"Broad targeting always wins now, let the algorithm decide" — not on cold offers

"Broad targeting always wins now, let the algorithm decide" — not on cold offers

The modern gospel: detailed targeting is dead, go broad, trust the machine. True sometimes. It cost me real money assuming it was always true.

The setup
Brand-new health offer, zero pixel history, broad targeting per the advice.

The move
After $1,400 of broad spend with no clear winner emerging, I switched to tight interest stacks to seed the pixel.

The numbers
Broad with a cold pixel: 1.3x, because the algorithm had no conversion signal to optimize toward — it was guessing. Once I ran narrow interest targeting and fed it ~80 clean conversions, then opened to broad, ROAS jumped to 2.7x. The machine needed examples before it could think.

The lesson
Broad targeting wins when the pixel already knows your buyer. On a virgin account it's just paying to teach the algorithm slowly and expensively.

What I'd do differently
Seed narrow, scale broad. Get 50-100 conversions into the pixel with tight targeting, THEN go broad. "Broad always wins" skips the seeding step that makes broad work. Figures illustrative; the cold-pixel penalty is real.