Home > Paid Traffic Sources > Mobile

---

How to do analytics and optimize campaigns on pop-traffic. (8)

---

Hey guys! I wrote this article in my follow along, but as people often can not find it, I post this material in this thread.
I want to say in advance that it's only my opinion and my experience. I'm not saying that it's an absolute truth. And sorry for my bad english.

How is analysis important

.
One of the most important skills in work with traffic, and especially with low conversion methods like popunder/popup/redirect is ability to analyze. The analysis and optimization can make your red campaign green and help to achieve positive ROI. And here I speak not about sites, platforms, Black or White lists as many people may think. Under the analysis I mean complete analytics of campaign which includes all elements, beginning with landings and offers, finishing with versions of OSs and hour of the day (step by step). I'll tell you at once that campaigns which I start, can be divided into 3 groups:

1. Super green without optimization (ROI from 50%) = 15%
2. The campaigns which needs optimization (ROI from-70% to +50%) = 70%
3. Campaigns which I can't optimize (ROI <-70%) = 15%

As you can see only 15 campaigns from 100 gives me a good ROI on start, this figure is very conditional, and it does not mean that from 200 campaigns you will unambiguously have 30 green campaigns, but the figure is close to it. And it is necessary to work with the others, further we just also will consider this work.



Landings and offers.


These 2 elements are key elements in the optimization. Remember the most important rule - You have to test ALL the offers which you can find from all networks available to you. Here we have a fresh example. The same offer in three Affiliate Networks:






As you can see one of the Affiliate Networks gave me -100% ROI, the second showed more than by 1.5 times the best result, than the third. What is the conclusion? To do spilt-tests of Affiliate Networks. However consider that with quantity of offers the budget on your test increases, therefore if you found too many offers (10+), you can ask your manager which of them:

• Are dead (I often see a lot of old offers in my networks which doesn't work. Especially it concerns offers with small IDs);
• Have enough statistics to unambiguously claim that the offer is bad.

This is a very important point. Surely ask the manager how many traffic this offer received. If offer received a little traffic - that is not a confirmation of this offer is bad/good, the best way for you to run by yourself and check everything for certain. Also don't forget that everyone works with different sources (Popup, Banner advertising, Native, Social Networks etc.) therefore the manager sees only an overall picture, he can't separate statistics on each type of traffic and, especially, on each source. Respectively, he can tell you that the offer is so cool, but it can be a situation that one person send on it a good banner traffic or native therefore your manager sees a good CR and EPC. Or on the contrary the manager says "Don't run it, the offer is shit!", but because just some people sent a lot of bot traffic from bad sources and manager sees very low CR and EPC.



How much money you have to spend to unambiguously understand, is offer shit or not? Answer: 10 payouts, but not less than $10. And I add one other rule: To run the offer at least one day for more exact analysis. Now I will explain why:

1. Differences in the time of day and day of week. Very often happens that one offer converts well in the afternoon and in the evening, and in the morning and at night he gives very bad ROI. And vice versa.
2. Conversions not always come at once. The postback delay can be 4-24h.

If you see that some offers gives your close results, run them together for finding of the best of them.



One more very important rule - test many landings. As much as possible, with different angles. What is considered different angles? At once I will tell what is not considered:

• Changing of color and the size of the button, and also the text on it;
• Changing of figures and data;
• Rewriting of the text;
• Change of a logo;
• Change of scripts.

All this can be considered as completions, but not different approaches. And what is considered now as a new approach:

• Complete correspondence of the text;
• Change of design is more than for 50%;

If you use two landers with diffetns texts, but with similar design - it's a new angle. It would be better if you play with design and make some new elements or edit elements which you already have. Even a simple detail like logo change can give you +30-50% ROI.

How many landers should I get to the split test? Answer: Not less than 3, it is better 5, it is even better 10. So you will precisely understand what works good and what works bad for you and concentrate on 1-2 angles. After I chose one best lander I have to create a three new landers with this angle but with different edits, additions (I can change a little bit my text, change button's color, add new scripts (alert, backfix, exitpop etc.), add some user's information (brand, model, city, carrier etc) on my lander. And I set my basic lander and three new lander to the split test and after it I choose the best one and on his base I create new 3 landers. And I repeat it until I have a good one lander and achieve a good ROI which I need.

Also you gotta know that if you have one good lander you can use it for any GEO, just translate it (I use onehourtranslation.com). I can say: A good lander is a good in any GEO. But if you see that your lander gives you bad results on a new GEO, you have to make a split-test with your lander and 2-3 different landers which you see in TOP in Spy-Service.

And one other rule - Check your LPs before running! Ok, you just downloaded some landers from Adplexity, add them to your server. You're happy. Yes? NO! A simple example:



What can we see on this picture? We see a static parameters from Spy-Service. When Adplexity's bot downloaded this LP, he used Samsung J7 and Hanoi's IP. And he copied this lander with this information. We have this code:

So if you don't clean this lander from "Galaxy J7", "Hanoi" and don't delete static dates, your users will see sth like this: "Your iPhone-6S Galaxy J7 is infected with viruses". You users will see double-parametes and double-dates.

Devices, OSs, Browsers and versions.
After I chose the best lander and the best offer I look at the Devices. What I mean when I say "Devices"? I mean: device type (smartphone/tablet), brand and model. Fist of all I look at the device-type stat. And If tablets gives me worse ROI than Smartphones (difference 30-50% and more and ROI is red) I turn them off.



As you can see Tablets gave higher CTR, but lower ROI. And I'll turn them off. My rule: The main thing is ROU. Not CR, EPC, EPV etc. Only ROI. Because some parameters can give you 80% CTR and -100% ROI, but it this time other parameters can show you 10% CTR and 156% ROI.

The next step is OS, Browser and their versions. If some OSs/Browsers/Versions gave me <-75% ROI, I turn it off. But I have to say one thing:

If these TOP parameters shows you a bed ROI, it means:

• Offer is shit;
• Landing is shit;
• Source is shit;
• This GEO+Vertical don't work on this source. Try to change the vertical or try new GEO.

Also as you can sometimes see Chrome 46+ shows not very good results on AV-PINs, but this versions gives you about 70% of traffic. So I'm trying not to turn them off. I'm trying to optimize other parameters to make them profitable.

ISP (Carriers).
The next our step is Carriers. I can say that often carriers shows similar results, but not always! A fresh example:



What we see? We can see that the 1st Carrier gave us only 15% ROI. And I think we can create a new campaign with this carrier with lower bid. What else? "Total access" and "Realmove" have almost the same amount of traffic but the 1st shows +347% and the 2nd only -77%. What I do in this cases? I'll turn off the 2nd carrier which gave me -77% ROI and create a new campaign with the 1st carrier with higher bid to get more traffic. But I have to turn it off in my basic campaign, because I would receive non-unical traffic if I haven't do it.

The categories and sites.
Actually, I do not like to delete any category, now I'll explain why. We have 3 sites and they gave us after test:

1. -87% ROI;
2. +71% ROI:
3. -100% ROI;
   Total ROI: -38%.

What we can do? We can turn this category off and cut the traffic from site which gave as a good ROI. Therefore I prefer not to disable categories, because it can be some good sites there. If I have 45 bad sites and only 5 good sites, OK, I'll disable all 45 bad sites and continue to run good sites.

How much money should I spend for test the site? Answer: 10 payouts, but not less than $10. If my payout is $2, I have to spend $20 for each site to make an opinion is it a good site or no. But! I can say that I disable all sites which gives me -75% ROI and lower after 5 payouts. There are no strict rules for me. I'm guided by my feelings. If I feel that site can show me a good result, I don't turn him off and continue my test. I can disable a site after 3 payouts and after 30. But If you see that some pubs gives you a very bad ROI on start (<-75%) you can to disable it immediately.

I have an other method. Filtering by CTR. If my average CTR is 20%, I'll disable all sites which:

1. Have at least 30 visits;
2. Have CTR: (Average)/2 = 10%
3. Have questionably high CTR (70%+)

A low CTR can tell as that this site have some bots or his audience don't interested in my ad. On the other hand an extremely high CTR means that a lot of bots on this site.

A little bit about WhiteLists.
If you see that some OSs, Browsers, ISPs, Sites gives you a better ROI than others, you can create a new campaign with only this parameters/sites and disable it in your basic campaign. But! You need to wait at least 2-3 days. Because you might have this situation:

The 1st day:


The 2nd day:


What can we see? We see that some sites on the next day gave worse ROI and some gave better ROI.
Therefore I always run at least 3 days each site to make an opinion about WL. Do not hurry to create WLs.

Often I have the situation when my WL campaign gives me more money than my BL campaign. And it's normal. I just check my BL campaign everyday for new sites which I can add to my WL campaign and disable them in my BL campaign.

What should I do if I can't disable some parameters?
If your source doesn't have ISP targeting or versions tageting you can just separate this things which gives you a bad ROI and move them for other campaigns. For example:
I'm running RO AV-PINs and one Carrier and one version of OS don't convert in my campaign (so they shows a bed ROI, but other shows a very green). I can just create a new campaign with sweep lander and sweep offer and move these Carrier and OS version on this new campaign. I just set up a rule in Voluum or I use traffic separator in Pixelk.

Algorithm.
So my algorithm is:

1. Offers. Choose the best 1-2 offers;
2. Landers. Choose the best 1-2 landers and optimize it. It is necessary because after lander's optimization sites which gave you a red ROI can give you super green ROI;
3. Devices. Disable all bad devices.
4. OSs and versions. Disable all bad OSs and their versions, but try not to disable the TOP of amount of traffic;
5. Browsers and versions. Disable all bad browsers and their versions, but try not to disable the TOP of amount of traffic;
6. Carriers. Disable all bad carriers.
7. Sites. Make Black list or White list of sites.

In conclusion.
I want to say again: To do a good analytics and optimize your campaigns - the best way to achieve your goal.

• Do not be afraid to test new angles, new landers, to write your own texts, to test a lot of offers and have a red ROI for a long time;
• Try to wait 1 day after each step of optimization;
• You gotta know that lander's optimization can give you +150% ROI. Therefore please don't be lazy and optimize your LPs. It's your main weapon, because people see only your lander. You don't have any banners or texts in your ad;
• Test a lot of offers. As much as possible, and you'll find only one which give you a super green ROI. I often have a situation when I tested 10 campigns, 9 were red and only one was green, but its profit covered all my spends for 9 red campaigns.
• Do not seat on the one case! Do tests every day!
• It's a normal situation when AV PINs don't work in some geo. Just test sweeps. Sweeps were bad? Ok, test dating. Test all verticals until you find a good one for this GEO.
• Test 1 offer from different Affiliate Networks, because you can get absolutely different ROI.
• Look at the spy-service and watch what people are running. You can use stolen landers on start, but you have to create your own landers if you want to a good profit.

Good luck!
P.S. If you have questions, you can write me here. If you want new articles, write me on FB what you want: https://www.facebook.com/mrbraun31

---

Hi Great Thread, thanks for taking the time to write this .
Here is a question for you
Landings and offers.


"These 2 elements are key elements in the optimization. Remember the most important rule - You have to test ALL the offers which you can find from all networks available to you. Here we have a fresh example. The same offer in three Affiliate Networks"

Why would you test the same offer from 3 different networks, since its the same offer, whether it came from network A, B or C, the result should be the same unless the payout is different, am I missing something here
let me know
thanks

---

Thanks man. This is one of the hottest post I've read to date on stm. pure value man! keep it up.

---

Do also seek out other threads and posts by mrbraun. Also, a little birdie told me he's cooking up something great for you guys - I'm eager to see what else he'll come up with!


Amy

---

Thank you for this. Will definitely read this again.

---

Thanks MrBraun for very valuable knowledge
I don't understand the part "Have to spend 20$ for each site". So "site" here is the landing pages right? Do you have to spend Payout x 10 for each landing page?
Thank you!

---

I was wondering about the same thing

---

The same offer can perform very different in different networks.

Can be redirect chain, scrubbing or whatever.

So when you have a winning offer available in different networks test it in all of them.

---

Home > Paid Traffic Sources > Mobile