis before editing. Change a small, explainable part of the page. Record its URL and actual release date.
3. Compare 28 complete days before and after the release using the same filters and reporting timezone. Allow for reporting delays. New pages have no equivalent before period.
4. Report clicks alongside impressions, position and query mix. A changed ranking or audience can change CTR without your wording being the cause.
5. Count resource clicks separately from downloads completed, inquiries, qualified leads and sales. A click is not a conversion.
6. Keep unknown values empty. Retain failures and flat results. Describe concurrent changes and the limits of a before/after comparison.

The CSV contains a clearly labeled example row, not Business Boomer performance data. No account connection, background automation or paid subscription is included.

Source: https://support.google.com/webmasters/answer/7576553