Stackman's Guide On His Recent $X,XXX/day 2014 Mobile Campaign. (31)
Stackman's Guide On His Recent $X,XXX/day 2014 Mobile Campaign.
I’m going to explain why it doesn’t matter if you’re a pro or if you’re brand new, and how it all comes down to testing!
Traffic: Mobile
After Thailand i got back a little bit hungrier as i always do after work trips. Lately it’s been hungrier to start a mobile team, but i keep delaying that. So instead i decided to do some MASSIVE testing. I had a list of offers i wanted to test from before Thailand, then after Thailand i sifted through 50+ emails and added another 10 offers or so to the list. Now was time to TEST, this was my method:
1. I chose 1 traffic source i like and have had success with.
2. Took about 30 offers, made landing pages for about 25 of them, and some i made 2 landing pages because i wanted to test different angles. I didn’t spend much time with design, i focused on the copy, and the angle (the concept / selling point of the offer). My theory is IF the angle was bringing in a decent ROI then i can play around with designs.
(Tip: DON’T worry about making your designs perfect, thats step 2. First find a hot offer and landing page angle or banner angle!)
>> I ended up with about 50 URLs, for 30 offers. This took a week.
3. Spent WAY too much time making each campaign, and formatting each campaign to whatever the targeting requirements were. Some offers accepted android only etc..
4. Sifted through a ton of data, but the core idea was to find some of the hot winners. I didn’t have a rule of thumb. I just wanted to see which campaigns yielded the highest ROI out of the group. I ended up with 5 campaigns for 3 offers. 1 offer hit a positive ROI off the bat from the angles i chose.
>> NOW IT’S GAME ON!
5. I took the 5 winners, and made MORE angles for each campaign and tested again on the same traffic source.
6. The 1 offer with a few different angles was winning consistently!
7. I made some very nice clean (app style looking) landing pages for the offer, and some shitty pages as and then some weird styles as well. I think almost 10 total style pages around the winning angle.
ex: Same my offer was 'Candy Crush', my winning angle was possibly then 'Why girls always win at candy crush', then i had different style pages around this winning angle. Make sense?
I tested them thoroughly. Got a bunch of mixed results, but after a while 2 pages took the cake. I chose 1 based on gut.
>> NOW IT’S SCALE TIME!
8. I paused the other 2 offers and have those on the backburner. Now it’s time to scale! Moved it to 4 traffic sources at once. I wanted to make sure it wasn’t a 1 hit wonder.
It wasn’t. 3/4 had similar ROI 
9 . Now i must be signedup with 30+ mobile traffic sources, so i spent ALL of today adding the campaigns to these sources and adding funds. Now what i’m doing is signing up to another 30 sources. I’ll probably be doing this all night. It’s beyond annoying.
10. Next step is a mix of getting the campaign live and active on the 30 sources i’m already on + the 30 new sources. It’s going to be a busy 15 days. I'm realistic and know probably this campaign will only work on 5-10 of these sources. This is just how it works, due to traffic quality, targeting options, small reach etc..
THATS ALL. There’s no secrets. If there was it would be TEST more. The answers to a golden campaigns come from creative testing + data testing.
This isn’t the only way to do it at all either. It’s just my method but my goal was to show that i don’t have any secrets.
Hopefully this inspires you to sit down and test until you find a big winner
awesome post stackman! One question: How did you deal with banner+LP combination testing? Did you simply make pairs that matched or did you split test both variables at the same time? The amount of combinations and data analysis required makes my head spin. Do you have a simplified way of testing initially or do you just brute force it?
I'm a pure brute force when it comes to testing and split testing. I'm sure there's a magic organized method to making everything easier, some forum techs will know.
My method is simple. I have a batch of banners that are generic and yield high CTR. Think Facebook/ angles other very well known type of things. So thats banner set 1. Then i make banners around the angle of the landing page. Those are banner set 2. Again i go pretty simple in terms of banners/amount for the first batch of testing.
So my simplified way of testing is:
1. banner set 1 are generic high ctr banners that sometimes work for any type of campaign, you build these up after running lots of campaigns
2. match your banners to your lander angle
I'm really curious what kind of ROI you're seeing on mobile?
I'm new here, so I'll try to ask this as "non-revealing" as possible, but if you spend $1.00 what kind of ROI are you looking at (if that's not an ok question feel free to put me in my place)?
I'm in the process of trying to figure out where I'd like to dive in first and put all my time and effort (I've given myself until midnight to decide ).
Thanks!
Nice thread dude, very inspiring and totally makes sense. I know that its all about the data collection and trying different angles on LP's. Sometimes 1 same banner with 5-10 different ad copies can yield some very mixed results so you are totally right. What was your testing budget for this and how long did the whole testing phase take you before you decided to cut the losers and keep those 5 winners?
Also, care to share your best list of mobile traffic sources? I was looking to start into mobile and would love to get the best traffic sources for it so I don't waste my time with shit traffic.
Thanks!
Hello Stackman
1. How many offers can we test as newbies and how much $ per day per offer?
2. Any newbie friendly software for creating mobile landing pages (except photoshop and dreamweaver)?
3. When you say (I took the 5 winners, and made MORE angles for each campaign and tested again on the same traffic source) do you mean MORE LP?
4. What tracker do you use?
5. Any method for split testing?
Thank you
How many impressions are you buying daily for that $XXXX ?
What is the avg. ctr of your wining banners?
What is the offer payout and the avg cr?
Is this INTL camp or US?
What is your goal for ROI before you start scaling?
I hope you can answer those withour revaling too much about your camp, i am really interested in the numbers.
thanks
Hey stackman,
Awesome post!
Were there non-english markets included? How did you manage the test: all english on first runs (then translated? or never translated?)
thanks!
Originally Posted by xteesy
I'm new here, so I'll try to ask this as "non-revealing" as possible, but if you spend $1.00 what kind of ROI are you looking at (if that's not an ok question feel free to put me in my place)?
I'm in the process of trying to figure out where I'd like to dive in first and put all my time and effort (I've given myself until midnight to decide
).Thanks!
Originally Posted by davidkingstonn
My overall testing budget to narrow down 30 offers to 5 was around $4000. I know this is high. If you don't have this kind of testing budget, it's your job to research the offers better and try to narrow them down and drop some off. This is done by asking your AMs to help you narrow them down, it's also done by asking traffic sources questions. Poke and prod until you can narrow things down a bit for a smaller testing budget.
Also testing 30 offers at once isn't always the best move. I'd recommend 10 at a time at max, but i wanted to be honest and explain how i did it.
Originally Posted by zakariya
1. I'd recommend 1 at a time if you're brand new. Per day doesn't matter too much, but spend 5-7x the offer payout for the offer. Then move on if its not looking even semi-good.
2. I use photosphop and hire a programmer
3. Yes and no, i DID make more landing pages, but it was around new angles. ie: new ideas to promote the idea (search on the forums for "angles", theres better explanations and threads dedicated to explaining angles)
4. CPVLab, STM Tracker & Voluumm are all amazing
5. Within the tracker and creatve brainstorming
Originally Posted by krisicash
I look at scaling in 2 phases.
1. Is perfecting the offer/angle/landing page combo. I do this usually on the 1st traffic source, and then try new things every so often
2. I then move to other traffic sources
So for #1 i aim for 50% MINIMUM, unless there's so crazy crazy volume available.
ex: If the campaign was in USA, for all carriers, handsets and OS, i'd be happy with 20% ROI.
Originally Posted by ricmed
About 60% non english.
Hey Stackman, did you need an Agency Website/company of your own in order to sign up to all the mobile traffic sources?
Big time post brah, Thailand put us all into hyperdrive! People these meetups give birth to energy you didn't even know was in you
Stackmans day: Test, test, test, tinder, test, test, tinder, test, tinder, test, test
Originally Posted by cosmeivan
haha, that's what I thought attila, I was wondering why that other forum recommends an agency website for these mobile traffic sources.
I've been running mobile adult/ dating for a year now and was looking to expand to other niches!
Originally Posted by cosmeivan
Originally Posted by maynzie
stackman,you rock! the guide is amazing!
Great overview on your mobile campaigns Stackman ! I'm testing a few smaller Mobile campaigns and got some good tips from you. Thanks!
Originally Posted by highflyer357
My method is mass testing, and finding what sticks based on many many ideas. Then scaling on what looks best.
Sick share stackman!
I am sure that more brote force testing will lead that effort to xx,xxx days.. Do you agree?
Are you going push it to that level?
Good luck!
Originally Posted by delash
REALITY UPDATE
Not everything is flowers and pony's for experienced affiliates. This is the real world, there's no magic.
- I tested about 15 new sources thoroughly.
- Lost $3000, and had 2 sources do well.
- Ended up scaling everything by about 20% for a TON of work.
..Definitely worth it, and i'm doing it again next week with more sources. All part of the hustle.
@:stackman(or any mods): hope things are going great for you!
been reading and re-reading this thread and your post quite abit and would like abit more clarification from you if you dont mind!
from this thread:
"My overall testing budget to narrow down 30 offers to 5 was around $4000. I know this is high."
"1. I'd recommend 1 at a time if you're brand new. Per day doesn't matter too much, but spend 5-7x the offer payout for the offer. Then move on if its not looking even semi-good."
vs
from a thread awhile back:
"I quite often do quick campaign test on mobile by direct linking, but i use my owner banners.
Here's what to do.
1. Double check with your AM the offer is good/converting
2. Make your own banners, try 3 angles and 3 banners for each angle (total of 9)
3. Run the traffic again and see what you come up with. $.80 CPI x10 x9 banners = $8x9. So spent $72 for this test. See your results.
... if anything bites at -50%+ then try it with a lander
4. Build more banners/landing page around the winning angles or try 3 more new angles direct linking "
i've been trying to put puzzles together. you mentioned you have 50 urls from 30 offers, and spent about $4000, testing. this roughly equates to about $133 per offer.
then you recommend 5-7x payout testing an offer, but theres a different formula that you used awhile back which was payout x10 xno. of banners.
with you spending about an avg of $133 per offer to test, it seems like you're closer to the formula payout x10 xno. of banners instaed of payout 5-7x. so im abit confused here.
also, is anyone good with the math behind this test offer for X * payout? i can kinda figure out why why we wanna spend approximately 3x payout per pub/handset/etc but i cant see a strong math basis on derriving X * payout. i can definitely accept if the figure is derrived through repeated trials and experience of season affiliates since i dont expect too many of affiliates to be a statistics expert but im just curious!
There is no exact magic formula. You have to go with what your budget allows for each test. 3x 5x 10x it all depends on how many offers you are testing and how well you think the offer should do. You get that from spying, talking to your AMs, traffic reps, other affiliates. If you know an offer is doing well, then you can let the tests go for much longer and test more angles till it works.
Stacks would have run some banners/offers/etc. longer than others depending on their performance. So, the end results will never divide out to some set rule... for this to happen you'd have to kill things religiously even if they were at -1% ROI.
The whole spend 3x, 5x, 7x the offer payout is mostly experience based and varies between platforms - the more variables in the mix, the greater 'window of opportunity' you have to give a campaign. Working with statistical significance is best left to banner and lander CTR/CVR where you can compartmentalise your comparisons.
Think of it like a job interview... you listen to the candidate for 10 minutes, decide they won't work well so let them go. Another comes in... he is more interesting so you listen for 12 instead. If it was a simple job you might know quite early on if the candidate has potential. If it was a complex job with a lot of intervening factors to come, or you knew some people really bloomed after getting hands on experience, you might wait it out a bit longer before sending them home.
It's very much the same for campaigns - you come in with some testing objective that is the product of your experience, offer history, traffic source complexity and you throw down a test.
Which different variables did you plug into each mobile source? Carrier, phone brand, manufacturer?
Do you group all of them in one or did you make separate campaigns for different carriers/phone brands web/app traffic?
honestly, the only way to understand it is to do it over and over again... you will lose money at first but one campaign that hits will make up for damn near a year of losses in a couple days.... Only definitive answers come from your testing... Once you see the data in front of you it will start to make sense.
Originally Posted by supeyrio
"3. Run the traffic again and see what you come up with. $.80 CPI x10 x9 banners = $8x9. So spent $72 for this test. See your results."
..don't do 10x the payout for EVERY banner, instead do it for every different angle you come up with.
ex:
Say 1 angle is 'hot girls play this game, try it' and another angle is 'this game is banned in 160 countries'.
Your plan here is to make 2 banners that clearly state each angle, test those 2 angle for 5-10x the payout, choose a winning angle.
..THEN make 10 banners around the angle and find the best type of banners.
Ideally you want to do this with 5 or so angles to start.
Originally Posted by stackman
When you're quickly testing 30 different offers to find potential, do you bother splitting them up by app/wifi, app/mobile, site/wifi, site/mobile?
It's high time I gave mobile a real shot. I'm so doing this - thanks for laying it out.
Honestly I'm surprised you spent ONLY $4k to get to those top 5 offers, I had this notion I'd need to be prepared to spend a lot more to really get things moving with mobile.
Seeing as this was a few months ago, how have things been going with the campaign(s) and trying out new sources?
Can you shed any light on how you came up with that initial batch of offers, excluding the ones I'm assuming you grabbed from offer report emails?
Also it seems like angles are the real key here. You have a particular set of emotions you run through, or just re purpose top Facebook ads?
I'm so gonna try this Maybe not with 30 offers but 10 - 15 would work. Are your offers from same vertical, or you just try the best performing one from various GEOs and verticals?
Home > Mobile >