RhinoFish - 1:04 am on Feb 21, 2013 (gmt 0)
eDub: we don't see #1 (client lineup mostly north america, plus almost always use "people in" versus looser choices) and believe [perhaps wrongly (but if it's a problem, it's tiny)] that the conversion optimizer saves us on #2 and #3. we also tend to work with small folks, where #3 wouldn't crop up often at all. and if #2 comes up, we'd negative geo-target vs IP block - again, our clients tend to be small and sell nationwide, so if the "stop" training doesn't work or a competitor is being a pain, we might exclude their city or zip code.
All our (situation specific) reasons aside, if I believed what 2kegx does, I'd appeal to G, it's very likely they are able to extend the count. Note, I didn't say likely they would, but likely could. Convince them there's fraud, and I believe they'll fix it.