First mobile campaign - Traffic source comparison (53)
Hi all,
After reading this forum few days and spend the rest of the time setting up a server, registering for affiliate networks, traffic sources, learning how to make a mobile landing page and creating banners I am ready for my first campaign.
I've decided first to test different traffic sources, there are a lots, and my account managers where all suggesting various sources. Best thing to do I thought is just testing some and go on from there.
Setup
- Server: Beyondhosting
- Tracker:
- Offer: A mobile card game download for Iphone & Ipad
- Landing page: Simple landing page focus on user benefits (redirect depending on Iphone or Ipad)
- Banners: All banners are the same, 10 sizes are created
- Traffic sources to test (simultanuously): iAd, Buzzcity, Inmobi, Airpush (all 10$/day, various cpc's)
Expectations
- Learning the difference in volume, cpc's & traffic quality for different mobile traffic sources
- Some conversions
Next steps
- See what traffic source works best
- Once a traffic source is selected keep optimizing the campaign based on data
I hope this setup makes sense, but feel free to hammer me down if it is a stupid thing to do 
Current screenshot of the setup in
Looks like a sensible split-test. What's the payout on the offer?
Looking forward to seeing your results!
The payouts are:
- € 2.40 for Ipad
- € 2.00 for Iphone
OK, cool. You'll need to get a decent sample size on all your traffic sources, then. I'd recommend at least $40 spend per traffic source before making any decisions.
Are you running split wifi/carrier/app/site tests on each source? If not, that's definitely worth doing too. At that point, probably best to run the standard $15 - $20 per sub-campaign per source.
As per Caurmens' suggestion I've made more testing adgroups.
- Not every traffic source makes it possible to target web or app. (only inmobi for my selected sources)
- All traffic sources made it possible to select Ipad & Iphone as different target devices
- All adgroups are split tested between direct link & landing page in
I am a bit worried about being able to generate enough traffic since the target groups are somewhat small, but we'll see. First things first, waiting for the new campaigns to get approved.
My new (improved) campaign setup:
That is a very good idea to see which traffic source is the best to focus on.
Just make sure you spend enough to have meaningful data.
Another thing to keep in mind that some networks are stronger to certain geos compared to others.
So I suggest for each of these networks, get an AM and ask for a list of volume by GEO.
Thnx bbrock32, I've send them all an email regarding cpc's & volume. I'll keep regular updates here about the responses.
It seems all ads are approved, which is great news... Making the banners/landing pages took lots of time.
Allright, first results are in!
Visits: 1,034
Clicks: 6
Spend: $12.18
Conversions: 0
Roi: -100%
Fact1: My landing page is not even close to what I was expecting, which is strange because I optimized it for Ipad & Iphone
Action1: Move to direct linking (now it was split tested)
Fact2: Iad seems to be the greatest provider of traffic
Action2: I've noticed there are some more possibilities to target which I didn't use. I will make 2 extra campaigns based on app versus web traffic. (it is called differently, that's why I didn't notice it)
Fact3: Inmobi & Airpush had no visitors. I've had a response from the Inmobi AM and he suggested to set the daily spent on 20$ to get traffic.
Action3: Increase Inmobi and Airpush daily budget
Thus the goal for next day is to increase knowledge about Iad visitors & get Inmobi & Airpush moving.
Interesting journey, but over 600 clicks to a game and no download? Something extra I should do for making people download something in iTunes?
What creatives did you use? What was your lander like?
The xxx clicks to game offer and no conversions is a pretty common outcome... especially when using stock creatives and if advertising in a way that garners a lot of accidental clicks.
Thank you for the feedback.
Here are some creatives I am currently using. (on Iphone I am having 0.1% ctr, so probably you are right about the accidental clicks. Any ideas how to minimize this?)
And below the landing page, very low ctr:
Thank you for all the feedback, I am following each advice given and learned a lots in 24h!
0.1% CTR is very very low - it won't be accidental clicks.
The creatives are very tame and boring. Who are you trying to appeal to? Engage them.
The lander is similarly very non-engaging.
Your ad copy:
Download free... - button already confers this message.
Enjoy sceneries - who cares, i like solitaire.
Have a great time.... booooring! I don't want a great time with unique card game variations. I want a reason to play solitaire!
This might be quite a hard sell but you want people to be interested in playing a card game... and to do that you need to appeal with more relevant/engaging text and images.
Avoid the logo, it's boring. Maybe try a picture of the end game shuffle effect from Windows Solitaire where the cards fly at the screen. Challenge the user - Solitaire pro? Bet you can't beat our unique twists on this old game. There are a lot of angles you could take that are far removed from the very 'stock' creatives nature of the current banners and lander.
Also, definitely focus on making much more interesting ads and direct linking first - i'm not sure people need much extra convincing to play a card game.
Zeno,
Your comment came like a hammer! Calling my ads boring with 4 "o"'s! After recovering I decided you where right. The thing is, I read a lots about engagement but just didn't include it. Not sure why...
Today something brilliant happened also! I've got my first mobile conversion ever! Bye bye -100% ROI, hello "yes, this can work" feeling and great, the tracking works brilliant!
After thinking I made some decision related to this campaign:
Fact 1. The other traffic sources are still not providing me any visitors after the changes yesterday.
Action 1. Cancel them, and focus as proposed by so many people in this forum on 1 traffic source for this campaign. This sucks a little for the people that are clicking this follow along to see the difference in traffic sources. But I think I need to cut my losses quickly on this and first earn some more basic experience.
Fact 2. My ads are not good
Action 2. Create 2 variants and see if it makes any difference
Enjoy your day all!
@dennisp - having looked at your creatives, they're all very well-balanced and nicely designed. Are you, perchance, a designer or similar by trade?
I'd recommend pushing hard to take a few more risks with your design and copy.
One thing I suggested to one contractor of mine recently who was working on banners was "try and get your banners disapproved" - without actually breaking any rules, try to just make them so ugly that the traffic source won't allow them.
None of them were disapproved (he's quite law-abiding in general, so he still didn't push it that hard) but he quadrupled his CTRs on the next load of banners.
Unleash your inner shady affiliate, at least a little bit! Try stuff that looks rough or doesn't feel professional, or even that you're worried might offend. You'll probably find it works much better.
(Note for anyone else reading: this isn't advice I'd give to everyone! And I'm definitely not recommending doing anything actually non-compliant. But if you're staying too safe, that's almost as bad as being too risky.)
After some great feedback on my current banners I've created new ones. Probably it is still quite boring, but I've tried to add more elements to convince people to actually play the game.
http://auberge-delapaix.com/newbanner/960x64.gif (this is not the server I am using for mobile campaigns)
I've created various variations and just uploaded them. let's see if I can get some better conv. rate on these ones! I'll keep you updated.
All my new ads got disapproved :/ I've made a stupid mistake using "fake" review stars. So now I've made all the ads compliant with the guidelines but unfortunately it is weekend so I don't expect them to be approved before Monday.
I'll be having a week vacation with my girlfriend, she would kill me if I would be working... To avoid that, I'll set the campaign on pause and I'll be back next week!
That's better! I would use a different background colour though, something that contrasts with the cards i.e. dark grey, black, dark blue, etc. Then just make the text white.
I thought you could not advertise on iAd as an affiliate. How are you tracking with iAd?
After one awesome vacation, back to the AM world!
Meanwhile I left my ads running to get some data. After I've changed the banners, the results are quite impressive:
Results:
My conclusions:
- There is not much difference between the 2 new banners I've created
- There is MUCH difference with my previous banners that had -100% ROI
- I am almost in the green zone!
- For a small 6$ I've learned a lots
Next steps:
- I will pause the 2 banners for now
- Create 2 new banners with a different angle
- Compare the results with these banners
@Fatmach, I didn't had a problem with it.
Looks good!
I'd recommend trying to up your spend at this point. You're spending little enough that you'll have difficulty separating out results from noise - can you boost your bid or your daily budget?
Today no results, I've made a stupid mistake and my account got momentarily suspended.
Now it is working again and took some action.
- I've done some research and found out that woman aged 24-35 could be a good target.
- I've created 2 new banners with new angles and decided to test them immediately on this target group.
- As per Caurmens' suggestion I've doubled the daily bid on the previous banner set
Tomorrow around this hour I am expecting to see the first results. Any idea what target I should be aiming for before heavily increasing the budget?
thnx
How quickly you scale budget really depends on how successful the campaign is.
Plan for $xx/day spend and increase based on real data that comes in.
How exactly did you get your account suspended?
Thnx for the feedback Zeno.
I've gotten my account suspended by using a prepaid cc (I keep hearing all these horror stories about overspending campaigns, so with a prepaid card I cannot spend more than is on the card), but I forgot to put enough money on it to pay for the expenses.
Today I was a little disappointed, traffic is only coming in slow... very slow. Last three days I had around 500 visitors and 1 conversion.
- A solution could be to increase cpc
Problem is that I am still testing if the female target group would have a better conversion, so this is my variable. CPC increase would add another variable and is planned as another test.
So to increase my having something useful to do I am going to start a new campaign. This will give me 2 campaigns to test thoroughly without giving in to the emotion of making changes before a test is finished.
Not enough clicks for the $ spent, need to work on much better banners. Up your CTR with shock/interesting banners.
So higher CTR = More money from the ad network = More views against low cpc ?
As per Stackmans' suggestion also a new bannerset has been created. I tapped in to the fact that lots of people think solitaire is good for the brain... "TRAIN YOUR BRAIN WITH THIS FREE SOLITAIRE GAME" style.
Curious to see the results
Are you bidding CPM or CPC?
In either case, a higher CTR will reduce click costs and give you more data for your tests.
I like how you structure your tests but here's some more ideas/suggestions
For angles/banner designs
-make sure you are testing at least 5 radically different banner design PER angle
-test at least 10 angles PER ad format/type as well
For traffic sources, be sure to test every ad type/size separately
Examples:
-a 300x250 placement can perform wayyy differently than a 320x50 placement both CTR & CR wise.
-a certain ad style format can be better than the next format (ie text vs display vs interestitial vs exit etc)
so test everything.. and the data will tell you which road to take after to scale
Thnx iAmAttila, this makes me think that I need a much much better planning for tests. Now things are a bit too much ad hoc. I will think about this for the next campaign.
Could you recommend mobile traffic sources that make it possible to test different banners in an ad group? Currently I need to make a new advertisement for each and every banner.
Today all the numbers where very red:
1184 clicks, no conversions.
- I've cut the banners that are lost (double payout without conversions)
- I've left banners with low traffic running and increased cpc
As I said, I will also start on another offer. This is definitely not a get rich quick thingy... Lots of hard work, but I like it so I decided to keep focusing on it.
It has been quiet here... but I was too busy setting up new campaigns
I realize you need data before being able to any feedback, so here I am with new data.
As you can see, this looks quite promising, -36% ROI and great possibilities to optimize the campaign.
- To increase CTR I can pauze obvious loosing ads
- The same for increasing ROI
The difference between lander & no lander, I need a bit more data for that. It seems to depend heavily on device.
But I have a question:
- I am looking for a statisical approach to make decisions based on data analysis rather than "hmmyeah, it looks like this ad should be pauzed". Anybody has any idea what models or tools I could use?
thnx a lots! I come back with more data soon!
Can you reupload those stats to e.g. imgur then insert into the post? The image is too small right now.
There are lots of tools you can use, ideally you need to come up with a sort of checklist or set of rules for your specific traffic source.
You can of course use tools for simple comparisons to decide if one ad variation is better than another, e.g. http://www.peakconversion.com/2012/0...al-calculator/
thnx for the feedback, the tool is exactely what I was looking for, thnx!
Best
Edited your post to put the full image in.
Note CTR is expressed as a percentage so change that column formatting to that, so it reads e.g. 0.239% rather than 0.002391.
I don't quite understand how the clicks, visists and clicks once more reconcile with whats going on - are some direct and some via landers?
It would be better to split that data and explictly show lander CTR, and label what comes from the traffic source, what's from the tracking system, etc.
So fare, I think the biggest issue is that your ad CTRs are very low and improving this should be your main focus!
Allright, here is a quick update. since 4th of july I started a new campaign. These are the results
Day1: (start)
Conversion rate: 0%
Revenue: 0
Cost: 1,86
Profit: - 1,86
Day2 (just look at the numbers)
Conversion rate: 3,9%
Revenue: 18,37
Cost: 25,83
Profit: - 7,36
Day3 (remove bad banners)
Conversion rate: 6,8%
Revenue: 91,85
Cost: 85,51
Profit: 6,34
Day4 (test with other ad formats)
Conversion rate: 7,8%
Revenue: 116,35
Cost: 133,54
Profit: - 17,19
Day5 (remove bad banners)
Conversion rate: 12,7%
Revenue: 189,82
Cost: 137,18
Profit: 52,64
Day6 (made winner & test campaigns but made a mistake with device targeting)
Conversion rate: 20,8%
Revenue: 91,85
Cost: 66,1
Profit: 25,75
Day7 (WTF is happening?)
Conversion rate: 12,6%
Revenue: 104,09
Cost: 133,28
Profit: -29,19
Day8 (Removed wrong device & included new landers)
Conversion rate: 15,6%
Revenue: 140,83
Cost: 103,72
Profit: 37,11
Totals
Total revenue: 753,16
Total cost: 687,02
Total profit: 66,14 (that's about 1€/hour of working 
)
After this mornings' data analysis I noticed that I can reduce neg ROI during the day by targeting specific hours instead of everything. (nights are big time losers). If I had done this from the beginning the data is saying my profit would have been 112$ so I think it would be a nice test. (during the weekend I'll leave the campaigns just running, people don't need to work, go to school,...)
So, that's it for now, I am extremely happy that I don't really need to pay for learning (except of hours working) My apologies for not very frequently updating, but days are busy with campaign optimization.
Day 9 (I didn't do so much, preparing for next week testing and another GEO for the winning campaign.)
Conversion rate: 15,3%
Revenue: 122,46
Cost: 97,90
Profit: 24,56
Probably I could increase traffic by going to other traffic sources, but I decided to stay with the same traffic source and focus on other GEO's instead, for now. (strange how I changed my idea. First when starting this follow along I wanted to test all traffic sources... But the truth is that a traffic source takes time and focus)
I thought the following image would be a cool share. It is the ROI per hour of the day. This image is the reason I will change the hourly parting of my campaigns.
Yes, you should be dayparting more aggressively
Hey, great follow along I already learned a lot and it is good to see you are turning green so fast!
I was wondering if you are still using iAD?
Decisive isn't working well anymore - no traffic.
Good luck!
Not sure about decisive, never used it. I like iAD, but I feel I am to green behind my ears to really have a good opinion about traffic sources.
Day-parting: looks like good data to base day parting on. Make sure you're calculating your confidence intervals on that data, though (see Getting Started Guide Part 10) - my guess is that the statistically likely range for those per-hour ROIs is still pretty wide and as a result you might end up killing winners or including losers.
Caurmen, you are right.
If I only look at ROI then I am between 60-90% significance for the hours I want to target, thank you for commenting to test it. Still I am fairly sure looking at the other metrics that it is an interesting test. (It is low budget so we will see)
Yesterday I got my ass kicked during the world cup final :/ luckely I pauzed the campaigns in time to minimize the loss.
Day 10
Conversion rate: 11,3%
Revenue: 67,42
Cost: 75,76
Profit: -8,34
Today the dayparting is on, and the 2nd GEO will be slowly tested!
Day 11
Conversion rate: 11,48%
Revenue: 85,79
Cost: 61,29
Profit: 24,5
That's about 40% ROI. Maybe I should start thinking about scaling, but to get more data with a small budget I'll just leave it running this week and see if it is consistent.
Nice one!
Day 12
Conversion rate: 13,00%
Revenue: 78,67
Cost: 57,72
Profit: 20,95
Good, another green day. Unfortunately no time for scaling due girlfriend attention needs (but I don't mind ).
But, I have some banners running in a test campaign, so I am still gathering data which is nice.
Nice!
Day 13
Conversion rate: 13,61%
Revenue: 127,82
Cost: 89,3
Profit: 38,52
Looks nice, problem is that half of the revenue is from the 2nd GEO. The first geo is just braking even. But like I said, just gathering some data for now.
Are you still direct linking? Or did you get a new landing page working?
In the beginning I did a test between lander and direct linking. I turned out that the lander was more successful. Now I am testing between different landers
Day 14
Conversion rate: 12,41%
Revenue: 115,65
Cost: 87,18
Profit: 28,47
This week I am continuously having green results. That's awesome, looking forward to analyse the data in detail tomorrow. Probably I am just really lucky to make this work. One of the things that I believe are helping:
- I am not using a popular traffic source
- I am not targeting highly competitive countries
- I am using dayparting aggressively
- I have loooots of banners in testing phase
- I am testing 2 landers at any given time
The main problems are:
- I am having next to all the traffic on this traffic source for my targeted countries. Scaling will be difficult, so I need to test another traffic source or implement a new country.
- I think I am leaving too much money on the table by not really adapting the landers to the banners, I should be using angles more efficiently throughout the whole sales funnel
My ctr on my landers are 41%
Then 14% actually leads to a conversion
Any experience that this is good, or should I improve this?
Nice work! And yes, all of those will probably be helping.
I'd definitely try some more traffic sources at this point - see if you can scale these campaigns out. Keep testing at the same time - you've got pretty decent figures but it's possible to get better.
Are you split-testing your offer across a couple of different networks? That's another thing I'd definitely do. And ask your AM if you can get a bump - more payout will mean you've got more room to scale.
Continue to day part, match ALL of your landers to your banners and start weeding out all the crap traffic (of which there is plenty on mobile) and then bid higher on the geo that is working.
^^That's a very good point from cmdeal.
Have you tried temporarily (for an hour or so) whacking your bids much higher, to see how much more traffic there is available if you raise your bids? If you're capping out the total volume available to you from this traffic source regardless of bid, you'll need to test other geos or sources (and optimise your existing campaigns, of course), wheras if there's much more volume available from higher bids, optimising on the traffic source you're using makes more sense, thus allowing you to raise bids.
One week later, week 2:
Conversion rate: 11,68%
Revenue: 730,32
Cost: 581,20
Profit: 149,12
Pure awesomeness! I wish I had the time to really optimize the campaign right now. So many things I can do.
For now, today:
- I removed bad performing banners
- Included good performing ones in the winning campaign
- Bid higher on successful placements
- Removed bad placements
- Removed a 2nd landingpage (didn't include a new one, since I am too short on time)
- Made a small change in the dayparting. Removed 2 hours, added 1 hour.
thnx for all the advice here, some points are really really strong.
Good job! You have found something that works. Focus now on optimising and scaling, while not getting too worked up about minor day-to-day fluctuations which are natural.
Home > Mobile > Follow-along Campaigns