Outbrain - geo inconsistencies when scaling (6)
has anyone else seen this ? we have an ecomm offer doing well across 15+ geos (tier 1 and 2) in taboola and just started scaling to outbrain a couple weeks ago.
in taboola, scaling from one geo to more was pretty seamless. once smart bid was dialing in eCPAs we wanted in one geo, it hit similar results everywhere else pretty quickly. in only 1-2 new geos did we have to pause completely (i suspect more due to poorly translated ads/landers)
in outbrain , it's a different story entirely, atleast so far. from the proven geos we are trying to move over, maybe 20% are instant home runs ($0.80+ epc) , 30% need major optimizations or revamp ($0.20-$0.30 epc) , and a full 50% of geos barely convert at all (sub $0.10 epc and sometimes no conversions).
we’ve tried all types of bidding from semi auto to target CPA, device/region/bid optimization etc, new ads/landers and even different products, and it just seems like a few geos don’t “work” for us on outbrain and that’s that.
was wondering if anyone else experienced this , and if so maybe we can brainstorm in this thread or open to PMs as well. cheers
As these are 2 separate sources with different inventory, there will definitely be significant differences in the performance too. The algo can play a role too, each of the networks works in a bit different way too.
But anyways, let me tag some people who are experienced with native traffic, they might have some info to share : @platinum @thedudeabides @jack_l
as far as i can tell this is an outbrain specific issue though. we’re running other native channels and even fb and nowhere else does the performance fluctuate based on geo much
How long have you had your Outbrain campaigns running and with what optimization approach? Based on what metrics are you turning off sections, publishers or ads? Any additional details would help with suggestions.
Generally speaking speaking there's a noticeable difference between Taboola and Outbrain when it comes to the optimization approach.
I think the best approach would be to go for semi-automatic, as that is the optimization mode most affiliates see success with. Then once, you've gathered enough data from Outbrain's pixel, you may try target CPA or Fully automatic.
Sometimes on Outbrain you may need to be really aggressive when it comes to blocking poor performing sections. Not publishers, but sections, since there is where most of the optimization is being done. Just keep in mind that by default you can have up to 30 publishers and 100 sections to block.
I guess you may have already seen lots of low reputation sites (or mobile apps) that have awful results, so it could be worth starting an aggressive blocking approach with them, then optimize tweaking bids on other publishers. Also, once you hit the default limit, you can ask Outbrain support to increase your blocking limit. An easy way out with them could be mentioning the results you're having on Taboola and that you're trying to reach the same with their traffic. You can even get the max limit they have 
We run a ton of INTL senior dating in a bunch of geo's on both Taboola and Outbrain, and it's definitely interesting seeing the weird patterns in how stuff performs.
I don't know what accounts for the idiosyncrasies, but they are definitely there. For instance if we run a big campaign with 20 geo's in English on Taboola Japan does great - but if we try to run in Japan by itself in Japanese it does terrible.
That could be the language issue (Japanese folks who speak English are more likely to convert) but we even see the same thing within English-speaking countries. Some do great as part of big INTL camps but badly on their own.
A lot of it just seems random, to be honest.
The only helpful info I might be able to give you is that in our experience, Taboola works well both targeting individual geo's and when combining lots of geo's together, however Outbrain does MUCH better when lots of geo's are combined, and significantly worse (by comparison) when targeting individual geo's. I think their algorithm is just much more sensitive and for whatever reason combining lots of geo's just gets you cheaper/better clicks.
So yeah, both in dating and in ecom and vsl's and any other INTL stuff we've run, 99% of our success on Outbrain comes from big camps with 10-20 geo's targeted at once. Even if some of the geo's within that group aren't profitable by themselves, it seems like it's better because if we start taking out the bad ones then the whole campaign dips in performance... it's weird...
And then yeah, we've been using Semi-Optimized lately on Outbrain, but we've tried Fully-Optimized with some success and also I've heard of people having a lot of success on tCPA.
Overall though we do much better on Taboola, so it seems your results are matching ours.
Edit: 2X on @platinum's excellent point about blocking the bad sections too - Outbrain does seem to have about 5-10 bad (0% lp ctr) sections that need blocked each time you start a new run of network camp. Usually it's not a big enough deal that it really changes whether and offer or geo will work, but it will definitely help your ROI if you block those (I like to block them on an account level when I see them so you don't have them pop up again in another campaign).
Great advice from both platinum and jack_l !! I knew I could count on you guys 
Thanks!
Home > Paid Traffic Sources > Native