Running Mobile? Boost ROI with Granular Offer Testing! - Part 4 (8)
Running Mobile? Boost ROI with Granular Offer Testing! - Part 4
Find Part 1 Here...
************************************
STEP 4: POSSIBLE NEXT STEPS:
************************************
I know you'd probably be more excited if I could run the above camps to profits before showing the results, but this post was meant to be a guide only and not a case study (which I have plans to do in the very near future), and I wanted to finish this post so I could spend some time replying to other posts. 
From here on, basically my next steps will be:
1)Continue running the promising camps and cutting offers.
2)When a camp is down to the best offer/offers, I'll clone that camp to test different bids to see which one will be the most profitable.
3)This would also be a good time to scale the camp to other traffic sources - because I know what the best offer is for the traffic segment.
4)I'll cut stuff - placements, devices, OSs, browsers. If a certain device/OS converts extremely badly I may also cut them even while testing offers. Usually I leave dayparting for last. If ROI is high I'll play with frequency too to see if I could increase overall profits.
************************************
GOLD NUGGETS & RECAP
************************************
-Different offers will perform VERY differently for wifi vs. each specific carrier. By testing offers for each separately, you can maximize your ROI for every traffic segment.
-Carrier traffic can often convert a LOT better than wifi traffic for carrier-billing type offers. By doing initial testing to find out which geos on which TSs have lots of carrier traffic, and testing offers for them, you'd be going after the gold!
I used Mexico as my example, which doesn't really show you the potential profits you can make with this method, because this geo consists mainly of wifi traffic, and I'm not about to give away my best geos, i.e. the ones with enough carrier traffic to capitalize on.
However, you can find this out easily by testing ALL the major geos using the "initial tests" I've described above - and you're not limited to PopAds either. Which brings me to my next point...
-You can use this approach with other traffic sources, not just PopAds. Other traffic sources may not give you a traffic estimate like popads does, but you can always run a bit of traffic and approximate to get an estimate. For example, target carrier traffic only and set a reasonably high bid, run for example $10 of traffic to it, drill down to "dayparting" data to see how long it took the $10 to run out. Let's say you got 300 impressions during the 1st hour the campaign went live, then 450 impressions during the second hour and then 200 in the third hour, then your $10 budget ran out. You could then take the hour with the maximum impressions (450 in this case), drill down into mobile carrier stats to see what percentage of total carrier traffic came from each carrier, then calculate an approximate daily volume for each carrier.
Explanation: So you got 950 impressions in total, and the max impressions in an hour is 450 impressions so we'll assume that's the amount of all carrier traffic you can get per hour. Say you drill down into mobile carrier stats, and see the following break-down:
CarrierX - 400 impressions (400/900=42%)
CarrierY - 350 impressions (350/950=37%)
CarrierZ - 200 impressions (200/950=21%)
You would calculate the percentages as above, then apply these percentages to the 450 impressions/hour approximation, then multiply by 24 to get daily estimates:
CarrierX daily volume = 42% x 450 impressions x 24 hours = 4536 impressions
CarrierY daily volume = 37% x 450 impressions x 24 hours = 3996 impressions
CarrierZ daily volume = 21% x 450 impressions x 24 hours = 2268 impressions
These estimations won't be accurate because of traffic volume fluctuations throughout the day, but should give you a rough idea on which carriers have enough traffic to be worthy of testing offers for.
-By doing initial testing to find out which carriers have enough traffic, you can then look for and test offers that accept those carriers. So for example, if you know there's very little traffic available for CarrierX, then don't even bother testing offers that only accept traffic from that carrier! On the other hand, if you see a good amount of traffic available for CarrierY, then you can look for carrier-billing offers (for example) that accept CarrierY traffic to monetize that traffic. You can also ask AMs for offers that convert well for that specific carrier.
-You may be asking "since most of all mobile traffic is wifi, why bother testing offers separately for each carrier at all?" This goes back to my first note about how carrier traffic can convert significantly better than wifi traffic for the right offers. These are pockets of profit that will allow you to bid high for and make a high ROI on.
***********************************************
And that's it! I probably made it sound a lot more complicated than it is. Do this a couple times and you'll see how simple it actually is to do.
I wish you luck in finding the best-converting offers for all major traffic segments! 
Amy
So to be clear, does that mean that you actually run different offer on each diffetent carrier as well?
If so would you just stick to the same carrier and offer combo when scaling across TS or would you re test?
Originally Posted by sushiparlour
I'm a bit lazy - I just scale the same carrier+offer combo to other TSs. You can by all means retest the couple top-performing offers for a given carrier, but I wouldn't recommend retesting all offers for each traffic source - I highly doubt the extra costs would be worth any extra accuracy that can be achieved, although someone may prove me wrong.
Amy
Lovely tutorial Amy!
What I got from it is - Identify which GEOs have lots of carrier traffic, then if you have a proven lander go in and test lots of offers with a low payout in a granular way on both Wifi and Carrier traffic with separate campaigns for every promising segment. Asking your AM for offers that convert well for Carrier X is another great tip too, should save us some cash!
I think a common question/concern would be - what if we don't have an already proven lander? Would you do the same thing but with landers in rotation, or would that get the testing costs up too high?
I'll be testing this method out as soon as I can!
Originally Posted by shishev
If you don't already have a proven lander, then just default to a more general approach: Pick a vertical+geo, ask AMs for a few top offers, rip 5-10 most-seen landers and run. After a couple rounds of lander testing (heck even after the first round) you'll hopefully find a lander that works well.
You can for sure test landers at the same time you're testing carrier traffic! In that case though you wouldn't want to mass-test offers at the same time or else there would be too many lander+offer combos and you'll have to run a LOT of traffic to get statistically significant results. Also, know that you'll get less (often MUCH less) carrier traffic than wifi traffic, so it may be quicker to use wifi traffic to test landers first because lander performance normally won't differ that much for carrier vs. wifi traffic (don't quote me here since I don't have proof lol - I just feel that if there are huge differences I would have noticed it by now).
Thanks buddy for chiming in - appreciate the feedback and the questions!
Amy
Originally Posted by vortex
If you don't already have a proven lander, then just default to a more general approach: Pick a vertical+geo, ask AMs for a few top offers, rip 5-10 most-seen landers and run. After a couple rounds of lander testing (heck even after the first round) you'll hopefully find a lander that works well.
You can for sure test landers at the same time you're testing carrier traffic! In that case though you wouldn't want to mass-test offers at the same time or else there would be too many lander+offer combos and you'll have to run a LOT of traffic to get statistically significant results. Also, know that you'll get less (often MUCH less) carrier traffic than wifi traffic, so it may be quicker to use wifi traffic to test landers first because lander performance normally won't differ that much for carrier vs. wifi traffic (don't quote me here since I don't have proof lol - I just feel that if there are huge differences I would have noticed it by now).
Thanks buddy for chiming in - appreciate the feedback and the questions!
Amy
Great post Amy, you make this forum more valuable! Regarding carrier vs wifi traffic volumes, I find that that you can often ask the traffic source and they will give you stats for that, sometimes they'll even send you a detailed spreadsheet with stats for every geo
Originally Posted by exclusif
I wished every traffic source would do that for you - but I'm finding that very few are actually willing to. But you're right - when they do, it can be a massive time-saver! And it never hurts to ask!
There are lots of carrier-billing offers in the adult niche as well! Adult video subscription offers for example. Another tip: Many mainstream offers will also convert well on adult traffic - just make sure the advertiser does allow adult traffic.
As for landers performing differently for each carrier - I actually haven't done this type of comparison. What I HAVE observed is that the same offer will sometimes perform differently for different carriers.
Amy
Home > General > Affiliate Marketing Forum