Home > Mobile > Follow-along Campaigns

---

Let's get the green at sonic speed! Newbie mobile pops + sweeps Follow Along! (4)

---

Hey STM! I've been meaning to start my follow along for a couple of days now but everytime I thought about anything related to my campaign, I ended up just wanting to work on it more/do research but I know that this will be super beneficial to me and hopefully anybody else that is trying to make it. I've been testing different things over the past week but in the past 3-4 days, I feel like I'm actually doing proper campaigns at least somewhat and that I have an idea of what I should be doing. Before, it felt like i was clueless but I kept trying things out until that feeling of confusion went away. My current state of mind is not confusion. Now I feel like, although I am not profitable yet, I feel like there's questions that I can think of to ask and things that I can constantly improve on. Before, I felt like I wouldn't even begin to know what to ask or what i should work on. That said, I'm totally expecting to catch things that I haven't thought about before and I'm excited for what's to come!

My goal is to become profitable at first but realistically within the next month or two, I'd like to be hitting double digit profit days. Doing $XXX and $X,XXX profit days are definitely on my mind but I'm realistic and ready to take this one step at a time. I won't get there without putting in the work and I'm ready to do that. I haven't had a campaign that was profitable yet but I can see small pockets of profit and I'm trying to reach those.

I made a thread earlier asking what I should consider when doing a follow along and a thing that stood out to me was that I should be transparent and I plan on doing that as long as it makes sense. Best case scenario could be that someone else learning along with me tries the same things that I'm doing and figures out how to make it work for them! Share if you'd like of course :P

Enough thoughts and rambling though, here's the hard info. My focus is on mobile pops + sweepstakes and I'm currently focused on South Africa. I started off with the mobile cookbook and did some stuff with Go2Mobi (without any spying) but through advice from a fellow STM user and from reading a lot of other threads here, I switched to POPS because that seems to be much easier for a beginner to handle at least for really learning the ropes(no banners to worry about). I saw Vortex(she gives awesome advice like.....everywhere!) mentioning in another follow along that it would be a good idea to eventually switch to banners from pops and that's definitely on my mind but I'm going to stick with pops for now. As for the specifics, I'm running the "free lotto" offer from Peerfly. Peerfly has been super helpful to me and the seem to have a decent amount of offers. I'd definitely like to get onto some other networks but I've been rejected for being not experienced enough (at least from Clickdealer and F5) so I'll have to stick with them for now. I recently was accepted with Addiliate so I'm going to look around and see what I can find there as well.

I've been using popads for traffic but I recently tried to expand to Zeropark because it seems like there's not a lot of traffic for South Africa on popads and I read that zeropark has good traffic for tier3 geos.

Now for the data. In the beginning, I grabbed the first lander that showed up in adplexity (THIS TOOL IS A AMAZING especially to me as a beginner - highly suggest investing in this unless you have another spy method that is working for you) that seemed to fit. I searched for "most traffic" in South Africa and this spinning wheel lander showed up where you click it twice and it gives you a button for the offer. There seemed to be more landers but I Was just excited to try something and rushed and set it up with the free lotto offer in Voluum.

I assume this is decent as far as RIO is concerned?


Unfortunately, I goofed hard with the tracking setup so the ROI for this campaign should be way worse than what is shown because as you can see here, I originally was not tracking costs OR placements (popads refers to these with "websiteid". I thought that Voluum would automatically do that with the traffic source template but it didn't work until I manually added tokens from popads to the url. You can see how that looked here


Well, I learned my lesson with that one. I prepared to properly setup the campaign and this time, I took the advice that Vortex fortunately repeats a bunch and decided to test a bunch of offers/landers at the same time. There were 4 other offers that were extremely similar on Peerfly and a couple of landers that also definitely were being used for these offers (one text based and 2 other variants on the spinning wheel). That left me when 4 landers with 5 offers each. I setup 4 campaigns (one per lander that would rotate each of the 5 offers) to test.

The results were interesting. The landers correspond to LP1-4. The top row is another test that I'll explain after.


There was one other offer that was close but after the second day, the original offer/lander combo that I started out with before mass testing was consistent on both days and slightly outperformed the two other offers that looked like they had potential. Big mistake #2 showed up here! I definitely rushed a little too quickly in making the campaigns in popads and accidentally targeted desktop as well. This was a money sink and I actually could've had some profitability $0.54 on LP1 without that error it seems. I fixed that in the next test which is in the first row while also removing an offer that just wasn't converting at all. Interestingly enough, after removing desktop and the bad offer, the ROI went down 2%. Another offer ended up not converting at all. Regardless, I believe the point of the mass testing is to pick the best combo, optimize around that, and then apply the optimized campaigns to other offers so I did that and ended up with this



This is the best performing lander/offer only while also blacklisting a couple of super bad placements that were getting a ton of traffic with 0 conversions. It got me ~10% increase in ROI. I find this interesting because the LP1 could've been profitable in the previous mass test if I had desktop disabled but I was also dayparting from 6PM to 12PM back when I was mass testing. This most recent test I've run the entire day to collect more data/traffic but it seems like the hours that the mass test were profitable at are different from the profitable hours of the most recent test with single offer/blacklisting.

That's where I'm at so far! I've also tried to duplicate the campaign on ZeroPark which I was recently accepted to and it BURNED through money like crazy. It seems like they have way more traffic for South Africa which is definitely a good thing but through running a RON campaign with them focusing only on mobile + non adult, the conversion rate is terrible relative to how the popads campaign performed. It gets like 4 times the traffic of popads for South Africa so I need to figure out how to not burn money in minutes.

My thoughts/questions/takeaways for now:

1. Should I run all landers/offers in the same campaign and have voluum rotate them all? I was suggested to make sure that voluum rotates the offers so that they are all getting the same placements. Should i apply this logic to landers as well?

2. I'm not really sure about how to approach whitelisting. The campaign on popads has only two placements with over a single conversion. A bunch of placements are profitable with one conversion but is having mostly placements that are profitable with one conversion each a situation where I should think about whitelisting?

3. THIS is probably a huge factor. Like a huge huge factor. I'm only using cloudfront right now and apparently, there was some crazy javascript loading going on and the full page render took 100 seconds when tested in South Africa. I saw the post about how increasing the landing page speed by a couple of seconds doubled the ctr and I've seen other posts about how loading time should be under 2 seconds. I've gotten it down to 10 with just getting rid of the javascript that didn't seem to actually be doing anything. I'm definitely gonna do my best to get my that lander optimized to see how well it can actually perform. Is there a recommendation on what I can do infrastructure wise to support this? VPS near South Africa or a CDN?

4. How important is a having a custom tracking domain? I've seen this recommended and I plan on setting this up anyway but I'm wondering how meaningful it can be and what I should consider as far as domain names? Is the name important?

5. I will definitely be double checking all settings for a campaign to make sure everything is working before I launch.

6. How can I test with ZeroPark without absolutely destroying my budget in minutes?

7. This maybe ties into the previous question but should I be dayparting? I've seen 6PM to 12AM recommended as a general thing to do. Should I follow this or should I test an entire day at first. Maybe this depends on budget?

8. This is just a random thought but I assume that once I have a real testing framework down and the budget to handle it, way way in the future, it's not a bad idea to develop a bot to run testing and automatically optimize campaigns (as much as APIs will allow me at least) I wonder how much of a human touch is required at the top level outside of making creatives.

EDIT 9. Almost forgot this one. Popads has a ton of traffic for Brazil and I've asked for suggestions on similar "free lotto" offers in BRazil since I see a ton of landers for that in Adplexity. I think I'm going to test those out as well. There's no harm in testing more geos right?

Thanks for reading if you actually made it this far! I'm pumped to keep testing!

Also, I figured I'd throw it out there although I'm not expecting any free help regarding the latter: If anybody would like to start a mobile pops/sweeps mastermind with me, I'd love to do that. Working together seems to be the way to go in the industry and hopefully at the very least, this follow along will help someone else! I'm also very much open if someone would like mentor me/needs an apprentice (again, not expecting any freebie help like this but I figured I'd throw it out there). I'm a web developer by trade and my freelancing schedule means I'm super flexible as far as available hours go and how much free time I have to work on campaigns. The end goal for me is to definitely do affiliate marketing full time!

---

Hey sonicam! Nice to see you starting a follow-along!

Wow what a post! I consider myself long-winded and it looks like I've met my match. Will comment on parts of your post below...

There's no need to set up separate camps for each lander. In fact it would be better if you didn't - because when you're rotating everything evenly in the same camp, each test candidate will be subject to the same conditions (traffic from the same placements etc.) which will make results more fair.

Also - you're not testing enough landers! There are tons of sweeps landers on adplexity. Your offer has 4 "prizes" available to choose from - basically just replace the prizes on ripped landers with those (this will take a bit of time - and you'll need to replace the images and product prices and specs on some of the landers as well). I've shared some sweeps landers in this post - feel free to grab and modify them:

http://stmforum.com/forum/showthread...stakes-Landers

Speaking of which: Which prizes are you showing on your landers?

Another question is: You said you're testing other similar offers - what kinds of prizes are they offering? If you are to use the same set of landers for multiple offers, then those offers have better be very similar, i.e. pushing the same "prizes" (e.g. iphone 7).

Another benefit of only testing very similar offers: You won't have to compare offer+lander combos and cut inferior combos to arrive at the winning combo. Although this is the more accurate way (and the way it really should be done), this approach will take more money to arrive at a winning combo. If you're testing similar offers, you can cut landers based on lander stats alone, and cut offers based on offer stats alone, without considering whether certain landers may convert better than others for any particular offer. This approach will be cheaper and faster - and speed of execution is important when running pop because camps are typically short-lived.



How did you arrive at that decision?

To cut offers - please check out the tool described in this post:

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

To cut landers - use the split-test calculator at peakconversion:

http://stmforum.com/forum/showthread...Banners-Part-1

I used to recommend a minimum of 90% "probability of being best" for the winner, before cutting the loser. But matuloo convinced me that it's be better to run more data to make sure - so now I'm cutting when the winner reaches 95% or even 100% "probability of being best". You're encouraged to do the same, but 90% is basically the lowest you should go.

Also - before cutting an offer, drill down into major traffic segments, such as OS and carrier, to make sure that "loser" offer doesn't convert exceptionally well for certain segments that other offers aren't. e.g. Offer1 may appear to be the winner, but when you drill down into stats for Offer2 you find that it converts very well for IOS, or CarrierX. In that case you may want to start a separate camp for Offer2 to targeting just IOS/CarrierX.



Yup!



Don't worry about meddling with placements for now. Blacklisting the really bad placements like you've been doing is good. But focus on testing offers and landers for now. Changing so many things at once isn't recommended - too many moving parts.



100 seconds?? Or even 10 seconds? Are you SURE? I don't think I've seen a lander that slow!

You're right - caurmen has done testing to prove that at the 2s mark visitors will start leaving in droves.

This is YET another good reason to TEST MORE LANDERS!

What about your other 3 landers? Are they ALL loading slow? Just want to see if its your infrastructure or the lander itself.

If you like, you can always test the landers I've shared above - I can assure you that none of them loads nearly as slow as 10 seconds.



You mean for voluum? If so, then it's important. The default voluum domain has been flagged by many places as being shady (not surprising since many of us run aggressive/misleading landers), so you'll want to replace that with a custom domain ASAP.



Good - but everybody makes mistakes once in a while so no need to be hard on yourself.



Since you seem to be getting better results with popads, why not just do your testing on popads first, and then scale to zeropark later when you get green?

That way you can test for cheaper. Also: no point in losing money on both networks when you're still in the red.



If traffic volume isn't a problem and you're getting enough traffic to test your stuff at a satisfactory speed, then dayparting will give you more bang for your buck.

It doesn't matter whether these hours are speculation or not - once you get green you can open up to all hours, and cut hours based on stats.



It would take a very sophiscated bot to analyze trends in data (for example what I said above about checking how well an offer is doing for major traffic segments, before cutting it). So human decisions and analysis will probably still be necessary.

However, it wouldn't be difficult to build something to automatically cut placements or tweak bids according to a set of criteria.

Building a bot to regularly check offers and landers, and cut them as they reach stat sig shouldn't be hard to do either.



No harm whatsoever!



You could also start a thread in the masterminds forum section. If you do be sure to link back to this follow-along so people can see how dedicated you are. I know I'm impressed!



Amy

---

Very nice to see a bit of green! And this is before cutting OS and carrier and placements etc. Definitely very promising, and exciting to see!

That screenshot showing the 2 carriers - are those stats from "Mobile carrier" or "ISP / Carrier"? If it's the latter, I would also like to see the former.

If this trend with Android hogging all the conversions keeps up - by all means just target Android.

You don't even NEED a VPN to test this. Just make 2 versions of that lander - one with the prompt the other without - and use any speed tester from any test location, to see the difference in speed. Just keep in mind that what makes a lander convert well is more than just loading speed - at the end of the day, finding out which one converts the best is the most important.


*thumbs up*


You got it!

Having a process is especially important when you're running pop, because pop camps are typically short-lived. So you'll always be needing to test new stuff and optimize and scale quickly to replace camps that die.

The advantage of running pop is that with all the landers available for ripping, you don't even need to come up with your own angles. Don't get me wrong - testing new angles or even spin-offs of your competitors' angles can give you an edge, but unless you have a way of keeping your original lander out of spy tools, this edge will be taken from you within an hour.

So a good strategy for a newbie would be to take advantage of spy tools like adplexity - basically just rip as many different landers as you see, and test as many offers as you can find. Then pick the low-hanging fruit camps that are at least close to breaking even right off the bat (or are even green), so that you don't need to spend a long time optimizing to see green. Then scale hard and fast. Rinse and repeat.

There's not much marketing skills involved, but for new people trying to break into AM, pop is still the easiest to learn, and is certainly the cheapest (compared to native or FB for example).

Great progress - looking forward to your next update!



Amy

---

Yup you're absolutely correct!

For pop, all you need is grab offers from aff networks, rip some landers from adplexity, and start testing.

(Once you find promising offers, you can always test lander variations or completely original landers. But ripped landers are enough to uncover the best offers.)

DO split-test offers and landers from the start though. Rip all the different-looking landers you can find for that geo and offer type, and test them using 2-3 AM-recommended offers (that other affiliates are getting conversions with, because you need conversions to test landers!) Then take the winning lander and test every offer you can find on all aff networks you're with. That's the best way I know of uncovering promising offers to run and scale.

Networks you can consider joining that are newbie-friendly: Adsimilis, Clickdealer, Affiliaxe, Appflood. (Disclaimer: STM does not endorse any aff networks - please join at your own discretion.)



It may be worth breaking out MTN and Cell C into a separate camp - if there's enough traffic volume available to be worth your time. Can test bids, pick the best-performing, and keep cutting placements.



Well - yes and no. Depends on how important it is to be accurate about whatever it is you're cutting.

If you're cutting offers and landers, or even some of the big placements - stuff that can make or break your camp, you'd want to use stats. Cutting corners can really hurt you.

For the less crucial stuff such as smaller placements, you can use rules of thumb. Here's a tool you can use:

http://stmforum.com/forum/showthread...t-Cutting-Tool



The trouble with targeting just one or a few placements - as you've found out - is that the traffic volume and profits may not be worth your time in maintaining the camp.

However - 28k uniques definitely has potential! And yes, I would recommend to test bids to see how much profits you could be making. Because we're only talking about one placement, it would make sense to start low and keep increasing until you start to see a decrease in profits. Keep in mind that it's the amount of PROFITS and not the ROI that you're trying to maximize here.




Only one lander and one offer and you're already doing -40% ROI!

As mentioned above - I would strongly recommend that you test multiple landers and offers from the start.



This is a common misunderstanding: When evaluating targeting options that are NOT mutually exclusive, you don't need to cut down to a winner!

Because you can target any and all OSs simultaneously, you'll want to keep running all OSs that have the potential to make money for you.

Unlike for offers and landers, where you only get to show ONE offer and ONE lander to each visitor. <--- For these, you will need to split-test to find a winner (sometimes multiple winners - but I digress).

You can certainly cut an OS because it converts like shit. Or, target a single OS because it has the most promise and you want to get green faster, and/or test for cheaper. But -60% for Android compared to -74% for IOS isn't that much of a difference. So you're encouraged to keep running both in this case.



The "Other" category is actual wifi traffic - as opposed to the other carriers listed which constitute carrier or 3g/4g traffic (i.e. when users are using mobile data and not wifi).

In general carrier traffic usually converts better, but is available in a LOT less volume, and is considerably more expensive than wifi traffic.

Since Brazil has a lot of traffic volume, you could definitely start a separate camp to target carrier traffic, and have your original camp target wifi traffic. The carrier camp will probably get green faster, but if you can get the wifi camp profitable it will likely bring in more profits due to wifi traffic being available in large amounts.



All sounds great! Keep on keeping on!



Ay

---

Home > Mobile > Follow-along Campaigns