Home > Paid Traffic Sources > Mobile

---

One Year Ago I Launched My First Campaign - $100/day case study (20)

---

Epic, tons of solid info. My hat is off to you sir

---

Is it just me or does the "big winner" animated graphic look phallic AND have a hand rubbing on it to make it grow? Is there any way to know (or maybe test) the CTR between male and female users? I know this is an old campaign but if my theory is correct, I'd bet that graphic would have converted better if you sent it to males only (if possible). Probably would have made more than $100/day??

Either way thanks for sharing. Great post.

---

Haha, I've been told that before about the best banner there

On Go2Mobi at least, you cannot filter for male or female so I didn't even consider it. Also, the campaign itself "started off" at $100/day. When scaling it made more and for a good amount of time.

The conversion rate was really very good from what I experienced afterwards, so I think that is not what was missing to make the campaign more profitable. Maybe a more WTF banner to increase CTR (lower price per click) should have been something to consider. The biggest way to scaling it at that point was to get more traffic, and also optimize what there was left to optimize with the existing traffic: phone models, OS versions, ISPs, etc.

---

Initially how many geo did you test when you split-test different offers? How much money did you spend and how many banners used to test your offers?

Thanks in advance.

---

I always went for Tier 3, so for the other offers I still always had India, Brazil, Philippines, Malaysia, Indonesia and the likes in. For DU Battery, it was not different...

I spent around $1.8k TOTAL, including services (hosting, translations, tracking, etc.) before being close to breakeven with this one campaign. Hard to put an exact point in time but let's say after <$2k in the red I was pretty sure this can make money.

How many banners is a question I cannot answer exactly. I made banners EVERY SINGLE DAY, for weeks and weeks, even when things started working, to improve and avoid banner blindness. So it was probably ~1000 banners if I include changing the colour as another banner. More than I should have now with more experience but more was better at that stage, most definitely. And yes, I did all this myself, I only hired a designer after 3 profitable months where I realized I couldn't do much else but churn out banners

Hope it helps, let me know if you have more questions

---

Wow ! It was a huge amount of money to spend to split test the offer, meaning to said that you spent 1.8k before you got the breakeven in DU battery offer. May i know what traffic source you using ? only go2mobi?

Thanks for the reply, it was a really EPIC post.

---

The $1.8k is from the VERY FIRST offer I tried it and then INCLUDING DU Battery Saver. Some people spent even more than that to get their very first profitable campaign.

For this I used Decisive and then switched to Go2Mobi like I mentioned above.

Now we use Go2Mobi when we do display, but most of our volume is pop traffic. We use all the commonly mentioned sources here on STM, aside from ZeroPark.

---

Thanks for reply.

To be more clear,when you split test the offer, let say for the int offer, how do you set up your campaign and pick up the geo ?

For example : Campaign 1 --> Offer 1, offer 2 , offer 3 and offer 4 (geo 1)
Campaign 2 --> Offer 1, offer 2 , offer 3 and offer 4 (geo 2)
Campaign 3 --> Offer 1, offer 2 , offer 3 and offer 4 (geo 3)

Is it the above campaign set up is proper to split test the offer ?

Please feel free to comment my follow along post--> http://stmforum.com/forum/showthread...-my-1st-50-day

Thanks in advance.

---

There's a difference how I did it then, how I'd do it now on pops, now on display, or other variations you can think of.

The basics are that you wanna make sure the geo variables can be compared. That means the others are the same.

How I did it then:

Make angle -> make banners with angle for X geos -> run direct linking to the same offer. This was split testing geos only, it was the only variable. You need the angle to be the same.

If you add offers to the split test, unless they can use the exact same angle, you will have a hard time comparing Offer 1, geo 1 with Offer 2, geo 2. Basically, once you have more offers, you can have combinations like:

Angle 1 + Offer 1 + geo 3 = green
Angle 2 + Offer 3 + geo 1 = green

That means both Offer 1 and Offer 3 are good to run. They are different geos, and even use different angles but profitable already.

How it happens more realistically is that none are green to begin with but you see significantly different ROI after enough traffic sent to them. This also applies for more affiliate networks. Offer 1 on Aff Network 1 could flop, but Offer 1 on Aff Network 2 could beat everything else. Same offer but the affiliate network made the difference.

Try to write down the variables you see so far when testing, then see how they connect with each other, then maybe it makes it clearer how you can test geos (or anyhing else for that matter).

---

It really helps me lots on understand how to split testing, Thank you for taking your times to reply my question.

---

My pleasure, glad it helps.

---

Yeah, it's makes an UNBELIEVABLE difference. And I am pretty sure you can have a worse setup than I had then. I am quite sure your tech setup can make it impossible to be profitable, if it's the completely wrong one.

Also, just so people don't expect -50 to 0 difference all the time: Neither ROI was with a HUGE amount of data so the real ROI could have been -60 and -20, or -30 and 0. Deviations are still possible but the goal was improvement, and it showed CLEAR improvement. There was enough data to compare to two but not enough to pin point the ROI of each.

Glad you enjoyed the read

---

Thanks for taking the time to write this excellent post! Learnt a lot from this

---

Thanks for the kind words and glad it was helpful

---

What was your approach to bidding at the beginning of the campaign?

And how many angles did you test before finding one with potential?


Great case study!

---

Thank you for an awesome post Wanted to know what are the best resources out there to build banners, landing pages etc. or what software language did you learn in order to build these banners and landing pages. Do most app install offers work better with a lander or sometimes direct linking is better.

---

Great case study. Thanks Manu!!!

---

I just bid around the average bid suggested in the planner Decisive had, then took the working bid to Go2Mobi.

I tested several headlines, with my mastermind group back then, around 15, but they were not all completely new angles between them.

Only one angle had potential, and I started building on it.

---

I know basic coding since high school, so it was relatively easy to edit pages written in HTML/CSS + JavaScript. I use Sublime Text 3 to edit the source code, I don't like visual editors, they produce poor code quality.

As for lander vs no lander, you should test both when going into a new vertical. I think very often landing page wins.. Extremely often, especially if it's not already a popular brand of app. That said, the question is very bad. You can find out every single time the answer with a quick test.

---

The banner text "belajar lebih" is not the correct translation.

---

Home > Paid Traffic Sources > Mobile