Home > Mobile > Follow-along Campaigns

---

It's not about the destination, it's about the journey (59)

---

Started off with Caurmen's mobile cookbook. I was looking at F5 and Adsimilis for offers and contacting the AM for a list of top offers. I was overwhelmed by the list of offers to choose from, I started out making multiples banners for different offers and losing focus. So, I decided to pick one of the top offers and work on it until I become successful with it. I don't want to put my energy into multiple offers instead I should pinpoint all my energy into one offer until it becomes successful or the offer becomes no more. So, let's get on to it:

Traffic source: Decisive
Vertical: Sweepstakes
Network: Adsimilis
Offer: Win the New Iphone6
Geo: Italy

Budget: Unlimited, but daily spend limited to $8
Payout: .50
Bid: .75

Preface: Currently, I’m direct linking the offer as stated to start out in the Appetizer section. I need to make at least $7 in payouts in each campaign to move on to the next stage. I already ran 4 campaigns that past the $15 point and none of them made the cut, so I’m trying again. Below are some stats:



Based on these stats, I think I need to increase my bid to gain more “wins”. The CTR for Iphone6DecAd19571ITMSite_v2 seems really high and no conversions which makes me wonder if it is my banners having issues. At this point I’m not sure where to go, but keep making new banners for my campaigns until I succeed. I know Caurmen said try 2 times before going on to a new offer. But, it seems like this offer is doing well according to the AM, so I’m going to stick for it a little longer.

What do you guys think of my banners?

---

Hi Simon! Way to take action!

Props to you on your decision to stick with one offer! Right now the most important thing is to learn the entire process of setting up a campaign and analyzing stats etc. Once you have all your basics down solid you can test a variety of offers.

For sweepstakes I would suggest that you pick a geo you know the language for. If you don't know Italian and want to give an english-speaking geo a try, PM me for a good offer recommendation.

Caurmen's offer cut-off guidelines are great, especially for offers that are somewhat unique such that if they die, you won't be able to easily find a replacement offer for. In that case, you'd need to test and scale very quickly before the offer gets pulled (which happens to mobile app download offers very frequently, many after a short time). However, for an "evergreen" niche such as sweeps where once you have good creatives you can potentially reap rewards for a long time to come because there will never be a lack of sweeps offers, I would focus 100% on optimizing my creatives and not on ROI in the beginning.

After a couple more days, you can pause 3 of your 4 camps and just leave the best one running - you can use that to test your banners and landing pages.

At this point, I would suggest for you to keep testing banners - lots of angles, lots of designs - just throw anything you can think of at it! After a couple more rounds of banner testing, only leave your best couple of banners running, and start testing landing pages.

While you're testing, keep an eye out for placements that are draining your budget. I have a spreadsheet that I use for cutting placements, but right now you don't need it - because you need as much traffic as possible to speed up your testing. I would just pause placements that have spent say 3-4 times your payout without a conversion for example.

Also, I like to blacklist the following placements from the start:

1)Grindr, and all flashlight apps.

2)The apps and sites that are on these Decisive Blacklists. (Note: Although these blacklists are from Decisive, you'll see similar placements on Go2Mobi because at least some of the exchanges they source their traffic from are the same.)

There are a lot of stackers doing the same thing as you at the moment, i.e. sweeps on mobile networks. Seek out all the follow-alongs you can find and see what they're doing, and learn from their experience.


Thanks for a great start to a great follow-along! Will be on the lookout for your next installment...


Amy

---

Based on your suggestions vortex, I started another campaign within the same vertical that allowed me to create banners in my language that I understand. I’m still Still waiting for Mundo Media to approve my application. The payout is higher, but I’m trying to understand the quirks within these campaigns that I just started.

Traffic source: Decisive
Vertical: Sweepstakes
Network: Adsimilis
Offer: Win the New Iphone6
Geo: Australia

Budget: Unlimited, but daily spend limited to $8
Payout: 2.40
Bid: 3.60

You mentioned that I should focus on finding good creatives in the beginning. By creatives, do you mean the banners that I create? Is there a guideline in which banners I should keep and which ones I shouldn’t? Should I look at each individual creative and focus on the click percentage? The reason I say that is because Caurmen’s guide mentioned that I should wait till I spend $15 on each campaign and then if none of my campaigns made less than $7, I should pause it. So, based on the current campaigns I’m running, it doesn’t seem like I’ll run into making $7 per each campaign.

In terms of the blacklist process, do you blacklist the sites/apps before the campaigns even start? I took a look at Mr. Green’s post on Decisive’s blacklist and tried to export the top 100 blacklist app/sites onto Decisive’s Batch Blacklist. But, it seems that it doesn’t work too well. So, I’ve been blacklisting the placements manually. Based on the screenshot below, I’m not sure if I should blacklist those targets even though it doesn’t fully match up with the name.



Here’s my progress on my campaigns so far…. I started running this campaign yesterday:

---

Hey Simon,

use this tool to decode the excel list: http://meyerweb.com/eric/tools/dencoder/ . After that there should be no problem. You can use this for both lists.

Wish you good luck. I'm on a same journey

Best regards,
Philipp

---

By creatives I mean banners and landers.

In the beginning, brainstorm some angles and create at least 3 banners per angle. The more angles you test, the more likely you'll hit on a good one. Once you find a good angle, make more banners and landers for that angle to increase performance further!

Caurmen's guide is wonderful! However, I think that when he wrote it, there were a lot of good mobile offers around - in which case it would be sensible to cut all except ones that appear promising from the start! However, you're trying to do sweeps - it's not easy to get profitable in this vertical without a lot of testing, but once you do get profitable you can scale this to a ton of offers in various geos! So like I've said before: don't fixate on ROI just yet! Focus on improving your creatives, then scale fast once you find something profitable, and use the good creatives to mass-test offers and geos as well as other traffic sources to identify where the gold is.

From looking at your data, it's apparent that your mobile camps are converting better than your wifi camps. I would let the 4 camps collect a bit more data to make sure, and if the trend remains the same I would pause the wifi camps and just use the mobile camps to optimize creatives (you can even wait to see which one of the apps or sites camps for mobile performs better, and pause the underperforming one and just use one camp to optimize creatives - doing so would certainly be less confusing for you).


Looking forward to your next installment...


Amy

---

Hey Stackers!

I wanted to provide an update on my campaign progress.

1) I have paused 2 campaigns that had a total of 16 different banner designs.
2) Based on the performance of those campaigns, I took 8 of the best performing banners either by the amount of conversions each banner had or the highest CTR.
-Btw, all these campaigns flopped.

Gameplan

- I have yet to create a lander for these campaigns to test. So, taking the 8 best banners I’m testing the particular “win iPhone6” offer across various affiliate networks ie Adsimilis, F5Media, Mundo Media.
- After testing network works the best with the iPhone 6 offer, then I will create 2-3 landing pages in rotation to see if it increases conversions.
- That’s my plan for right now, I’m not sure if it’s the best plan, but that’s all I got for now.

Because of the new execution of I might have to get some new tools like Whatrunswhere/Mobile Ad Scout, and get tracking software from Volumm to track landing page performance.

I’m sticking with one traffic source(Decisive) at the moment because I don’t want to test too many variables at once. It gets me distracted. Instead if these campaigns flop, I might transition them over to zeropark to see how it performs.

Stay Hungry Stackers!

---

Welcome. It is about the destination because thats what buys you stuff and pays for your travels

When running an offer take the offer url plug it into wrw and see what were the top 2 or 3 traff sources all time. This is a good way to start as they serve as an indicator where the offer ran the most thus where it was profitable.

Good luck.

---

I have ran the network's offer link into wrw and it doesn't seem to have enough information about it. Based on the prevalence(1) and duration(1) of the information it seems that it's a offer that doesn't get too much traffic. Therefore, not making it a hot offer as opposed to something like Candy Crush Saga.

---

Check the top sites in similarweb that send traffic to offer url and then plug those into wrw

---

Stackers! I hope you're making 10 bands, 50 bands, 100 bands that you forget which bands you're on . For any hip hop fans, they'll know where I got that lyric from.

So, I know I haven't posted a while because I was still learning about the different concepts I'd have to grasp for affiliate marketing. I switched over to a new traffic source because then when I can become profitable with a pop traffic source, then I could switch over to a display network like Decisive. Let's get to it!

Traffic source: ZeroPark
Vertical: Sweepstakes
Network: F5 Media
Offer: Iphone6
Geo: Australia

Campaign Budget: $100
Payout: 2.40

I ran this campaign for one day. The campaign budget was set at $100 because for each lander I wanted to check it's performance either 10-15x the payout before I make a decision on the campaign. I tested 4 landers to one offer and it is apparent which lander did the best.

Campaign/Landing Page:



Things I noticed:
1) I started this campaign early around 9:00 and there wasn't really much conversions coming in until the afternoon around 13:00 - 18:00. It seemed like the offer would perform best around the early morning in Australia from 4:00 AM - 8:00 AM.
2) Based on the landing page performance, it looks obvious that landing page 4 performs the best and I would be using that page to make variations of it.
3) With the amount of visits I received, I feel as if my CTR is really low. Maybe I should populate more elements within my mobile landing pages to make people stay for the offer. I heard one way to make users stay on the offer page longer is an exit pop script. Is there a good exit pop script that I could use for these mobile landing pages?

ZeroPark Placements:


Things I noticed:
1) Based on the first placement you see I spent around $31 for that one placement. I read one of Ruby Tune's post about cutting a placement after it reaches 2-3x it's payout. But, since I'm in the initial testing phase I didn't want to bother with the data too much and just let it run.
2) The thing I did adjust was my bid position, I noticed my position would sometimes fluctuate between position 1 & 2. What that meant to me was that someone else was competing for traffic, so I was constantly monitoring my bids every 30 minutes or so today to make sure I'm the number 1 bidder. This maybe a bid excessive and I should probably avoid that next time. What do you think?

Next Steps:
I'm not sure about which step would be better and was hoping the STM community could provide some insight.
1) Based on landing page performance, create 3 slightly different variations of landing page 4 by changing some of the elements such as CTA button(looping, color scheme change, font size) or change the background color of the page or replace the image. Right now, I'm planing to change each page by adjusting the CTA button and leaving everything else the same.
OR
2) Keep the 4 landing pages regardless of performance, and for the next test, test more offers along the same realm(iphone6 sweeps) and see which offers come up on top. So it'll be 4 landers with 4 offers.

Any input is always appreciated!

Thank you for your time for reading this!
Simon

---

What's up Stackers! Here's another update on my campaigns!

Still working on my sweeps campaign. At this point I'm focusing on a particular iphone sweeps offer on F5 before jumping onto another offer. I know the best option is to work on a hot offer, and I'm not sure if my offer is considered a "hot" offer. But, it seems to be converting better than the different offers I tested on other networks. Thanks to Mr. Greens previous post on sweepstakes. I learned several things, but the most I got out of it is that advertisers set caps on the amount of leads that affiliates can generate and that by the end of the month I'd receive fewer conversions. So, I experienced something like that with a iphone offer on clickdealer, where the webpage was redirected to a coupon site instead of the iphone sweeps towards the end of the month. Alright, so let's get on to the campaign...

Traffic source: ZeroPark
Vertical: Sweepstakes
Network: F5 Media
Offer: Iphone6
Geo: Australia

Campaign Budget: $100
Payout: 2.40

Taking my existing campaign, I tweaked a few things around.

1) Based on the previous campaign performance, I took the best performing landing page and created 4 different variations. For each landing page, I changed 2 elements.
- CTA Button - I turned the button into a animation which would loop around 5 times.
- Headlines - I changed the headlines to see if there was a change in conversions.

Here are my headlines:

- "You have been selected to take part in our monthly contest. Please choose a gift you'd like to win." Landing Page 1 Headline
- "We're giving Australian residents a chance to participate in our monthly prize drawing! Please choose a gift you'd like to win." Landing Page 2 Headline
- "Every day we choose a user to take part in our monthly drawing! You've been selected today for a chance to win one of our prizes below!" Landing Page 3 Headline
- "Your phone has been randomly selected for a chance to win 1 of 3 Apple products below! Please choose a gift you'd like to win!" Landing Page 4 Headline





Things I've noticed/did differently:

1) LP 1 was my winning lander from my previous campaign and through peak conversion, it seems that it still ends up being the best lander. However, LP2 has a higher amount of clicks and same amount of conversions as landing page 1, it is considered the last lander I should keep. This is interesting to me because LP2 has the most clicks which makes me think because I targeted Australians by using that word in my landing page that it enticed more people to click.
2) I didn't touch my bid price, I just bid 30% more of the top bid. This advice was given in one of iAmAttila blog post about pop/redirects. So, I'll stick with the bid price and work on landing pages till I can turn the negative ROI into a positive ROI.
3) I also tried to optimize my landing page speed. I already implemented a CDN onto my LPs and read another one of iAmAttila's blog post about compressing gzips and that has also reduced loading time.

Next Steps:
1) I'm pretty optimistic that I can turn this into a profitable campaign, I just have to find ways to tweak around certain elements. The next step is I'll take landing page 1 and change the arrangement of the pictures.

If there is another way I should approach this, I'm all ears

Any input is always appreciated!

Thank you for your time for reading this!
Simon

---

Wow - awesome follow along! I really like your observations!

Some comments on your last two posts:

-To cut landers: You're not using the split-test calculator correctly. A lander can only be cut when its "probability of being best" (as compared to the current-best lander) falls below 10%. Please see this post for more details. (However, you are correct in your observation that LP4 in your first campaign was the best with 7 conversions.)

-For exit pop, stick the following code at the bottom - perhaps just before your </body> tag:

<script language="JavaScript" type="text/javascript">
function exit_a1(){alert("Your prize has been reserved! Enter your info on the next page so we can send it to you!");}
</script>

Then inside the code of your CTA where you want the pop to appear when the visitor clicks on it, add:

onclick="exit_a1();"

-In your first campaign's placements stats - that first placement should have been paused long ago! $31 spent $4.80 made - that's such unnecessary drainage of budget. Because you're in the testing stage, it's important not to limit traffic, but at the same time you don't want to drain money needlessly, so be sure to pause the worst placements right away. You can always re-test placements later when you have optimized the rest of your campaign. Instead of wasting that money on the first placement, you could have increase the bid for the profitable placements.

-It would certainly be a good idea to test offers as well! I would recommend that you start a separate campaign for this purpose. Use your best lander and mass-test all iphone 6 offers for your geo that you can find across different aff networks.

Some ideas for your lander testing:

-Try other angles and write completely new copy.
-Try different themes - the fb theme for example has been done to death. Instead, try system messages, whatsapp, snapchat...
-Call out user information (such as geo, isp, browser, device, model...)
-Use testimonials (don't copy from ripped landers - use new headshots and comments, and names that sound local to the target geo)


Very impressive progress! Looking promising!



Amy

---

Ah Amy! I should of read your cutting landers post more thoroughly. I will have to revise my game plan.

- Thanks for the exit pop script, but is there another exit pop script when the user is trying to exit my landing page? I think the one you gave me will automatically redirect them to the new page and I think users would not lose interest that much.

Revised Next Steps:

1) Test the best lander vs the other landers one by one. At this point I set a budget of $30 dollars to each test. So, I'll be spending $120 for 4 tests. I'm not sure if it's that important to test which lander is better based off of the headline. In my landing page, I just changed the headlines and didn't change the font/size/color. The CTA button is the same across the four landing pages. I think doing that kind of test is only recommended if you revamp a noticeable chunk of the page. Can you chime in on this Amy?

2) So, I was stuck today about what to change about the landing page(fb win type of lp) till I thought of mimicking the interfaces of popular social media. So, what I did was google "Australia's top social media usage" or something along those lines and found social media interfaces to input onto the top of my landing page. This will be interesting! I decided to use interfaces of snapchat, twitter, tinder, and youtube.

3) Is there a cumulative list where all the scripts are to call out user information?

Thanks again! Will post another update soon!

---

Hmm...not sure what it is you're looking for. Could you please elaborate? Are you asking for a script that will redirect the visitor when they click the back button? If so this will work:

<script>
history.replaceState(null, document.title, location.pathname+"#!/remover");
history.pushState(null, document.title, location.pathname);
window.addEventListener("popstate", function() {
if(location.hash === "#!/remover") {
setTimeout(function(){
location.replace("http://wherever.com");
},0);
}
}, false);
</script>


Oh - you don't need to only test 2 landers at a time! You can test as many as you like (but keep in mind that the more you test, the longer it will take for some to reach statistical significance so you can cut them). However, when COMPARING them on the split-test calculator, you need to compare each lander to the "current-best" lander and only cut a lander when the "probability of being best" falls below 10%.

When it comes to testing landers, always start by testing completely different designs / angles - i.e. the "big" things - then move onto the "smaller" things. That way you cast a wide net and gradually zero-in on the best optimizations, by optimizing aspects that will have the biggest impact on conversion rates first. If your headlines are significantly different, I would say go for it - test them at this point in time. And your decision to keep everything else the same is a good one - if you change more than one thing you won't know which change will be possible for the differences in conversion rates.


*Applause!* You're DEFINITELY on the right track!


This tutorial by Caurmen will tell you everything you need to know to set that up:

http://stmforum.com/forum/showthread...ing-Parameters


This is getting fun!



Amy

---

Brief Updates on my current campaigns:

I've paused my campaigns due to the lack of volume it receives in Australia. The offer I'm promoting around several networks claim that the volume is pretty low. I read a post from Malan Darras that said something along the lines of I rather be a small fish in a big pond. That's where I'd like to claim that piece of the pie!

With that being said, I have asked the affiliate networks I'm working with about the type of hot offers that they have. Here are some of the questions I asked them:

- Is there any hot offers out there specifically in the iphone sweeps arena?
- With that hot offer is mobile traffic considered the top performing traffic source for that offer?

So based on those questions I received a bunch of USA offers that also comes with pretty strict guidelines on how the creative(LP) should look like. It was tough not able to use certain words like "win", "gift", or "prize". Another rule for some of the offers is that you can't provide testimonials unless they're real. My first thought, how am I going to attain something like that? How am I going to seek true testimonials, lol

I sent my landing pages to affiliate managers to approve and it's been tough trying to stay 100% to the guidelines.... Also, waiting for affiliate managers to contact advertisers to LPs take time, so I'm facing downtime by not starting any other campaigns since I want to be focused solely on iphone sweeps.

But, I digress so I wanted to upload my landing pages here to see what you stackers think of it and what I could improve upon.



Notes:

- For these landing pages, I wanted to mimic the style of mobile apps social media interface. Let's see if you can distinguish between the different social media interfaces

Tracking:

I looked at my previous campaigns and how I was tracking conversions and it looked like I was doing it incorrectly. This process dawned on me when I read a post about CTR being incredibly high using a back button script.

After a few tweaks, I believed that I fixed it. However, I want to see the difference of CTR when a user actually clicks the CTA button on my lander vs. the back button script that will redirect the user to my another offer. I'm honestly stuck on this part...

Is my volume tracking link supposed to look like this?

In Voluum Offer page area: http://adstrx.com/?a=xxxx&c=xxxxxx&s1={clickid}&s2={campaign.id}&s3= {var1}

In postback URL area of Adsimilis: [url]http://xxxx.voluumtrk2.com/postback?cid=#cid#&campaignid=#campid#&var1=#s3#



Based on the screenshot, I labeled it correctly on the 4th row and I got it to work. But I'm trying to append a 3rd sub id to find the difference of CTR between CTA on lander vs. backbutton script.

Additional Questions:

1) When you send affiliate managers your creative landing pages, do you just send them screenshots of the landing page? Or do you give them your landing page link? The reason why I'm asking this is because I read something about double meta refresh being able to hide landing pages from AM so that they won't take it and reveal it to other affiliates.

---

I've been running sweeps campaigns for a while and after attending affiliate summit and received numerous tips from individuals I have gained more insight on how I should run my campaigns.

Campaign Process

Traffic Source: ZeroPark
Offer: Iphone
Affiliate Network: Adsimilis/F5 Media - Leaning more towards F5 Media
Budget: $20-25 per lander.

1) Scraped 11 landing pages to test - Didn't test them all at once though.
2) Picked two to three offers from the network mentioned above and at the end picked one specific offer to promote from F5.
3) Tested several landing pages at first to see which one performed better than the others and then took the winning lander to test against new landing pages to see which lander performs the best.

Next Steps



1) I plan to take landing pages 3,10, and 11 because they all have a ROI less that -50% ROI.
2) I'm also planning to use these landing pages and test them against new offers since the iphone6s release date is tomorrow.
3) Testing LP(3,10,11) with the targets I've whitelisted and not with the targets that I've blacklisted due to amount of spend on each target.

What are your thoughts?

---

Nice plan! And using best-converting targets for testing is an awesome way to keep testing costs low. This is assuming you're getting enough traffic from the whitelisted targets to not have to wait forever for each round of testing to yield results.

One concern though regarding the way you're choosing "best landers" - basing that on ROI alone is not the best approach, because it doesn't take into account how many impressions and conversions you have on each lander at the time of comparison, i.e. you don't know if you have enough data to make a sound decision. For a straight-forward method on how to compare and cut landers please see:

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

I spent a couple of minutes to plug your lander stats into the split-test calculator at peakconversions - and the second lander on your list is clearly the winner - with 99% probability of being best when compared to the other 2 landers you selected:



(I've gone through all the landers on your list and the second lander is the winner amongst all - not just the other 2. Please feel free to verify this yourself. )

This is not to say that you should eliminate all other landers from further consideration/optimization. It really depends on HOW YOUR LANDERS DIFFER. For example, if you've ripped 10 landers from spying tools and they're all very different in their lander styles/types, after gathering some data you may want to pick the SEVERAL most promising lander types to further optimize. On the other hand, if you're in latter stages of lander testing, say you're testing 10 landers that are exactly the same except for the headline, then you WOULD probably want to keeping cutting until you're left with only 1 winner.

Hope that helps! This may seem trivial, but knowing how to cut stuff based on statistical significance is extremely important, because if you make decisions before statistical significance is reached, you may cut something that could actually emerge the winner in the long run. Also, by cutting things based on statistical significance, you may be able to save on testing costs by not letting your test variations run longer than they need to.


Amy

---

Campaign Update!

Taking Amy's advice I am currently running one LP with one offer that is prevalent than all the other offers that I'm working with.

I have also narrowed down the targets on my traffic source now. I'm running my current campaign with targets that have at least one conversion and excluding anything that doesn't have any conversions or the spent amount exceeded 2-3x offer payout. I'm looking at a -30% to -40% ROI. But at least it's not -60% to 80% like it used to be .

I think I'm bidding aggressively on my traffic source.




Next Steps:

1) Spying on profitable targets to see what new landing pages are showing up.
2) Taking those ripped landing pages and testing on those targets
3) Once I hit a ROI of -20%, I will start to make variations of the lander and retest on those targets.

What are your thoughts on my next steps?

Question:

1) When I test landers after running the campaign for a while, I don't know what's the optimal budget to test for each lander. So let's say I ran a campaign for 7 days and I got some significant data, what would be the testing budget for 1 LP at that point when I accumulated so many targets from testing?

---

Hmm...why would you set a budget for 1 LP? Aren't you just plugging lander stats into a split-test calculator and cutting inferior landers?


Amy

---

Because after running 10 landers and see that only one lander has the best CR/ROI(even though it was -Roi) I'm using only that lander to test on the traffic source. Now only using one lander I want to see what budget I should set for it. But I'll post about the progress I made with that one LP.

---

If you're only down to one lander then what are you testing right now?


Amy

---

I took that one lander and created a separate campaign on zeropark that had targets with at least 1 conversion and didn't spend over 2-3x the offer amount.

So right now, it's one lander running one offer.

Next Steps:
1) Spy on the targets that has already had conversions to see what landers are on there.
2) Take those landers and setup a new test with the 120ish targets.
3) If one lander reaches -10% ROI take that lander and start optimizing till I get a positive ROI.

Here's the progress:

Zeropark Targets Campaign: Currently have about 120-125 active targets.

---

That's a fantastic plan you've outlined! Eager to see results.


Amy

---

Update!

Tried to spy on the targets this weekend, but the web links I got were pretty obscure. Some of the web links looked like this(i.e. fasd3fdsf.com, easkdfu2.com, etc....). However, I understood the general gist of what traffic I'm receiving. It seems like I'm getting a lot of torrent/downloadable music/movie websites. Looking at all the sites I spied on there wasn't too much content in terms of landers I could grab.

So, I just went another route. I read a post by Malan Darras(10 Ways to Optimize a Landing Page) recently and with his insight I adjusted my best landing page by changing the headlines, body copy, and the main image. Some of the ideas I came up with is from googling popular things in Australia.

I've complied a list of ideas that I could write about that would pertain to Australians:
- Australian Prime Ministers
- Vegemite
- Timtams
- Neighbours
- etc.....

With this list I started developing new landing pages off the existing lander that wasn't profiting but it was at a -20 to -30% ROI scale. I created 3 new landing pages each with a testing budget of $20 dollars, total of $80 including the original landing page.

It turned out that the original landing page was the 2nd best performer and L10_v3 turned out to be the best lander. Here are the results:


As of right now, I think I'm going to use the prevalent lander and go back to the drawing board/Malan's post and find other things I could change around.

Questions:
1) When you have a specific whitelisted placements campaign in ZeroPark do you start bidding agressively on a specific placement? i.e. Let's say I have one placement that has the most conversions but my bid position is 3, should I increase the bid on that one manually? Should I just leave my bid as is?
2) Am I putting too much budget on testing landers at this point? Since I have made a campaign that solely consists converting placements, is it safe to say that I need a higher testing budget?
3) Should I start working on different traffic sources? And if I do, should I start from step 1(i.e. testing 10 landers then find which one is the best)?

---

I think your lander testing plan is great! Ripping may be the quickest and easiest way to establish an initial benchmark, but I find that by writing my own copy and putting my spin to a lander that's already working, I can almost always make it better.

However, for future reference: There are plenty of spying services, such as Make Massive and Mobile Ad Scout (and a plethora of new ones have either just come out or will be coming out soon), that will save you so much time. I know there are new tools coming out that will be cheaper and will come with discounts for STM members - so stay tuned.

1)Unless you see a REALLY good placement, I would suggest that you focus on improving your "funnel" at this point, which is your offer + lander combo. Doing so will most likely bring you the biggest increase in ROI. When testing, always try to tweak the variable that has the most potential of yielding the biggest increase on ROI first. Having said that - the only way to find out the effects of bidding higher for that placement, would be to test. As long as all your landers are receiving equal impressions from that placement, tweaking the bid on it shouldn't skew your test results.

2)I don't entirely understand what you're asking here so please feel free to elaborate. Using a camp comprised of whitelisted, well-performing placements to test landers is an awesome way to keep testing costs low. Why would you worry about having to allocate a higher test budget?

3)As strange as it may sound, when you test the same batch of landers on different traffic sources, the winning lander will often be different! It may be because every traffic source has different placements comprised of different types of sites that attract different demographics, and some landers just appeal to a particular demographic more than others. The safe thing to do would be to re-test all the landers you've tested in the first traffic source, but I get lazy so I just pick the few most promising landers to test on subsequent traffic sources. Then, every new lander I make after that, I'll rotate it into all traffic sources to split-test. (Sometimes I get lazy and just rotate new landers into 1 traffic source, and only if it ends up being the best will I rotate it to other traffic networks. May not be the best way to do things though.)


Amy

---

I noticed that as well by setting up a whitelisted, well-performing placement campaign that my budget doesn't burn that quickly. But here's what I meant, suppose when you start a new campaign the general guidelines for testing budget would be for each lander that you test you would allocate 5-8x the offer payout. So for instance let's say I have 5 landers and the payout is $3. Then I would have to allocate $75(5landers*3offerpayout*5). With that being said, since I developed a new whitelisted campaign after testing for a while do I still need to allocate that much amount of budget for the amount of landers I test? So if I test a winning lander, would I set a higher budget for it or a lower budget?

I also looked over at another follow along that you posted http://stmforum.com/forum/showthread...t-got-Confused and I feel lost by not having a system. But, the post you made over there gave me more clarity on how I should approach my campaigns in a systematic manner. So, I think I'm going to start back from scratch and carefully detail my progress. So here's ultimately my game plan.

Game Plan:

1) Test all existing 10 landers I have at the moment again.
2) Test 5 landers at a time and take the offer that worked best for me with those landers.
3) While testing those landers cut placements that spend 3x the offer payment regardless of the amount of conversions. Focus on the lander and offer as you stated
4) After testing those 5 landers, see which lander came out on top.
5) Take that top lander and test the remaining 5 landers. In this phase, I think it's good to reopen up the placements that I've deactivated to see if there's any increase in conversions.
6) After testing all those landers, you come out with 1 lander that is the best.

I'll leave it off here. But, I think this is something I'm seriously considering by restarting my campaign because I didn't take a systematic approach to it.

---

Before moving onto anything else, we need to clarify something:

When split-testing offers or landers, DO NOT use rules of thumb (e.g. x amount of payout) for cutting. Split-test 2+ candidates and ONLY cut when the "probability of being best" falls below 10% when compared against the current-best candidate!

Please read my previous post above again where I took your lander stats and plugged them into the online split-test calculator to find the winner. Also see this post for a step-by-step. Knowing how to cut stuff correctly is extremely important. If you use rules of thumb, you'll either cut something before your data reaches statistical significance, or spend too much money running stuff too long after stat sig has already been reached. Just set up your split-testing, keep checking for stat sig, and keep cutting candidates whose "probability of being best" < 10% against the best, until you're down to your last one. Then start another split-test.

(In some case you will not want to eliminate until you're only left with the last one. For example if you're doing your first round of lander testing using landers from the wild and they're all different copies and themes etc., you would probably want to only cut the worst ones, take the promising ones and create variations. But if you're testing the same lander with 10 different headlines with everything else being the same, then you would most likely want to cut until only one is left. Always know what you're trying to achieve with each split-test you set up.)

You've addressed one of the most major obstacles encountered by many newbies - and I know and remember it well because I was a newbie not too long ago - and that is not having a campaign optimization strategy or system. Once you have one you will "optimize your optimization system" as you gain experience. Psychologically it makes things a LOT easier knowing every step you take, is taking you one step closer to making your campaign better. The trick is to sit on your hands and wait out each split-test and not change a thing before stat sig is reached. Oh yeah - you bet I've been guilty of doing that in the past!


Several thoughts regarding your game plan:

-Why not include offer testing in your game plan? A round of offer-testing using your current-best lander, followed by a round of lander-testing using your current-best offer, and so on - would be a good way to increase your ROI fast. And if you want to speed things up at the price of more ad spend (pun-intended sorry ), you could even start a separate campaign, use one to test offers and the other to test landers. If you don't have a lot of traffic in the first place it would be best to just stick with one campaign though.

-If you're going to have 2 rounds of lander-testing back-to-back, consider testing all 10 at once as it would cost less - because in the second round you'd be re-testing your winner lander.

The point here is you want to get as much traffic as you can for the testing, but you also want to keep costs low. The "cut placements that spend 3x the offer payment" may be sensible, but doing so "regardless of the amount of conversions" is not. If you have a ton of traffic, consider cutting placements based on gut-feeling or rules of thumb - but ONLY the ones that are in loss! You would not want to cut placements that are making conversions AND in profit or close to breakeven. If you don't have a lot of traffic to start with, I'd suggest to only cut placements that are in major loss. Also, keep in mind that if you cut based on gut-feeling/rules of thumb, that it would be a good idea to re-test those placements once you have a good offer + lander combo - at least re-test the bigger placements that don't look like they're total losers.

In short, my suggestion would be to continually split-test offers and landers until you have a combination that's doing a solid 30%+ ROI. After that start monitoring ROI of placements, devices, models, etc. etc. and cut anything that doesn't give you your target desired ROI (I set mine at 30%+). At the same time, scale the best offer(s) and lander(s) to other traffic sources.

Hope I haven't made things even more confusing!


Amy

---

Thanks Amy for following me along this journey! It's been one hell of a ride.

I gathered some insight around various threads and have ultimately come up with a finalized game plan that I feel pretty confident about.

Game Plan:

10 Landers
1 iPhone6s Offer(I know it's been converting better than the Zalinco one, but that's just me lol)
ZeroPark Traffic Source
$100 Testing Budget

Note: I will set the top bid to only .015. I noticed in my time in running campaigns that I'm usually in bid position 2 by doing this.

1) Test all existing 10 landers at once. Given that ZeroPark has a traffic volume around 292 million, in the initial testing phase I would like to test an amount of $100. That way I am accounting around 6.5 - 7 million in traffic(7mm*0.015(Bid Rate)/1000 = $100 in testing).

2) With the initial testing phase I will only allow each placement to have 135,000 impressions(135*0.015=135k impressions). Although, a lot of placements won't reach criteria of surpassing 135k impressions, the reason why I'm setting a "impression limit" is to have a lot more placements and not just a bunch of placements in the initial phase that are spending more than $3-4 or the rule of thumb. So, ultimately allowing only $2 max on each placement in the beginning. Since I'm testing 10 landers at the same time, I want to also be able to allow breathing room for all the landing pages to get its shot at each placement .

Note: At this point I'm not cutting placements to completely disregard them, I want my landers to be exposed to more traffic on different placements.

3) After spending $100, take another $100 to test but this time testing the landers that came up on top. At this stage, if there really isn't a clear winner lander, I will take the top two landers and create 2 variations of it. So, in the end I'll be testing 4 landers in stage 2 of the testing phase. Whilst opening up all the placements again and allow each placement to spend more. Ex. 270k impressions(135k*2). My thinking behind this is because I know which landers are dominant in the first testing stage, so I'm more confident that it'll convert in stage 2 of testing.

4) After Stage 2, I will be able to see which lander is dominant and if I still have to test which one is better, I will have to narrow it down by creating more variations. So when I find one prevalent lander I will start to split the campaign in half. So as Amy mentioned, one campaign with current best lander against 2-4 offers and another campaign with 1 best offer against a handful of landers(not sure how many landers, but that will come later).

So, this is my ultimate testing plan as of now. I will keep you all posted on the results. I plan to launch the campaign tonight. It will require me to do a lot of micromanaging of the placements. Stay tuned!

---

Back to back testing these past few days! I have followed my game plan pretty closely and didn't stray too far from the path. I'm going to mention some key distinctions that I've noticed running the campaigns this time.

Guidelines/Variables/Budget:

- Spent $320-$330 for testing
- Didn't let a single placement spend more than $2-3 each testing phase. I accidentally spent $9 more on this one placement....went to get coffee and then came back to see the placement taking all my money lol.
- Did the AB testing on landers and I have finally narrowed down to one lander.

Let's get onto the campaign stats:



Thoughts:

- Every time that I went into a new Phase of testing I had one goal in mind. I remember looking over some threads and someone mentioned the importance of a goal out of every testing. Yes...it's good to see profits at the end of each testing. But, in this case I was more involved in looking what kinds of landers I could cut, the CTR on those landers, etc.... This provided more clarity in what I would need to do next. I believe starting off in marketing and in my opinion that gets newbies like myself intimidated is not having a clear guidance on what to really do. But, I think having a systematic approach to campaigns is one of the ways that could provide you more confidence in your campaigns.

- Using the A/B calculator that Amy provided I was able to narrow it down LP9 being the most prevalent lander. The strangest thing about LP9 is that it was previously the top lander as well. But, in the last testing I just didn't test my campaigns systematically. So, I will now use the variations I created from LP9 and test them with my current offer while I will create another separate campaign using only LP9 to test against other similar offers.

- Since I have got a significant amount of data or at least I think I did. I have about 42 pages worth of placements in my campaign now. As I said before, I limited $2-3 per placement each testing phase. But, I will reopen up those new placements again when testing those variations and so forth. I think the only time I will really cull the placements is when I optimized it to the point where I'm making profit and those placements are not making me a dime at all is when I will draw the line to cut those placements.

- After optimization and getting my campaign hopefully to greener days I will start messing around with the bids on individual placements and looking at the patterns of dayparting.

Random Tangent: I'm trying to spy on my profitable placements, however for some reason I see these obscure links that don't lead to any website. Also, I see some websites labeled as adcash. Is ZeroPark brokering traffic from AdCash?

---

Wow Simon! You're been busy - lots of progress here - and thanks for posting so many details!

Comments:

-There's still some misunderstanding regarding how to cut landers. Basically we don't care how much money has been spent - we do not need to know that in order to check for statistical significance!

Basically what you do is, start split-testing the 10 landers, then once in a while plug all the numbers into the split-test calculator to identify the "current-best" lander (as in TEMPORARILY-FOR-NOW best, not statistically significant best!), then use the calculator again to compare that current-best to EACH of the other landers, and cut the ones that have less than 10% "approximate probability of being best". You'll do this check once in a while and cut, cut, cut until you're left with landers that you feel have potential. (Or, if none of the landers are even close to breaking even, you may even want to just take the "current-best" and test entirely new landers against it.)

Then, you can take the leftover landers (the ones that have potential) and create variations of them. Then start another round of testing using the leftover landers plus the variations.

So now - using your data as an example - in phase 1 LP7 is clearly the "current-best" (again, as in TEMPORARILY FOR NOW, we won't know whether it's statistically significant until we use the calculator!) So we compare LP7 against every other lander using the calculator, and we compare one by one...

So first we compare LP7 with LP1. The "approximate probability of being best" are 87% and 13% respectively. So LP1 isn't ready to be cut yet!
Next we compare LP7 with LP2. The "approximate probability of being best" are 99% vs. 1%, so LP2 can be cut.
Then we compare LP7 with LP3...etc. until we go through all landers.

Result: All others can be cut except for LP1, LP3, LP7 & LP9.

So at this point you have several options. You could keep running traffic to these 4 landers and do further checks on the calculator to eliminate more landers, and THEN create variations of the remaining one(s). Or you could just decide that all 4 landers have enough potential to warrant making variations of them to start another round of testing right away. Which path you choose will depend on so many things - such as how different your landers are, how different their conversion rates appear to be, etc. Just do what sounds right and feels right to you.

One more thing: Since you've continued to test LP7 and LP9, I added up all the impressions on the landers and the conversions and plugged the results into the calculator and got 37% vs. 63% "probability of being best". I would definitely suggest at this point that you not continue testing just these 2 landers against each other because they're obviously performing similarly and there'd be no point in waiting forever until one emerges a winner. I'd suggest testing variations of at least these 2 landers next. And if you've found other interesting-looking landers while spying, feel free to throw them in there too.

You are definitely on the right track by testing LP9 variations. As mentioned, you can consider testing some variations for LP7 as well - especially if LP7 is vastly different from LP9 in some aspects.


-Limiting the spend on each placement like you're doing WILL expose landers to a wider range of traffic quality from more placements. However, this is overkill IMHO because it will require a lot of monitoring which may not be worth the trouble. You would indeed still want to pause placements that are draining your budget, in order to save money. But as long as your landers are being rotated evenly, we can assume each lander was treated "fairly", i.e.that each lander received approximately the same # of impressions from each placement.


-Yes Zeropark brokers traffic from many places, including Adcash. So many of the urls you see are probably ad server urls and not actual site urls.


Excitedly waiting to see what the results will be after the next round of offer + lander testing! Making good progress - again, thanks for taking the time to post all these details Simon!


Amy

---

The one issue with my lack of successful campaigns is offer testing.

I'm using a proven lander based on spy tools. For example, I'd use a lander that runs the longest and has quite a bit occurrence on the spy tool. The one lander I use primarily with all my campaigns currently is the Questionnaire(Where was google founded?) lander.

I haven't drilled down to OS yet, I thought that it's best not to blacklist anything at first when testing landing pages and offers. But, then again seeing your post makes me think differently now.

Is my testing process off? I noticed that not only in my thread but in other threads you recommend affiliates to use 2-3 AM recommended offers along with 5-10 ripped landers to test offers. But, wouldn't it be quicker and more cost efficient if I just took at the Australia iphone 6s offers and link it to a proven lander to test the offers?

In my testing strategy I feel as if I'm not getting enough data to confirm whether or not I have a strong offer. I could be testing 13 offers like I am in Australia, but there's only 2-4 offers that gets conversions and it's not too a point where it seems statistically significant.

Let's look at the campaign example below:

Preface: I haven't been running this Australia campaign for long. As mentioned early in my posts, I took all relevant(13) iphone 6s offers from my affiliate network and put those offers up against a proven lander.

Body: The results weren't that good. For once a GEO that worked alright for me was starting to take a shift in performance. But, that's expected since I'm testing all the offers they have. After plugging all the numbers into my spreadsheet and looking at all my Voluum there are some things I've noticed that I didn't think I paid attention before.

1) I'm bidding at a pretty high CPM even though the offers only pay around $1-$3. And because of this perhaps my bid should be way lower than what it should be.
2) Mobile Carrier converts better than Mobile Wifi.

End:

Within my Australia Mobile Wifi campaign, I've went to check "pockets of green". Narrowing it down to "OS" doesn't show anything green. If I take the red "OS" placements and narrowed it down to "browsername" I see green placements. But then, again they're not statistically significant for me to proceed with the Wifi campaign. I think it's okay to bury this campaign or start again using with Mobile Wifi using Amy's suggestion(2-3 recommended AM offers vs. 5-10 ripped landers).

Within my Australia Mobile Carrier campaign, I find two offers that work the best among the bunch. Both offers have 5 conversions each.
If I narrowed down to OS-Browser, no real significant pocket of profits. I think I'll take those two offers that show the best performance and test them again to see if there's any significant growth in ROI. After that I'll start getting rid of bad placements/targeting.

---

The definition of a "proven" lander, is when you've split-tested a good number of landers and a particular lander has won. Just because you see a lander the most often in a spy tool doesn't make it the best lander. I may have used this term loosely in the past which may have caused some confusion.

Because you're testing so many offers, having a well-converting lander is essential - otherwise 1)you can't really accurately gauge the profit potential of the offers, and 2)you'll be overspending when weeding out offers.



There's no definite way of doing things, and every situation is different. However, seeing as you're wanting to mass-test offers in multiple geos, it would be more of a priority to get green faster to generate some positive cashflow - unless you have a strong heart and a big test budget.

You can test each geo thoroughly, or test a bunch of geos simultaneously but less thoroughly. It all comes down to choice of approach. If there was a one-size-fits-all optimization approach, someone would already have written a guide on it.

For example if you're seeing that the vast majority of leads were from Android OS, then by only targeting Android, not only will you get green faster, but you'll spend less money when testing all those offers.


Yes you're right of course! But the key here is having a proven lander.

The 2-3 AM recommended offers + 5-10 ripped landers, is for people that have no experience with the particular geo+vertical so don't know what landers work. In that case, casting a bigger net is almost crucial - because if you're just testing 1 offer with 1 lander, not only would the chances of success be small, but when you DON'T see good numbers you're left asking the question: Should I keep testing or ditch the camp?

On the other hand, If you're casting a wider net of 2-3 offers + 5-10 landers, and none of the offer+lander combos shows promise for any major traffic segments (e.g. a particular OS), then you can say "OK maybe there's lower-hanging fruit elsewhere". And if you see a combo showing promise, you could continue to run until you have a winning lander - and the chances of the winning lander actually BEING A GOOD LANDER will be higher. (Analogy: There're 10 balls in a bag and only one has a star on it, when you draw 8 balls you have better chances of picking the one with the star, vs. drawing just 2 balls.)

And having a good lander will then allow you to test more offers. Because you have higher confidence in your lander, if the next few offers you test aren't at least close to breaking even, then once again, you know there're probably easier geos.

So you may ask, how high does the initial ROI have to be for me to continue optimizing? It would be near impossible to provide a cut-off because every situation is different and everyone's test approach and goals are different. This is where experience comes in. Some considerations are:

-How many more things can I optimize to increase my ROI? Say you see an offer converting a lot better on IOS than Android, and/or you're running on a source that has lots of bad placements you inevitably need to weed out in the beginning, and/or you see one bad placement eating up a big part of your budget - all of a sudden the low ROI doesn't seem so bad anymore.

-How worthy of your effort is this camp? If it's a big geo with lots of traffic, you may decide to test more landers and offers to make that geo work, cause you know that when you do you'll be rewarded in a big way. So even if the initial ROI doesn't look too good, you may want to put in the work required to get green. Or, if your friend has told you a particular geo is doing well, or you've read several follow-alongs where members got green for the same geo+vertical, then you may want to put in more effort.

-How much traffic does the geo have? Example: Say you're running in a geo that has a ton of traffic. Even if you only identify one major traffic segment, e.g. Android + mainstream + carrierX + only the best placements + only the best hours of the day, you could still make good money. In this case you won't care too much if the initial ROI was lower, because after all these optimizations you can raise ROI by a lot. Whereas a smaller geo does not have as much leeway - if you cut too much stuff you'll be left with small traffic volumes.

-How creative can you get with your landers? Say you're just breaking into a new vertical, and have only been testing ripped landers. You have several great lander ideas you want to test out. In that case you won't care too much if the ROI as on the lower side - as long as you're getting conversions you can test out your new landers. However if you're having trouble coming up with lander ideas and choose to stick to ripped landers or variations thereof, then you probably won't be able to increase the ROI significantly through lander testing.

Of course, it's easier if you test several geos at a time - that way you can just pick the one(s) with the highest initial ROI to optimize.


A lot of people have the same reasoning, which may be flawed depending on the traffic source.

On many traffic sources where people can bid higher for traffic from good placements, the higher you bid, the higher conversion rate you'll get.

It's NOT like when you're bidding higher, you're sacrificing more of your margin in exchange for traffic volume. In many cases you'll get an increase in CR too. Whether that increase is worth the additional cost is what you'll need to test to find out.

This is why you need to test several bids to find out which one gives you the best ROI. Before mass-testing offers, you may already have tested a couple of offers (the ones you were using to split-test ripped landers to find the best one). Simply use the better offer and best lander, test a few bids to find the best bid, then mass-test offers at that bid.



Suggestions for this AU Sweeps camp:

-If you haven't already, test more landers first to find a decent one. Can use the best-converting offer when doing this.
-Drill down to carriers in your "carrier" camp to see if there's any specific carrier doing good ROI that you can target solely.
-Drill down to OSs to see if either Android or IOS converts especially better than the other.

At an ROI of -70% and over, you'll need to either find a significantly better lander or cut out some major traffic segments that aren't converting. Cutting placements to reach green probably won't cut it (please excuse the pun). You've tested 13 very similar offers so chances are you won't find another offer that will all of a sudden give you 200% ROI over your current best (anything is possible, but you know what I mean). I just want to make sure the camp still has potential before you spend more money, so please provide some stats on carriers and OSs, and also let me know how many landers you've tested so far.


BTW what are some of the geos you're considering testing?



Amy

---

I was thinking about it the other way around. In your Granular Testing for Mobile Post, I decided to create a campaign for Wifi and Carrier with a bid +.15% of the avg bid. If I take the approach 2-3 best offers against 5-10 ripped landers, once I reached down to the best lander among each offer do I just take those landers and start mass testing offers like the scenario below?

*Scenario:
Landers - 1,2,3,4,5,6,7,8,9,10
Offers - 1,2,3

Initial Test: *Really scrutinize the placements and targets(OS/BROWSERS/DEVICENAMES) during the initial test. That means I'll have to keep a close eye on these campaigns to make sure any placements or targets are going overboard with my testing budget.*

Recommended AM Offer 1: Test Landers 1-10. Landers 2 and 3 performs the best
Recommended AM Offer 2: Test Landers 1-10. Lander 8 performs the best
Recommended AM Offer 3: Test Landers 1-10. Lander 10 performs the best

Secondary Test: Gather all the different iphone6 sweeps offers from Australia in my affiliate network and start testing them against the winning landers, so it'll look like this?

Previous Recommended AM Offers 1,2,3 left out of test
New Offer 4: Test with landers 2, 3, 8, and 10 based on previous performance.
New Offer 5: Test with landers 2, 3, 8, and 10 based on previous performance.
New Offer 6: Test with landers 2, 3, 8, and 10 based on previous performance.
New Offer 7: Test with landers 2, 3, 8, and 10 based on previous performance.
New Offer 8: Test with landers 2, 3, 8, and 10 based on previous performance.
etc....

With my test, there's anything that sticks out like a sore thumb.

I'm going to work with France, Germany, and Australia at this point.

Every time you post I get more information about how to approach affiliate marketing and this may sound like something I should already know. But, it's not reinforced in my head enough .

Taking your approach I have a lot of work to do this weekend, and probably I'll see a difference in the results. I'll definitely keep you posted.

---

A sample scenario - what a great idea!

Hang on a sec - are these 3 offers very similar? The 3 offers should all be similar enough to be able to use the same set of 10 landers. So say these are all iphone6s offers, then you would be using 10 iphone6s landers. And you'd mostly likely cut these landers based on "overall" stats across all offers, and not for individual offers.

So say that after you've been running these 3 offers with the 10 landers for a while, you got:

LP | Imps | Leads
LP1 5000 0
LP2 5000 0
LP3 5000 4
LP4 5000 0
LP5 5000 1
LP6 5000 0
LP7 5000 2
LP8 5000 0
LP9 5000 0
LP10 5000 1

You'd probably NOT care, for example, which offers were responsible for the 4 leads made by LP3. Because the 3 offers are all iphone 6s, you kind of assume the landers would perform similarly for all 3 offers.

So in this example you'd keep running LPs 3 and 7 and cut the rest. Once one of them wins you can use that to mass-test all iphone 6s offers for that geo.

Note: This example assumes offer payouts to be similar. If you have offers with really different payouts (e.g. yesterday I was split-testing 2 offers - one with $8 and the other with $12 payout), it may or may not be a good idea to cut landers based on "overall" stats across all offers, because each lead will have a different value. However, chances are the same lander that converts the best for offer 1, will be the best converting lander for offers 2 and 3 as well - as long as the offers are similar enough!

Note 2: If upon more testing LPs 3 and 7 look like they're performing around the same, just go ahead and either just pick one or use them both to mass-test offers. We don't care which lander we use as long as it is / they are good enough to gauge offer promise AND keep costs down in the mass-testing.



Again - the idea here is to use the best lander to mass-test offers. i.e. You won't want to start mass-testing before you find the best lander of the bunch you were testing.

Exception is if landers 2, 3, 8 and 10 were all performance similarly, going head-to-head with one another in the initial testing of 2-3 offers above.



I've never liked sweeps very much, but I've been running them lately for a case study (that will be posted soon) and now also as preparation for the 3rd round of the AMC course.

Coincidentally, I'm planning on testing the geos you've listed above, next (I'm talking about tomorrow haha). I've been testing lots of sweeps offers for the Asian geos up to this point - for TH, VN, MY, and SG. Oh and the UK.

So - while I can't tell you whether FR/DE/AU are easy or not, I would suggest MY and SG and TH. For TH, stay away from the low payout offers that pay $0.10-0.15 - it's hard to make those profitable and you'd have more luck testing the ones in the $1-1.50 payout range. I didn't manage to make UK work but then I didn't test a lot of offers for that particular geo.



That's a real nice thing to say Simon, but I can't help feeling I may be going into too much nitty-gritty here. There are so many different possible scenarios and different styles and approaches when it comes to testing offers and optimizing a campaign, and I'm far from being an expert.

Basically, as long as you believe that your optimizing decision/approach makes sense, just keep doing it until you feel otherwise. The more camps you launch, the more you'll revise your strategy. The last thing I want would be to limit you with my current paradigm, cause I know you can come up with better strategies on your own given you're such an analytical thinker.

Just try not to fall into the trap of analyzing all day every day instead of running camps. Make decisions as are necessary and don't try to come up with a perfect optimization plan just yet. We'll get there.



Amy

---

Would you setup your campaign in Voluum like this:

Path 1 - Lander 1 - A, B, and C Offer
Path 2 - Lander 2 - A, B, and C Offer
Path 3 - Lander 3 - A, B, and C Offer
etc.... all the way to lander 10.

Is there a specific reason why you wouldn't add LP10 & LP5 to the test even though it has one conversion? The payouts could vary depending on the GEO I'm working with. For example, Australia I'm working with offers that payout around $1.80 - $3.20. Not much of a spread between the offers. But, for France I noticed that I'll be testing offers w/$20 payout(These are probably subscription offers) and offers that are SOI based $2-3. So it's probably a good idea for me to do the granular testing on mobile for France to see which carriers can tailor to the specific offer.

These iphone offers that I'm testing are similar in one way or another. Some of them probably require a longer flow for conversion like France as mentioned above. I ran a iphone 6s indonesian campaign that paid out around .04. But, the amount of leads I'm getting was substantial which allowed me to make a better decision on what to cut and what not to cut. Plus given the amount of leads it was probably an offer with a very easy flow. Now, this is strange..... I just came up with a thought. I stopped running this campaign because it wasn't making much. I spent a total of $20. $10 to wifi and $10 to carrier. I made a total $8.04 which has the potential of becoming a easy $50 a day campaign if I optimize it? Now, there's only one iphone 6s indonesia offer in my affiliate network and I haven't tested multiple landers yet. I think I'll revisit this campaign soon once I have more offers from other networks and creating more LPs to test.

---

Gosh no haha! That would be a nightmare to set up. What would be the benefit of doing it that way?

Why not just set up a single path with the 3 offers and 10 landers?



That was why I included this note:

"Note: This example assumes offer payouts to be similar. If you have offers with really different payouts (e.g. yesterday I was split-testing 2 offers - one with $8 and the other with $12 payout), it may or may not be a good idea to cut landers based on "overall" stats across all offers, because each lead will have a different value. However, chances are the same lander that converts the best for offer 1, will be the best converting lander for offers 2 and 3 as well - as long as the offers are similar enough!"

When testing offers that have very different payouts WHILE testing landers at the same time, this is what I do: I would treat each offer+lander combination as a single "split-test candidate" and compare them in this calculator:

https://win-vector.shinyapps.io/CampaignPlanner_v3/

For instructions on how to use it please see:

http://stmforum.com/forum/showthread...+offer+payouts

Then I would cut offer+lander candidates just like I would for a group of offers, or a group of landers.

Regarding carrier traffic vs. wifi traffic: You could set up separate camps to test offers for carrier and wifi separately. OR, you could just test offers for wifi first, get profitable, and then test offers for carrier traffic to further increase your overall ROI. OR, you could just test offers for carrier traffic. It's really up to your personal preference.

Running wifi and carrier traffic in separate camps will allow you to bid higher for carrier traffic, but that's assuming you have an offer that converts great on carrier traffic (or a specific carrier even). But if you want to run volume nothing beats wifi traffic. So if you're running in a geo that doesn't have enough carrier traffic, you may decide to just focus on testing offers for wifi, and then only test offers for carrier to further increase ROI. As well, when you do this, you can run on sources that don't have carrier targeting - in which case you'd just set up rules in your tracker to direct carrier and wifi traffic to their respective best offers.



Yup - that camp looks like it has promise! Next time someone tells me "this offer's payout is too low - I can't possibly make it work" I'll be sure to refer them to your post and screenshot!

And if you find an offer+lander combo (possibly an offer+lander+OS+? combo haha) that can do high enough ROI for WIFI traffic, be sure to set up a camp on mediahub - they have a ton of ID traffic. Sad thing is they don't have carrier targeting - at least that was true the last time I ran there.

Just thought of something else too: If you ask for a 1 cent pay raise on that $0.04 offer, you'll already be in the green for your carrier camp.




Amy

---

For some reason, I think I saw in one of Finch premium posts why he split it up like that was to get even distribution among the landers being tested. However, I can't reference back to his premium posts.

At this point, since it's only one offer that's available on the network. I'll scour other networks to see if they have additional networks. If I do find a few offers to test, I'll start comparing them with the multitude of landers.

Thanks for pointing out MediaHub, I'll be sure to check it out.

BRB, asking AM for a payout bump .

---

Hmm....if you rotate 3 offers with 10 landers in a single path, all offers and landers will be rotated evenly unless you specify otherwise.



Haha glad to hear!



Amy

---

I'm going to split this posts in two parts. The first part is a general overview of my campaigns. The second post is looking at drilldown data and you'll see through my thought process of cutting items.

Over the weekend and weekday I was busy translating landers and tweaking some coding on the landers I ripped. I'm going through Phase 1 of testing which is just testing offers against ripped landers. I'm hoping to take it slow this time and not rush on making irrational decisions. After all that, I gathered this for my testing:

Australia:

7 Ripped Landers | 2 Offers($2.80 & $3.15)

France:

6 Ripped Landers | 2 Offers($.80 & $22.40)

Germany:

7 Ripped Landers | 2 Offers($2.80 & $1.60)

Campaign Criteria:

- Created campaigns including Mr. Green's blacklist(PopAds). So that means in the excluded websites tab on PopAds I restricted 100+ siteids to be promoted in these geos.
- Divided campaigns between Wifi & Cellular
- Bid was set slightly significant(oxymoron :P) above average. So if the avg bid was .002240 I would set it around .003240.

Phase 1 Testing Results:



Quick Thoughts:

1) Off the bat, when it comes to testing I don't like when I'm testing a GEO, one campaign exhausted the funds while the other is still gathering data. In this case, between Wifi and Cellular.
2) The reason why the budget is higher for France is because I'm testing two offers that have a wide variance in payout.

Now let's examine some campaigns!

---

Australia - Cellular:

Offers:


Can't cut anything here. Both offers have one conversion, although one payout is higher, I don't want to cut the lower offer because it might be the winner here. In this case, I'll still keep the offers running.

Landers:


This is the hard part for me. Do I just cut everything here and leave only LP 6 & 7 running? My thoughts are that I should just think it won't be enough conversions to cut any landers yet.

OSNAME:


I think it's safe to cut IOS / 9 / 3 because it's spending way too much of my budget. But, as always I feel that I don't ever want get rid of anything that generates me a majority of my traffic. Should I get rid of IOS /9/3?

BROWSERNAME:


Right here, I could see that I should cut Safari/9 and Google Chrome/50. Again the thought comes up should I cut something that generates the majority of the traffic for me?

DEVICENAME:


Same ideology as browsername and osname.

WEBSITEIDS:


Placement 835224 is on it's way to banville.


Takeaways:

1) I think the next step is to cut the landers with 0 conversions. The reason is maybe getting rid of the bad landers that I could finally see an increase of conversions in the OS/BROWSERS that I'm targeting.

---

I've been working with one trusted affiliate network and asking AM for 2-3 recommended offers as well as ripping 5-10 landers to test with those geos. So far I've tested with Australia, Indonesia, Germany, Singapore, and France. I will test Italy this weekend. I'm probably going overkill with the initial stages. But, I've been making more accurate decisions now than I have in the past. I started contemplating less.

As a rule of thumb for me, If a GEO doesn't have high traffic I don't really bother with blacklisting as much problems to get to profitable as opposed to a GEO that has high traffic I'd take more consideration in the things I'll be blacklisting.

Here's a recap on my campaigns:

1) I have got rid of running Australia and Germany campaigns. Today, I tested Singapore and it's close to becoming a GEO I'm blacklisting for now. The reason why I blacklist these countries because I generated less than 5 conversions individually for Wifi and Cellular campaigns. I split my campaigns between Wifi and Cellular. Also, I want to focus all my emotions/energy that generates conversions at the moment when I'm trying to get winning campaigns.
2) Working on France, Indonesia as the main campaigns right now.

BLACKLISTED GEOS:



WHITELISTED GEOS:

FRANCE:


INDONESIA:


In a separate posts I'll go a little more in depth in my France and Indonesia campaigns.

Questions:


1)Does anyone have a explanation of the below. Are the promoting terms below only stating you can use Wifi or Cellular with only these Internet Service Providers.

---

Its a pin submit. doest matter what the ISP (carrier or wifi) but takes only pins sent to those two carriers only for double opt in.

You have to find out if its MO or MT flow. MO converts better but MT pays more.

---

iPhone6 Sweeps Indonesia campaign - PopAds - Background:

- Split Test between Wifi and Cellular
- Did 2 trials of testing already.
- 1st Trial was taking one offer from my network and using it with one "proven"/not really proven lander. Mistake: Just because the lander appears a lot on the spy tool doesn't necessarily mean it's a proven lander. Corrected mistake by ripping 6-7 landers to test. Since there's only one offer that's available in indonesia. I went to two other affiliate networks to find the exact same offer to test. So, that starts the beginning of the 2nd trial.
- 2nd Trial - 3 exact offers from 3 different affiliate networks vs. 5 ripped landers. Results are the below from 2nd trial:



Let's take a look at the Cellular campaign:





Although offer 1 has 66 conversions, offer 2 has a higher payout rate. So, I blacklisted offer 1 & 3 and running only offer 2. Drilldown to the landers that offer 2 has, I am now using landers 1, 2, and 5 and blacklisted the rest. Now, time to wait for the results......

Onto the Wifi campaign:





Same thing again.... Offer 2 is the winner, and offer 2 is the same offer and network from the cell campaign. Drilldown to the landers, and I choose landers 2 and 4 to keep running. Now, time to wait for the results....

Notes:

1) I'm taking my time with this campaign. There's a lot of volume in this geo. If this geo didn't have a lot of traffic, I would start looking for pockets of profits. But, I'm willing to sacrifice some money in order for a higher volume and profit campaign.

---

Regardless of whether a network is "trusted" or not, you should be trying to test more than one. An offer (or similar offer if they don't have the same) can perform differently across networks for a variety of reasons, not just because one is a shadier than the other.

Singtel/M1 will probably be one/two click flow, and wifi traffic will be MO or MT. Usually the payout is the same, so focus on the one click carriers first.
Wifi traffic will still convert but at a lower CR due to the harder flow - if you can find really cheap wifi traffic it could be worthwhile, but in my experience it's better to hammer the main supported carriers.

---

iPhone6 Sweeps France campaign - PopAds - Background:

- Split Test between Wifi and Cellular
- Did 2 trials of testing already.
- 1st Trial - Tested 2 offers($20 offer against a ~$1-2 offer) vs. 6 ripped landers. The lower payout offer did better which is kinda expected. My next step was to find the similar offer from my preferred other 2 networks.
- 2nd Trial - 3 exact offers from 3 different affiliate networks vs. 6 ripped landers.

Results are the below from 2nd trial:



Let's take a look at the Cellular campaign:





Offer 3 is winning offer with 5 conversions. Drilling down from offers to landers, not much significance. Next test is take the winning lander and test against the landers to see which lander sticks out like a sore thumb.......

Let's take a look at the Wifi campaign:





Offer 1 is winning offer with 3 conversions. Drilling down from offers to landers, not much significance. Next test is take the winning lander and test against the landers to see which lander sticks out like a sore thumb.......

---

What's MO/MT? One/Two click flow is easier because there's less steps for the user to complete the requirements, right?

When it comes to finding out one click carriers, that comes to a lot of testing I'm guessing?

---

Yes - one/two click flow is far easier. Everything is done on the offer page.

MO flow is when the user enters their cell number and gets sent a text, which they then have to reply to in order to confirm the subscription.
For MT flow the user enters their cell number, gets sent a text with a pin in it, and has to enter it back into the offer page.

Both of them take the user more effort than just clicking a button on the webpage... although perhaps MO/MT offers back out better for the advertiser because there is more commitment involved (that's just me speculating tho)

There's higher fraud and cancel rates on one click offers, which I believe is why not all carriers support it (and as an offer owner it'll be harder to get approved for/bigger bond, etc).
Basically it's much easier to force an impulse decision - I'm sure you can imagine why one click antivirus converts like crazy...

---

Asking my AM for good converting MO/MT iphone sweeps offers. Those pay more

Thanks for this! It's no wonder why a lot of AMs are recommending AV offers in Asian geos.

---

They usually pay more yes, but again the flow is harder. Don't stick to them exclusively just because the payout is higher.
You also might have trouble finding sweep pin subs... they're generally form submits.

---

Campaign Recap:

I've been running sweeps campaigns in several geos trying to find out which geo shows somewhat of a potential. So far these geos are out of the picture:

Australia
Germany
France and Singapore(Still contemplating whether or not to pass these geos aside for now).

Right now I'm running in Indonesia because it generates me the most conversions and with those conversions I'll able to make more decisive decisions than a offer that has 2-3 leads.

Here's the progress, and I'll explain each phase:

Indonesia:


Phase 1: Took exact offers from 2 different affiliate networks and tested against 6 ripped landers.

Phase 2: Found a winning offer from one of the 3 networks I tested. And running the most profitable landers against the winning offer.

Phase 3: Established winning lander and offer combo.

Phase 4: Created separate campaigns to staggered bids.

Phase 5: Found the winning bid that I'll continue to keep running. A quick note about this is that the high bid and mid bid performed about the same, but the mid bid gave me more impressions which would be better for me in the long run.

Phase 6: Testing the best bid with the best offer/lander. However, after drilling down to specific stats, there's nothing I see that could really cut without decreasing a significant amount of traffic. I'm currently stuck right here.



I'd like to get rid of any targeting that's 2x the payout. But for instance Android/4/4 generates a majority of traffic for my campaign.



Same for browser, but I can eliminate Google Chrome / 47 & Samsung Browser / 4



IPNAMES not really showing anything potentially good as well.

If you're approaching this campaign what would you get rid of and why?

---

I was taking a quick nap, and I don't know if some of the STM members were speaking to me in my dream lol. But, something told me that the geos I'm disregarding that I should test other offers with other affiliate networks. Not just sticking to one affiliate network to test offers, but let's say one geo doesn't work out with AM recommended offers, try another affiliate network.

---

Hey there Simon! First of all I apologize for the delay in replying - have been busy with other projects and not been as active here on the forums as I'd like to be. Let's take a look at your stats! (To avoid confusion I'll reply to one post at a time...)

Meet my new favorite tool for assessing and cutting offers:

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

You're right that you can't cut offers at this point.


As always, use the split-test calculator:

http://www.peakconversion.com/2012/0...al-calculator/

And once again, you can't just cut the other landers yet. I really wouldn't recommend that you do.


You can take either of 2 approaches:

1)If you're not getting a ton of traffic, and don't want to slow down your testing, try not to cut anything unless you have to. Con: This approach will require more test budget. Pro: You'll be able to cut things faster.

2)If you wouldn't mind having to wait for a long time to cut stuff, feel free to just target the most profitable traffic segments and use them to carry out your lander and offer testing. Con: You may have to wait forever to cut stuff. Pro: You'll save on test budget.

Remember though that an offer won't be around forever, so if you hit on a good one, test and scale quickly. But, if you're just wanting to test landers, then speed of execution wouldn't be as important.


I wouldn't focus on placements at all at this point, because you don't have big placements eating up your budget. I mean you could still cut placements, but that's not going to increase your ROI by a lot compared to testing offers and landers.


I wouldn't do that - not when your "best landers" only got 1 conversion each.





Amy

---

Completely agree with what you're saying about having a system, and being able to make better decisions that involve less thinking.

However - why did you stop running AU? The landers that got conversions were at -30-40% ROI, and both conversions were for IOS! In the very least, I would suggest to target JUST IOS and keep running all landers for a while to see if you have a winner. On top of that, you haven't played with bids, or cut placements yet! So yeah if you want to give this camp another $10-20 say - that's what I would do.

You may also want to take another look at your DE camps in a similar way, i.e. drill down to OS -> Landers to see if there are combos close enough to green, so that you can push the camp into green by playing with bids and cutting placements etc.


I could give you an SG offer to play with, unless you've completely given up on this geo. Will PM you.

Casting a wide net to identify hanging fruit - yup that's my approach too.


Best thing to do would be to ask your AM. A lot of offer requirements are ambiguous and convoluted.



Amy

---

Nice approach! Yup Offer 2's the winner (I checked the calculator). Just a note about cutting landers: If the offers are similar enough (which yours are - you have the same offer basically), you can just cut landers based on total stats across all offers. (Unless you see a particular lander working particularly well for one offer - which doesn't happen all that often, assuming your offers are similar enough.)



Haha it's no surprise that offer 2 is the consistent winner! It has a similar conversion rate as the other 2 but has a payout almost twice as high.


Good idea! Conquer this one geo and scale to multiple pop sources and you'll be looking at mid-to-high 3 figures a day at least. Next step would be to join more aff networks, and gather up all ID sweeps offers to test. Once you're making profits from multiple traffic sources, it may be worth it to test more landers/variations. (TIP: When you're ready to scale, start with mediahub...)



Amy

---

You're got the process down. Just going to repeat what I said about cutting landers: Given the similarity between offers, you can just base your cutting decisions on overall lander stats (across all offers), as opposed to lander stats for the winning offer.



Amy

---

For France - what happened with Offer 1 + Lander 5? Again - did you drill down to OS?



As pointed out in an earlier post from today: Test more offers! You've said yourself that there's a lot of traffic in this geo, so it would be worth your trouble to improve your funnel as much as possible before cutting stuff.

Test all ID offers you can find. Sign up to more aff networks if you need to. I wouldn't even test lander variations before finding a good offer.

It may also be worth it to set up a quick test on mediahub even now - spending $10-15 there will tell you whether the camp is ready to be scaled to there.



Amy

---

Good point!

You've already spent the money in those geos, and in some cases even started cutting landers. Testing more offers would be a good idea.

Having said that - tier 1 geos have more competition, so if you want to conserve test budget, testing lower-tier geos with low payout offers may make more sense.

Having said THAT, I would definitely suggest focusing on ID first, before reconsidering past geos or considering new geos.



Amy

---

Thx Amy for the advice! Also for the PM

I keep on feeling like I'm jumping around. Yesterday I started another strategy to create campaigns. I think I'm constantly changing the way I do things because I want a good foundation on creating campaigns.

Every campaign that I create within a certain GEO has phases now.

Yesterday, I created multiple campaigns in multiple geos still targeting the same countries as before. This time, I paired them up with varying iphone offers from different affiliate networks. The stats from Phase 1 are from below.



Phase 1:

Indonesia: This is my main campaign now, I am taking your advice Amy on approaching this. I am scouring through several affiliate networks to see if that offer is available. I also just used my previous winning lander LP2 to go against multiple offers. How do you determine when you had enough testing for offers? How do you know when to stop testing offers? I understand that we want the offer with the best payout and conversion rate, but when do you know when you have enough info to stop testing offers?

Before scaling to other traffic sources, do you recommend improving the funnel first meaning even if I found the best offer/lander combo keep trying to change up the winning lander variations to see if you can spike up the conversions or do you just scale to all the traffic sources and then once a several of the traffic sources do well then you try to make lander improvements.

Australia: This is a competitive geo to work in but the volume is good here as I know from previous experience. I know in popads right now there's a lot of competitors in the space. In the first phase, I tested 3 offers from one network because of a recommendation from a AM who has direct relationship with the offer. The result it didn't turn out too well. And just an hour ago, in phase 2 i upped my bid to be among one of the top 10 bidders and it only generated 1 conversion with $30 worth of testing. So at the end of phase 2, each offer(total 3) that I'm promoting each has 1 conversion(not enough data, plus spent $60 to get only 3 conversions). It's time to try another affiliate networks offers. The thing with Australia is many networks have multiple AU offers. So, I ask the AM to give me the top 2-3 offers AU offers and then I test them, if they don't show significance then I move on to another affiliate network.

[Germany, France, Singapore, Italy] - All of these countries haven't really provided me significant conversions yet in the first phase. I'm testing offers from various networks. But, with these countries I was targeting them mostly through the slow times and not the peak times(6PM-12AM). So, these geos are going into the second phase of testing which is targeting only at peak times. If there is any rise in conversions then I'll continue on. Otherwise, I'll archive the campaign and keep working on Indonesia. It's more important to me than ever to organize my campaigns now, so I created a Voluum naming convention for myself to keep track of these campaigns. In case in the future, I want to revisit my old campaigns.

---

After running a bunch of traffic to various geos, I'm sticking with a few geos right now.

This Indonesia iPhone6s offer is really bugging me as I'm not sure where to go from now.

Here's the progress:



Offer Payout: $0.08

I'm at a point where I'm not sure what to blacklist. Again here's my hierarchy on how to blacklist items.

1) Browsers
2) OS or OSname
3) ISP Name
4) WebsiteIDs

Let me show you some things in which I would blacklist

This campaign is PopAds - Targeting only Android in Indonesia
DATA: 6/26 - 6/28

BrowserName



OSName





I did drilldown from OsName -> BrowserName and Google Chrome 51 is one of the budget drainers. However, Google Chrome 51 is one of the majority conversion generators, so it's not good if I were to blacklist Google Chrome 51. So, it's better if I get rid of Android 6 / 0.

ISPNAME



WEBSITE IDS



I don't want to do too much micromanaging here, but I would blacklist all the placements here. But, right now as I'm writing this post, I drilled down from websiteid----browsername and see 1219839 would be profitable if I cut out Google Chrome 51. So I wouldn't cut out Google Chrome 51 because it's generating a lot of conversions for me, so it would make more sense for me to just blacklist 1219839



Let me know your thoughts on my blacklisting approach.

---

Home > Mobile > Follow-along Campaigns