Case: an H1 template A/B test on 3,000 pages moved average position 4.1 spots

Case: an H1 template A/B test on 3,000 pages moved average position 4.1 spots

Two H1 formats tested at template scale. Setup:

— Step 1 (analyst): split 6,000 matched pages into two 3,000 cohorts by hash. Gate: cohorts matched on baseline position +/- 0.3.
— Step 2 (dev): Cohort A H1 = '{Product} Review'. Cohort B H1 = '{Product} Review ({Year}): {KeyAttribute}'.
— Step 3 (analyst): held 6 weeks, no other changes. Gate: no template edits during the window.
— Step 4 (analyst): compared average position delta between cohorts.

Outcome: Cohort A moved +0.4. Cohort B moved +4.5 — a 4.1-spot edge from the attribute-rich H1. Rolled B to all pages.

Lesson: at scale, one template variable is worth thousands of manual title edits.

Guardrail: change exactly one variable per test, or you can't attribute the win.

Ship gate: don't publish until all boxes are checked.