Case #044: Scaling past the winning audience

Case #044: Scaling past the winning audience

My ecommerce lead-gen campaign maxed out one tight audience at $120/day, 64% ROI. To grow I had to find new people — and 'broaden the targeting' usually means broaden the losses. Here's the audience-expansion playbook that grew the campaign without diluting it.

— Build the seed from converters, not clickers: I fed the lookalike a list of actual converted leads, not everyone who clicked. The source list quality decides everything downstream.

— Expand in defined steps: 1% lookalike first, prove it, then 2%, then 3%. Jumping straight to a broad 5% pool drowns the signal in cold traffic.

— New audience, new ad set, isolated budget: never let the new pool spend from the proven set's budget — you'll never know which audience earned what.

— Hold the creative constant while testing audience: I changed only the audience this round. Change creative and audience together and you can't attribute the result.

— Set an expansion floor: any new audience tier must clear 40% ROI in its first $300 or it's cut. The proven set stays untouched as the anchor.

— Layer interest-stacks last, not first: only after the lookalike tiers were mapped did I add interest refinements. Order matters — broad-then-narrow reads cleaner than narrow-then-broad.

The 1% pool cleared the floor at 51%. The 2% came in at 44%. Both scaled. Combined with the original audience: $2,820 spent, $4,420 back, 57% ROI — more than double the volume at a manageable ROI dip.

The lesson: you scale an audience the way you scale a budget — in steps, isolated, with a floor — because the second-best audience is profitable only if you never let it spend the winner's money.