Home > General > Affiliate Marketing Forum

Running Mobile? Boost ROI with Granular Offer Testing! - Part 2 (10)

04-07-2016 02:04 AM #1 vortex (Senior Moderator)

Image fixed - sorry guys!

Amy

04-19-2016 03:40 PM #2 dandyman (AMC Alumnus)

Thanks Amy for the great tutorial.

2 questions though..

1) When testing an offer (AV pin submit), if the offer allows both carrier and Wi-fi traffic, how is it better to set up the initial test campaign? One campaign for both carrier and Wi-fi traffic or 2 campaigns, 1 for carrier traffic and 1 for Wi-fi traffic?

2) For mobile antivirus offers, does it make sense to test pop unders, or in general, pop ups always work best with mobile AV?

Thanks.

04-21-2016 03:47 PM #3 vortex (Senior Moderator)

Originally Posted by dandyman

The initial test campaigns are to tell you which carriers you can target exclusively, and which ones you can't, so that you can make better decisions when you're optimizing your campaign.

Example: Initial testing tells you that CarrierX can be targeted by setting "connection type" and "connection speed" to carrier/cellular. But for CarrierY, even when you set to "carrier/cellular" you'll still get a lot of wifi traffic with the carrier traffic. In this case, if your offer converts well for CarrierX, AND if CarrierX has good volume, then you can expect to get that camp profitable. But for CarrierY, say your current offer converts well for the carrier traffic, but not the wifi traffic that's "tagging along", then you know NOT to continue to optimize for this carrier - not unless you can find another offer that converts well enough for the wifi traffic. (BTW what you can do in this case, is take this info, go to traffic sources that allow you to target by carrier, and target CarrierY exclusively for your offer.)

Although, if you've already decided to run in a specific geo, you don't even need to run the initial CAR/NOCAR tests. You can simply set up your camps that way, i.e. one targeting carrier/cellular and another targeting everything else. Then you can just cut offers/ISPs separately for each camp based on stats.

The initial CAR / NOCAR tests are more useful if you're wanting to run offers that convert especially well on carrier traffic (i.e. some carrier-billing offers) and trying to decide which geos to focus on. So say you do this testing on the top 20 geos in popads (with the most traffic volume), identify the geos that have the most amounts of carrier traffic and then mass-test offers in those geos, that's where the gold will be.

As for testing popunders vs. popups for antivirus: I never thought about that too much. I just start testing using popunders because that's where most of the volume is usually, and once I get profitable I test popups and tabups/unders. Are you seeing some interesting trends related to that?

Amy

05-04-2016 12:51 AM #4 bradh_ ()

Originally Posted by vortex

I compiled a list of offers for MX, and the carriers and OSs each offer will accept.

When coming up with this list, are you choosing offers that are all Mobile Content and of the same type of content

EG
Offer 1 - Mobile Content - Antivirus
Offer 2 - Mobile Content - Antivirus
Offer 3 - Mobile Content - Antivirus
Offer 4 - Mobile Content - Antivirus
Offer 5 - Mobile Content - Antivirus
Offer 6 - Mobile Content - Antivirus

Or is it possible to test different types of mobile content offers against each other?

EG
Offer 1 - Mobile Content - Win iPhone6
Offer 2 - Mobile Content - Antivirus
Offer 3 - Mobile Content - Download Now
Offer 4 - Mobile Content - Whatsapp Wallpaper
Offer 5 - Mobile Content - Antivirus
Offer 6 - Mobile Content - Cute Girl Wallpaper

I previously assumed it's the first, as I can't imagine a lander that would work across offers that are all somewhat different, however I noticed that you have some mainstream offers and some adult offers. So based on that, it would seem that the latter is possible?

05-05-2016 06:12 AM #5 vortex (Senior Moderator)

Originally Posted by bradh_

Great question!

You're right - you won't want to use the same landers for multiple types of content, but if you're wanting to see which offers are converting well for a certain geo, you can certainly implement your scenario 2 above.

What you can do is group similar offers together and use a separate set of landers for each group, then set up a separate path in your tracker for each group, e.g.:

Path 1 - Whatsapp wallpaper offers - Whatsapp wallpaper landers
Path 2 - Antivirus offers - Antivirus landers

And so on.

Although, it would probably be more efficient use of time to focus on one vertical, and test multiple geos for that vertical. That way you'll know which landers work (although this can be different across geos, you'll most likely see certain lander elements that will work well across multiple geos). Plus you can outsource your lander translations for several geos at a time for example. Imagine having to do this for 6 different verticals per geo - could turn into a nightmare unless you have employees doing some of the lander preparation for you.

There's also an in-between scenario: Different offer types within the same vertical. Sweeps for example:

Path 1 - iphone 6 offers - iphone 6 landers
Path 2 - iphone 6s offers - iphone 6s landers
Path 3 - samsung s5 offers - samsung s5 landers
Path 4 - samsung s6 offers - samsung s6 landers

You get the idea. For something like this, you can actually use the same landers for each group - by simply changing the "prizes" and "testimonials" to suit each type of offer.

Amy

05-25-2016 05:04 AM #6 hangman (Member)

Regarding the "Proven lander" you mentioned above. Did you mean that it is a lander that you have been testing and know it is converting well or just rip the common used landers from spy tools and test it using the method on this thread/

05-25-2016 08:28 PM #7 vortex (Senior Moderator)

Originally Posted by hangman

Yes I mean a lander you've experienced success with.

If you don't have such a lander you can just rip a bunch of landers, run 2-3 best offers from AMs for a given geo+vertical and use the winning lander to mass-test offers. This strategy is basically what I use for all pop camps, carrier or wifi, adult or mainstream.

It really depends on how many offers you're testing as well. But if you're testing large numbers of offers, then you'd definitely want to use a proven lander, or it will get unnecessarily expensive to test all these offers.

Amy

07-23-2016 11:59 AM #8 koutrouss (Member)

Hi Amy,

What about running offers for a specefic geo with multipule vertical ? since when asing my AM for 2-3 proven offers running now and has high CR, I think there will not be much proven offers for a specific geo+vertical, that will be hard to find.

Also I'm still confused a little bit
When you are testing, how much the total budget for the initial test per offer, and how many rip landing pages should be use in test to find the winner one/ones?

07-24-2016 01:18 AM #9 vortex (Senior Moderator)

Originally Posted by koutrouss

When it comes to "mass-testing" offers, you won't JUST include AM-recommended offers - you'll want to test as many offers you can find on as many networks as is reasonably possible.

As for budget per offer: You don't need to set such a budget. Basically you're testing offers for each carrier to find the best offer. So you would keep cutting inferior offers until the winner emerges. Exception: If none of the offers look like they could be profitable for any major traffic segment (e.g. Android/IOS) - in which case you'd want to stop the camp until you have a new batch of offers to test. Here's the best tool I know of for comparing and cutting offers:

http://stmforum.com/forum/showthread...ou-pick-offers

To find a winner lander that can then be used to mass-test offers, I would suggest to rip and test 5-10. Obviously the more you test the better chance you have of finding a good lander.

As for running multiple verticals per geo - I would test EACH separately, because different verticals will require different landers, and running multiple verticals in a single camp can make tracker stats unnecessarily messy. If you want to cast a wider net, consider testing multiple GEOS for a given vertical instead. That way you know what kinds of landers are being used in this vertical - it would just be a matter of ripping the same landers in different languages. (Alternatively you can rip landers in one language and have them translated into other languages.)

Amy

07-24-2016 03:51 PM #10 koutrouss (Member)

Originally Posted by vortex

Ok, thanks for the additianal infos.
I have adplexity tool, I'm gonna pick 5-10 landers there, copy & paste them. and test them in one camp using high BID, since testing that way gonna give us high quality traffic and better results to take decisitions in them.
my question is about when to stop the the test, since runing the camp for monday will have different results of thuerday, I think the test should be at least for 3 days, with a minumum budget at least 10 dollar targeting high quality traffic only, that should give us better stats, and we could make better decisions.

Also for people who get them traffic from RTB like Avazu.. I always test there with low BID (half of average "CPC") with a daily budget 10 usd for each traffic segment (banner & Native ads) for 3 days, once I find a profitable site/app, then I BID high for those profitable sites/apps only.
After the 3 days, If none of the offers look like they could be profitable, then I stop the camp.

What do you think in that way?
Does I should test there seperatly for each traffic segment (Android/IOS) with 2 camps instead of 1 for each plan (Banner & Native Ads)?

Home > General > Affiliate Marketing Forum