Test 3 distinct angles
Outcome-focused, fear-of-loss, status/identity. Three fundamentally different angles. If one wins, you know which resonates.
Control is the current headline
Always test against current performer, not against a straw man. Otherwise 'wins' are fake.
Document learnings explicitly
Not just 'variant B won'. 'Status-framed headlines outperform outcome-framed for this audience'. That's transferable.