Hey STM
These questions maybe completely NOOBIE-ish by nature, so apologies in advance if this has been already answered somewhere else. If so, I would a link to that thread.
1) Global CDN
It looks like different CDN providers have different speeds in diff parts of the world. If you are running multiple geos, do you sign up for multiple CDNs or is 100ms latency acceptable? I've noticed rackspace, while awesome in tier1 countries, kinda sucks in tier 2/3 countries. (And this data is constantly changing... as it depends on internet traffic trends).
Random idea for a tool: Wouldn't it be great if a tool automatically re-routed the traffic to the fastest loading CDN? I was thinking I can make a tool that measures latency times for all cdns around the globe and how fast they are in each of the cities/countries. Than "copy" the CDN content to the fastest CDN provider. Of course, this would have to be done real time.
2) Offer cap reached
If you get capped out.. how do you keep a campaign alive? My guess is.. have "backup" offers in the order of highest to lowest ROI and route them when your offer gets capped.
Random idea for a tool: Do this automatically by accessing CPA networks' API
3) Bot Detection
Maximize ROI by automatically cutting out bot ridden traffic sources / campaign spys
[edit] This thread is here
4) Keep campaign lasting longer
This isn't a well thought out question... but what if there was a way to somehow keep the campaigns going longer. Have banner / LPs "queued", and an API automatically follows your "optimization algorithm". So you have your creative people churn stuff out... and let the automation tool fight banner blindness or what not.
Maybe this is an overkill?
Would appreciate any thoughts
Good ideas all!
1) There's actually already a tool that measures those speeds - see http://stmforum.com/forum/showthread...ty-data-Part-2 . The automatic switching would be interesting, but generally you're only running a campaign in a single country anyway. Not always, but usually.
2) This is correct - but usually you only need one or two backup offers, and you'll carefully pick those by hand. There's no way within the network's API to tell what the best offer is, and you're far better to simply pause the campaign than to run a bad offer.
4) Optimisation automation is definitely something that quite a few people have tested out. Simple rules can be very useful, a la POFpro's automated hurdles. More complex automated optimisation turns out to be quite a hard problem - I'll be touching on it in my talk in London.