Proving an SEO Change Worked With Split Tests, Not Before-and-After Charts

How Do You Prove an SEO Change Worked?

Search Engine Journal's piece argues that before-and-after charts cannot show whether an SEO change caused a result, and sets out how page-level split testing builds a fair comparison. The author, writing from the agency side, explains that traffic moves for many reasons, including core updates, seasonality and AI Overviews, so the real question is whether results beat what would have happened anyway. The article walks through matched control and variant groups, a forecast built from the control pages, and a pre-launch check to test the model before trusting it.

For brand builders, the useful part is the reporting discipline. The piece shows how to set a metric and a win threshold before launch, why conversion should act as a guardrail, and how to read intervals and probabilities without overstating a result. Anyone who answers to a finance team, or who is asked to justify a site change, should read it. Teams with smaller sites will also find sensible fallbacks for pre and post comparisons.