Home > Mobile > Follow-along Campaigns

---

My First Mobile Campaign - POPs for a Newbie (8)

---

Hi everyone,

This is my first entry into the mobile scene. I was first looking at doing mobile display and following the mobile cookbook, but then advice from here said that app installs weren't too great, but pin submits were in. Then based on my entrance budget, someone suggested I try antivirus.

I started looking at antivirus and thought POPs would be easier then banner -> lander -> offer, so I just went with it.

Here is the offer details:
Network: Adsimilis
Payout: $0.50
Traffic Source: ZeroPark
Country: Vietnam
SpyTools: Adplexity

I chose an offer that was less than $1 payout for cheap testing. I also chose a less popular country.

I used Adplexity to look at what landers were working. It seemed to be a lot of the same stuff: pages that made it look like the device was infected.

I ripped a few, and made some changes to make it my own. Then found out the offer wouldnt let me say your phone is infected, and I had to say your phone may be infected. But alas, I got the landers approved.

I had 3 landers approved, but I have only tested one of them so far (I want to find out if the offer is any good - am i doing the right thing here?)

I ran it for 2 days at $5 a day. Towards the end of the second day however, I got approved for another offer that was similar that I started to split test it with.

When I set the campaign up, I set it up as "Domain Redirect traffic" campaign. Pardon my ignorance, but what is the difference between this and PPV? I tried searching it and someone says that domain redirect is when someone parks a domain. That didnt really help explain what that meant though. Can someone just explain how a person would get to my lander from domain redirect traffic?

How would this compare to PPV? My thoughts would be that because I am going for a lander that appears like a system message, that a popup would be better but I am not sure. Could someone tell me if I'm right here?

When would it be better to use domain redirect vs PPV?

Offer 2:
Network: Adsimilis
Payout: $0.80
Traffic Source: ZeroPark
Country: Vietnam

Here are my results:

Visits: 10,842
Clicks: 1421
Conversions: 3
CTR: 13.11%
CR:0.21%

Revenue: $1.50
Cost: $10
ROI: -85%

I have since paused the campaign before I clarify a few things:
1) Should I be doing domain redirect or PPV for this type of campaign?
2) Should I be split testing landers at this stage? (I just wanted to know if the offer was any good)
3)Now that I have seen conversions, is that enough for me to see it is a working offer and go to the next stage, which is split testing landers?
4) I have some data from Voluum when I sort by target and I can see some sites getting lots of visits with no clicks, and some that have more clicks then visits. Should I be blacklisting these sites even this early on if I am seeing 2000 visits and no clicks? The same if i see double the clicks to visits (does that indicate bot?)?

Thanks guys.

---

Ok so a few things I have done since:

I have launched a campaign alongside this one for POPs. So now I have one for pops and one for domain redirects. I want to see which one does better.

I have added another offer, so I am split testing 3 offers right now.

I have also added more landers, so that I am split testing 3 different landers.

I have set the daily budget at 10x the offer payout (although I am not hitting this daily limit even though I appear to be in the top bidding position)

I still have no extra conversions from my data.

I have been able to blacklist some targets that have spent above 3x my payout with no conversions.

I have also created a whitelist campaign and added my three targets that converted.

At the moment I just need to run more traffic before I can cut more placements. I cant choose the best lander either because none have conversions since I have been split testing so I cant say what is better.

---

Thanks for documenting your campaigns in such great detail!

Good geo, and nice observation about the commonality between popular landers on adplexity.

Actually, it would be better to test a number of landers. If you only use one lander, and you don't see conversion, you won't know whether it's the lander's problem or the offer's. If you rip and test 5-10 landers, and you don't get conversions, then you can be fairly confident that it's the offer that sucks.

Furthermore, to increase your chances of hitting on a good offer, it would be good to test 2-3 offers recommended by AMs from different affiliate networks. Ask your AMs to recommend offers that other affiliates are running. The more affiliates there are running the offer, the more certain you can be that it will at least generate some conversions for you. Conversions will give you data so you can cut landers to identify the best one.

Once you have a winning lander, you could then use that to test more offers. Testing offers must be one of the best, if not THE best way to increase campaign ROI. Don't fall into the typical newbie trap of taking one offer and spending all your money and time to make it work - even if it means having to cut a sh*tload of placements and other stuff to get to green, and be left with a trickle of traffic in the end that will only make you pennies/day.



Awesome! Now we're talking!

Be aware though, that the split-test started at the time you introduced the second offer. Meaning, when comparing stats for the 2 offers, avoid using data collected by the first offer before you added the second offer. Otherwise the results will be skewed.

This is because the performance of a campaign will vary from day to day. Also, a campaign will often perform better in the very beginning, before saturation sets in. So comparing data gathered during different time periods will often not be the best approach.


AFAIK, domain redirect traffic is when a website no longer exists, so visitors that arrive at the domain are automatically redirected via a script (to our landers/offers).

I haven't put too much thought into the nature of domain redirect vs. PPV (pop). What I have observed at that based on stats, domain redirect traffic on zeropark is available in lesser volumes than PPV, and the former converts better which is why it's also more expensive (driven by higher demand).

So by all means give PPV a try as well.



For this camp, again, I would suggest testing more offers and landers to cast a wider net. Also, by testing more offers and landers and cutting down to the best combination, you have better chances of hitting on something promising, as opposed to just testing a single offer.

You can just take your current offer and split-test landers, but by doing that you'd basically be keeping your fingers crossed that you'll find a lander that can convert 700%+ better than your current lander. Not that that can't happen, but why not cast a wider net by testing a couple more offers?

Regarding bots - you're right - if select placements have really high or really low CTR, it could mean bots are at play. If you're getting a lot of traffic and you're not waiting impatiently for enough traffic test your offers and landers, then feel free to pause any placement that looks suspicious, or that are in loss. However, keep in mind that you would probably want to retest them once your campaign is green. A placement can only be as good as the offer and lander you're running. Just because the offer+lander you're running now aren't good enough to convert a placement, doesn't mean a good offer+lander won't.



I'll be on the lookout for your next update! Thanks for starting this follow-along!



Amy

---

Amy gave you a ton of quality advice so I'm not gonna interfere much, just wanted to add a thing or two.

- domain redirect, as Amy already said, really is traffic coming from sites that no longer exist. There are people who make their living by hunting for dropped domains - domains that somebody didn't want to renew or forgot to renew. Many times, these sites were still getting some traffic from search engines or from link trades etc ... or they were live sites that someone forgot about - so when someone picks up the domain and makes it live again, the traffic can start flowing again.

People monetize these sites by either putting a default search-like template on it or they directly redirect the visitors to a monetization platform, like zeropark for example. As you can guess, this traffic is a mega mix, people got there expecting to land on some site and they are redirected to something completely different - that's why the quality of this traffic will usually be very low.

PPV are usually Pop-ups or Pop-unders, you can read more about this traffic type here : http://stmforum.com/forum/showthread...inition-1-POPS

I would also suggest to read this thread, the method described in it should work well in this case too : http://stmforum.com/forum/showthread...-of-my-methods

All the best

Mat.

---

Here is my update:
I have continued to run traffic to my pop test and my domain redirect test.

My domain redirect campaign has done better then my popup, so i have made a whitelist for that and this is what is happening there:

What I am now doing is looking for a conversion and then adding it to a whitelist campaign. I now have 4 targets in that.

I am also black listing any placement that spend 3x the payout with no conversions (that is slow however).

My whitelist campaign is already running OK with a -75% ROI (which is beter then the other ones). I have my 3 landers and multiple offers running still.

I am starting to look at some optimization now, specifically landers. Here is my analysis. Please let me know if I am on the right track or if I am too early:


Now when I run the following numbers through a calculator (http://www.splittestcalculator.com/) it says there isnt enough data, so I havent eliminated a lander yet. I did realise I forgot to add a lander into my whitelist campaign though so thats why there is only 2.

What I am a bit worried about is one lander is getting more clicks then visits (even though its on my whitelist campaign). What should I interpret from this? It appears my CR is lower, but if I go on visits, it is about the same. Should the placement with higher clicks then visits not be in my whitelist campaign? Is it bogus?

I have looked at my 3 offers here:

Now from a statistical significance point, it says there is no winner when comparing Offer 1 and Offer 3, but I have spent 6 times the payout and seen no conversions, so I am stopping this offer and will put a different one in. (I think Im doing the right thing)

Actions
1) Add a new lander to my split testing
2) swap out the bad offer with a different one

Thanks guys.

Oh btw, this is fun!

---

Ok, hoping to get some opinions on what I should do here.

Here is my campaign stats:


Here is the stats from Voluum (i havent put the cost in there because it wasnt the same as zeropark):


Now, I have been blacklisting targets when they hit 3x payout; i have been testing through 3 landers (none of which show a stastically significant benefit over another) and I have tested 4 different offers, and eliminated one offer from these. I have used the most commonly seen landers using spytools, and modifyng one of my own.

I have tried running different bids with this too. At first I tried running bids WAY above the recommended bid so that I would get the best traffic, but even with that I only got 5 conversions!

I have spent $85 and only made $7. Even my best campaign (my whitelist campaigns) had ROI of -82% and -90%. I mean on this forum I have read people say if you start at -80% then you might be able to get to profit with some optimisation. Well I have been doing optimisation (of placements, landers and offers) and only could get to -80%.

I have some thoughts I would like the more experienced opinions on please:
- When I was looking at spytools for VN antivirus pin submits, I saw there wasnt any long term campaigns running on zeropark (the network I ended up using). Is it likely that there wasnt anyone running on this network because its traffic is no good for this type of offer?
- I have read people say that if you see an offer being run, its likely a good one because people wouldnt run it if it wasnt working. For the future, if I dont see an offer is not running on a particular network, should I not start with that network. (I understand I can try and scale it horizontally later on, but maybe I shouldnt start there??)
- Should I try another traffic source with this? I mean I have seen that there is a particular offer that is converting, but its just not working well on this traffic source.
- Did I spend way too much on this campaign then I should have? I was cutting placements quite soon after they hit 3x payout, but because there was a lot of placements, all getting under $1 of traffic, it meant I didnt have enough traffic to cut the placement, but it was chewing up the cash.
- The other thing I did with this campaign was to copy what I had seen on spytools for what to target - i.e. I saw most of the campaigns targeting Android and only doing carrier (not wifi). Should I have run these split tests myself rather then starting my campaigns with this targeting?

Thanks

---

You don't need to start a whitelist when you don't even have profitable targets. The whole point of having a whitelist is so you could bid higher to get more traffic from better placements.


Too early as you've found out by using the split-test calculator.


Some landers have backbutton scripts and auto redirects that can lead to there being more clicks than visits. CTR shouldn't be considered when trying to decide which lander to keep - ROI is the key factor here. You're aiming to make money, and a high CTR doesn't mean a high CR.

The same applies when trying to decide which placement to whitelist.


Actually if you compare the "current-best" offer (offer2) against the other 2 offers, you'll find that offer 3 is ready to be cut. To learn how to cut offers properly, use the tool as described in this thread:

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

When cutting offers/landers, use this method:

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

Do spend some time to learn how to cut landers and offers - this is important!


When split-testing, you would ideally keep cutting until you have a single winner. Exception is when you have 2 candidates that are performing around the same - then you would either keep them both or just randomly pick one. Then you would start another round of split-testing with the previous winner as the "test control", by adding new candidates.

(Caurmen talks about what to do when you have multiple landing pages that are performing the same, in this thread here:

"3. If you've had a few hundred clicks worth of data for each landing page, all of them have probabilities of being best between 25% and 75%, and you're seeing high, overlapping graphs like the ones below, the remaining landers all perform about the same. Pick one to continue running, pause the others, and start a new split-test")

This applies to offers as well as landers. Don't arbitrarily pull things out and add things in. Cut offers and landers as they reach statistical significance, and only add new candidates to a split-test after the previous round of testing in over and you're left with a winner.

And yup - if you're having fun, then you KNOW you're headed in the right direction!




Amy

---

Oh boy! >_<

I have no comment for your first two questions, because I'm not good at doing intelligence gathering. I mainly focus on testing offers to find ones that have enough potential to optimize and scale.

As for your last question, just because SOME offers do better on android and/or certain carriers, doesn't mean all offers will be the same.

First of all, how many bids did you test? For next time, I would suggest testing 3 bids - low, average, high. But any way you look at these stats, your landers and/or offers aren't working. When you're at such a negative ROI, you just can't rely on cutting placements to get you to 30%+ ROI. Not that it's impossible, but it will be costly, and you'd be way better off putting that money on testing more offers and landers instead. Here's a very rough test approach you can follow until you develop your own:

1)For a specific vertical+geo, ask AMs from at least 2 aff networks for recommended offers. Get 2-3 offers.

2)Rip 5-10 landers you see are quite popular / frequently seen.

3)Set up 3 identical camps with staggered bids - low + average + high - and run 10-20x payout to each camp.

4)Drill down into offer -> lander in each camp to see how many combos are green. The more green, the more promising the camp. Can also drill down further into major traffic segments such as android/ios. All you need is one profitable offer+lander+major traffic segment to have a profitable campaign.

5)If you conclude that the camp doesn't have promise (i.e. no green combos that are green or even close to breaking even), then either just end the camp or replace the current offers with new offers and repeat the test. If you feel that the camp has potential, then pick from the 3 camps the one that has done the best ROI, pause the other two, and continue running the remaining camp to cut offers and landers.

6)Once you're down to a winning offer and winning lander, test more offers. I like to go to all my favorite aff networks and grab all offers available for that geo+vertical and test them using the winning lander. Finding a good offer is key. I don't understand how most people would test thousands of placements willingly, and yet be so reluctant to mass-test offers. Generally speaking, finding a better offer will likely increase your ROI by a lot more than cutting placements could.

And of course you will be cutting stuff like placements in this whole process, but we'll cross that bridge when we get there. Please feel free to ask questions. It will be easier to suggest a placement-cutting approach once I see some stats.

The gist of the whole test approach is to not spend too much money on testing any one offer, but instead to test more offers. The first round of testing is mainly to a)identify a good lander which you can then use to mass-test offers in the second round, b)see if the offers even have potential at all, and c)identify the best bid to run the campaign at.

It's a great geo you're targeting! I would suggest to go through the steps above for the same geo and vertical. Alternatively of course you can pick another geo and/or vertical. Rip more landers, and get some new offers, and get testing!




Amy

---

Home > Mobile > Follow-along Campaigns