<b>The smartlink that quietly subsidized my worst geo</b>
<i>The setup.</i> Mainstream pop traffic into a dating smartlink, 14 geos rotating. Illustrative numbers, real mistake.
<i>The move.</i> Blended ROI sat at a comfortable 1.22 for a month. I scaled spend 3x because the top-line looked clean. Spend went from $90/day to $270/day.
<i>The numbers.</i> At 3x, blended ROI collapsed to 0.84. When I finally broke it out by geo: three tier-1 geos ran 2.1 ROI and were capped on volume. Eleven tier-3 geos ran 0.4. The smartlink averaged them into a pretty 1.22 that hid the fact 78% of incremental spend went where I lost money.
<i>The lesson.</i> A smartlink's blended number is a comfort blanket. It can stay profitable on top-line while every dollar of NEW spend goes to the losing tail, because the winners are volume-capped and the losers aren't.
<i>What I'd do differently.</i> Never scale on a blended smartlink ROI. Break out per-geo, find which segments are capped, and scale only those — manually, into direct offers if needed. The blended view is for reporting, not for decisions.
Arb Files
@ArbFiles