Home > Mobile > Follow-along Campaigns

---

Building A Small, Profitable Pops Campaign And Scaling It (34)

---

Introduction
I spent the last two months learning the basics of mobile affiliate marketing, building my infrastructure, and getting comfortable launching campaigns. My goal now is to build a small, profitable campaign using pops traffic and learn how to scale it.

This feels like an ambitious goal. I spent the past two months working full-time on affiliate marketing without a profitable campaign. But I believe that this goal is achievable. I will be working full-time until I get there.

I am starting this follow-along to document my journey. I hope that other STMers can get some benefit from this, especially those who are just getting started with pops and/or the traffic sources that I'll be using (Popads to start).

My Current Position With AM
Here are some details on what I’ve achieved over the past two months. I include this so that other newbies - and anyone who’s interested - can know where I’m starting from.

1. Coding Landers - I have very basic HTML & CSS skills. I can modify ripped landers in a very basic way, but for anything original or intermediate, I outsource on the cheap using Elance.
2. Traffic Sources - I have a lot of Adwords search experience and some Facebook experience, but not as an affiliate. I spent the past two months exclusively learning Google Display Network as an affiliate. I have no experience with pops.
3. Spying - I have been spying on pops campaigns for about a week. I learned by searching the forums.
4. Infrastructure - I am using DNSMadeEasy & Rackspace Cloud Files
5. Other Tools - Photoshop (low-intermediate skills), Sublime Text (for coding), PureVPN, Voluum

For more information about my background with digital marketing, you can read my introductory post here.

First Pops Campaign
Offer: Antivirus pin submit
Geo: South Africa
Payout: $3.60
Traffic Source: Popads
Budget: $8/day
Max Bid: $.03
Lander: One ripped lander to start; made some modifications to make it compliant
Targeting: All categories, English, Android

I set the budget low while I get familiar with Popads.

Popads Setup
Quality: Average quality and better
Frequency Cap: 1 view/24 hours
PrimeSpot: All traffic
Adv. Type: Popunder
Allow other methods when the chosen one is not available: No
Throttling: Disabled
Connection Targeting: Connection Type = Cellular; Connection Speed = Cellular/Carrier; Internet Service Providers = MTN-SA, Vodacom, CELL-C
Website Targeting: Disabled

I am trying to target three cell providers. The above Connection Targeting settings are my best guess at how to do this, because there isn’t a cell provider targeting option.

I’ll post an update when I have some data!

OJay

---

I spent my first $8. I have to say, it’s nice to get data in a matter of minutes or hours rather than DAYS with Google. I feel so free and nimble.

So far not too bad…. it’s way too early to tell, but I’m happy that I don’t have to stare a goose egg in the face for the rest of the night.

Campaign Update
Spend: $8.60
Revenue: $3.60
Profit: -$5.00
ROI: -58%
CPV: $.023
Lander CTR: 24%
Conversions: 1

The placement that got the 1 conversion had 10 impressions. This brings me to my first question:

QUESTION: Should I do anything with this placement or is it too early? If it’s too early, what is a good rule of thumb for when to break placements out and target them?

On a similar note:

Guidelines For Cutting Placements
Currently I’m planning to cut placements that:

1. Have spent 2-3x payout, AND
2. Have no conversions

The closest site to meeting these criteria currently has 152 impressions and $3.88 spend. I have my eye on you, placement #xxx658.

My next steps are to add a new lander and start throttling up the spend.

---

Just some suggestions:
- It's too early to cut placement, let it run for 3 days. From my experience it's better to test over the weekend which can get you more traffic .
- For bigger payout, I would set bigger daily budget (maybe 30$ or more) and set hourly throttle. Bid higher if necessary.
- Don't check the tracker every 5 mins, let it run and collect data.
- BE CAREFUL for those placement that gives the most traffic with bigger click-loss. The way to check this is to compare report on your traffic source and the one on your tracker.

---

Agreed with Nuel.

I would say it's WAY TOOOOO EARLYYYY to cut any placements if this is only day one. You are going to need to let it run for a solid one or two weeks minimum IMO (of course everyone has their own process). This also depends on how much your daily spend is or the amount of conversions you are getting as well. However, if there is a placement that is taking a good chunk of your daily budget with no signs of conversions or 0 clicks, I would consider it. It will be tough to get this amount of data to determine which placements are good or not with the current budget and offer you have set. At this point, unless your 10000000% certain your lander is Gold (and not a lucky conversions), there is no telling whether or not the placement that had 152 impressions and $3.88 spend is blacklist worthy. There are many variables that play a role that also need to be taken into account: Day of the week, time of day, category (for popads), etc... These can play a big factor into why there are no conversions. Well... those and the lander of course. You could be cutting a potential winner without knowing it.

Speaking of daily spend, as nuel stated it is recommended to have a higher daily budget for the current payout you have. I would personally recommend running something with a smaller payout first like an app install. I know you hear a ton of people say the same thing, but trust me you will learn much quicker when it comes to bidding, placements, and all that AM stuff. Main reason being is that app installs are much easier to convert than a sweep and it requires a much lower budget. More conversions will give you more data and with the data you can see have a more accurate determination on which placements to work on. I started with app installs at the start myself and it personally helped me out in terms of learning a process. Later on I moved on to another vertical which I solely focus on now.

Lastly, be sure to check all of the variables in your tracker. If you make any changes I would recommend having it documented. It may be a hassle at the beginning, but if you keep at it you will be doing it without thinking about it. This is one thing that helped me out a TON. When I started off I was playing around with ton of different bidding levels, blacklists, whitelists, you name it.. I ended up hitting my first profitable campaign back then and started messing more with the variables so much that it just tanked. I had to try to back track as much as I could remember, which was pretty almost impossible at the time because I didn't know what I was doing or did. With this practice, you will know what you had done for each day and what bids, placements you removed, etc... that affected your campaign. It may not be something you want to keep doing, but I believe it will help at the start.

Hope this helps. I would also say to either start or join a mastermind. This was probably the best thing I have done and I still work with my team. You will learn so much, so fast. No joke.... Anyone feel free to correct or add to anything I said.

Anyways I wish you the best OJay and remember to keep trying and don't give up (emphasized way too much for a damn good reason.)

P.S. Just reactivated my STM account (sorry!), glad to be back!

---

First, apologies for the long post here…. I’m using headlines so you can scan for whatever you’re interested in.

For today, I did quite a bit of work… not much to report in the way of campaign successes, but I learned a lot about cutting placements. (I'll share that at the bottom of this post.)

Thanks For Your Comments!

Thanks, I hadn't thought of that... it looks like I'm getting 10-25% click loss on my top 3 placements. What number should I keep an eye out for?

Yeah, I started with app installs and ran them exclusively for two months on GDN. I learned a ton running them. It was a tough decision to switch gears, since I know from reading here that it's best to stick with one traffic source and one vertical. Ultimately I decided to make the switch, based on a number of factors that I'm happy to share if anyone's interested.

Great advice. I just started a spreadsheet to keep track of changes.

Yeah, I was thinking of doing this. I will look into it soon.

Thanks, Exitius! It's really encouraging to read your guys's posts.


Before I get to the campaign updates, there were some questions about my budget, so I thought I’d clarify that.

My Budget
I’ve budgeted so that I can lose $1,000 per month for the next year on AM and be OK. Even if I don’t make a single dollar of profit, I can lose about $35/day on campaigns for the next year.

Of course, I plan to make some profits, so I can increase my daily spend above $35/day if I need to. But I don’t want to give myself a heart attack, so I’m going to aim for $35/day spend and increase it when there is a good reason.

If anyone has any thoughts on this strategy, I’d be really interested to hear them.


Campaign Updates
Campaign-A
(This is the campaign I started yesterday.)

Total Spend: $24.39
Impressions: 1,093
Total Conversions: 1
Revenue: $3.60
Profit: -$20.79
ROI: -85%
Budget: $15/day

This is getting a little ugly. I’m still just collecting data, so I didn’t cut any placements or take any other actions. I did increase the budget to $15/day and may increase it again tomorrow so I can collect data faster.

Campaign-B
This is a new campaign recommended by my AM. I am using a similar lander to Campaign-A, but it’s a very aggressive lander (as compared to Campaign-A’s friendly little compliant lander).

Offer: Antivirus pin submit
Geo: South Africa
Payout: $4.40
Traffic Source: Popads
Budget: $15/day
Max Bid: $.03
Targeting: All categories, English, Android

All other settings are the same as Campaign-A. No data yet.


Popads Update

1. It’s interesting to see that Campaign-B is getting the same big sites for its top placements as Campaign-A got on its first day. (Website #xxx658 has the most impressions for both campaigns.)
2. The settings I used to target carrier traffic worked.

Finally, my big lessons for today….

Placement Cutting Guidelines - UPDATED
Nuell and Exitius both made comments about not starting to cut placements yet. My head was spinning trying to figure out when to start cutting placements, so I did some research here on STM, and this is what I came up with:

Step 1: Set Target ROI
It seems like setting a target ROI is the first step in deciding when to cut placements (this thread and this thread helped me learn this). Once I have a target ROI, I’ll be able to test placements for statistical signficance and determine how likely they are to hit my target ROI. (I'll explain that step next.)

In this thread, Caurmen suggests a 40% target ROI for this type of campaign. (I should say, I’m not positive that guide was written for mobile, so I wonder if 40% is right for my campaigns here.)

I’m setting a 20% target ROI, because I’m still looking for my first profitable campaign, and I don’t want to be greedy.

Step 2: Calculate # of Impressions Needed to Cut the Placement
There are two steps to this:

1) Determine the placement's minimum viable conversion rate
2) Calculate significance for the # of impressions needed

(I'll skip the calculations on this, but if anyone’s interested in how to calculate these two things, I’ll be happy to share. Caurmen talks about it here, although I couldn’t get a usable number by copying his formula to Excel.)

1) Determine the placement's minimum viable conversion rate
For website #xxx658 for Campaign-A (my closest placement to getting cut), my minimum viable conversion rate is about .84%. This means that website #xxx658 needs to convert at .84% in order to hit my 20% target ROI (at its current CPV of $.025 and my current payout of $3.60).

2) Calculate significance for the # of impressions needed
I used the split test calculator (which Caurmen explains here) to calculate significance for how many impressions this website would need before it’s worth cutting. By entering 0 for Successes (i.e. conversions) and changing the Trials until I got “95% chance conversion rate between” 0 and .84% (my minimum viable conversion rate), the number of Trials ended up being 350. This means that I need to let website #xxx658 run for 350 impressions with no conversions before I can cut it with confidence.

[On a side note, my girlfriend’s a scientist, and she has some opinions on this method. She says that you need to repeat the experiment twice and calculate standard deviation between the three results. That would be expensive here, but I can confirm from my conversion optimization experiments on the web that not calculating standard deviation can result in completely bogus data... (Ex: "the light pink button converts 47% more than the medium-light pink button with 95% confidence!")]

Anyways, I’d be really interested to hear if anyone can confirm that this method for cutting placements will work. In searching STM so far, I wasn’t able to find this method for cutting placements, but I easily could have missed it. (And I just as easily could be 100% wrong on all of this.)

On to cutting campaigns... this is WAY easier.

Campaign Cutting Guidelines
Caurmen says here: “As a rule, if a campaign is at -30% ROI or more after you've completed a couple of rounds of landing page and offer split-testing, kill it unless you're very sure that you can make it work.“


So that was a lot of work... and a lot of reading effort on your part if you read all of this. I'm hoping this method will be a guide for me for the rest of this thread and in the rest of my affiliate career. And if it works, I hope this will help other readers here too.

Do you think it will work?

---

I just wanted to post a quick update, because I found a thread that vortex wrote three weeks ago which goes into this approach in detail. It's the most definitive thread on this topic that I've found so far in STM. Here it is:

http://stmforum.com/forum/showthread...Banners-Part-2

I'm definitely going to use this approach in this follow-along.

---

Vortex’s post was a huge help. I spent much of the past four or five days trying to better understand when to cut campaigns, offers, landing pages, and placements.

I feel that this was time well spent. Now I’m going to limit my research time and spend as much time as possible on my campaigns.


Campaign Cutting Guidelines - ADDITION
I am using vortex’s spreadsheet to test whether a campaign will reach “ROI > -30%” and testing for 80% confidence.

I also found this post by vortex very helpful for making a decision on whether or not to cut a campaign. In addition to the 80%/-30% confidence/ROI factor, I will use the guidelines in that post.


Campaign Updates
Campaign-A
Total Spend: $62.02
Impressions: 2,863
CPM: $21.66
Total Conversions: 3
Revenue: $10.80
Profit: -$51.22
ROI: -83%
Status: Killed

This campaign had -83% ROI. Using vortex’s significance spreadsheet, I can see that there is 99% confidence that it will not reach -30% ROI. There are no promising placements to break out, so I killed this campaign straight up.

Campaign-B
Total Spend: $62.00
Impressions: 3,057
CPM: $20.30
Total Conversions: 4
Revenue: $18.80
Profit: -$43.27
ROI: -70%
Status:Killed

There is 90% confidence that this campaign will not reach -30% ROI. However, there is one placement that is worth breaking out, so I’ve broken it out into Campaign-C.

Campaign-C
Total Spend: $5.80
Impressions: 203
CPM: $31.50
Total Conversions: 2
Revenue: $9.40
Profit: $3.60
ROI: 62%
Status: Active

There isn’t much traffic to this site. I’m keeping my bids high to try and reach statistical significance as fast as possible.

Popads Update

• I found this very helpful post from Mr.Green on 100 sites to blacklist on Popads; I added them to my Popads excluded sites

---

OJay, could you point us to Vortex's spreadsheet? sounds like a useful tool. Thanks!

---

Nevermind, I found it. Thanks OJay/Vortex

---

Wow OJay - great follow-along!

A lot of amazing suggestions already. Here's mine:

Thanks for your compliments on that spreadsheet calculator! However, for an evergreen niche like antivirus where you can potentially bank from for a long time once you find something that works, don't fixate on ROI in the beginning!

Instead, focus 100% on testing and optimizing your lander and finding the best offer!

Meaning, don't even use that spreadsheet for now! Just cut the worst placements that are obviously budget drainers. If you cut using ROI criteria at this point, you will seriously limit your traffic which will slow testing to a crawl after a while.

For something like antivirus, there are tons of offers in tons of geos, and landers that work for one geo will usually work for many other geos as well. So by focusing on mass-testing landers, you'll be getting the most bang for your buck in the long run.

As for detailed test strategies - I've talked about mine several times already so won't repeat myself here - see this post for example:

http://stmforum.com/forum/showthread...l=1#post227509


Good luck my friend - I'll be on the lookout for your next update!


Amy

---

You cant test a 4 dollar offer with 8 dollars a day budget. Each placement should at least get 3x the offer payout to get significance. You prob need at least in the 100s budget/day.

With 8 dollar/day you could prob test a .10 app install offer

---

Things are moving slowly. I’ve hit a point where I need to get some coding done, and now I’m remembering how slow I am at that. It’s frustrating, but in the big picture I’ve learned a ton this week, which I’ll go through here.

The biggest thing that I feel I’m accomplishing right now is that I’m working out a process. I’m trying things systematically, researching why things aren’t working, identifying new methods to try, and taking lots of notes.

My hope is that when I finally get profitable, I’ll know why I’m profitable and exactly what I did, so I can replicate the process. So hopefully all this time will be worth it in the end.

First, thanks Amy!
It’s nice to see your post here. I have learned so much from reading your insights throughout the forum. Once I get profitable, I will certainly owe quite a bit of it to your generosity.

Once again, you are one step ahead of me. I found the thread you posted a few days ago and reworked my strategy accordingly. Here’s what I’m working on now.

My New Approach
Taking Amy/vortex’s suggestions in her comments here, I am casting a wide net to first identify a winning offer.

To do this, I’ve identified 3 top performing antivirus offers with the help of my AM, and I have also identified 10 landing pages to test the 3 offers with. The landing pages are mostly ripped, with a couple of original ideas thrown in.

QUESTION: If it’s not recommended to focus on ROI when testing for a good offer, how will I know when to cut an offer? It seems like I could spend forever on an offer that will never get anywhere.

Stackman recommends here to spend 5-7x payout when testing offers. I’d be interested if you agree that that’s a good guideline, Amy? In the numbers I ran, statistical significance actually was reached around 5-7x offer payout. I thought that was interesting and took it as a sign that stat sig was a good guideline.

What I’m Doing Now
Getting the landers ready (10 landers x 3 languages). My coding skills aren’t great, so it’s taking a while, and I may end up outsourcing the coding so I can focus on other things to get these tests ready.

I’ve also been working on translations. OneHourTranslation was more expensive than I expected. It would have cost about $400 to translate 10 landers into three languages. So I’m using Google Translate and hope to find people on Elance to proofread them at a much lower price.

Campaign Update
I’ve killed all of my Popads campaigns. Campaign-C wasn’t reaching significance, and the traffic was miniscule.

Traffic Source Update
Popads has been nice and easy, but it hasn’t given me a ton of traffic. I set up another ten or so traffic sources, including Zero Park. I was hoping to stay away from Zero Park since everyone uses it, and Stackman strongly recommends here to find unique traffic sources. But I read in another of Amy’s posts that Zero Park has a ton of traffic and is therefore a good source to test pops with.

QUESTION: What are some other great pops traffic sources for testing offers?

Thanks for your comment, Sleenirvana. As you can see, I’ve taken some steps back and am not cutting placements any more...

---

$400 for 10 landers translated? Must be pretty text-heavy. Check out fiverr - might be able to get it done at a much better price still with good quality.

Also, the ROI comment about not focusing on that also said... 'in the beginning'! Of course ROI ultimately is very important. Right now you just need more data for testing and then you can start worry about ROI.

---

@OJay,

I concur. Yes Zeropark is competitive, yet due to the volume Zeropark is a good place for you to test your stuff out, and fast.

There's a volume breakdown at Zeropark that you can refer to on the bids, go easy with the bids first when you get in.

Test everywhere!

---

I never learned html/php/css/js either and have always tried to just "wing it". However, recently I finally got sick of not having any basis at all and set aside a weekend to go through some online tutorials - codecademy has really easy-to-follow ones on HTML/CSS, PHP, JS and JQUERY. If you want your landers to contain interactive elements I'd highly recommend the JQUERY tutorial. To further your learning you can try w3schools - they have tutorials that aren't interactive like the ones at codecademy but they cover a LOT more detail and all the functions you'll ever see. These should be enough to allow you to at least mod landers faster!

Also check out these awesome threads by caurmen (and seek out all thread started by him if you have the time!!):

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

And of course the entire programming subsection is littered with gems as well!

If you need to do more than just minor modding - there's a coder I've used from Fiverr that's good, cheap, and fast! To avoid having him bombarded with jobs from everyone, I'll PM you his link instead.

With some luck, after you've gained some experience, you can choose to partner up with a fellow AM'er that codes well. I'm extremely lucky - someone like that joined our group and we no longer need to outsource except for the most heavy-duty coding.


Very solid thinking! Having a process is very important. Blindly throwing shit at the wall and not seeing results is disheartening. Based on observation and personal experience, not having a system is one of the major reasons why many people quit.

Everyone has a different way of doing things, but it's nice to have a system that you can modify over time. Also, it's important to set it up so that the various parameters of your testing (i.e. creatives/offer/traffic) will continuously improve, such that the overall combined effect of all the parameters will result in a gradual improvement in ROI.

Your initial plan sounds great! Looking forward to seeing how you do.


The idea is to identify the best offer of the bunch that you're testing, so we're judging by relative performance and not absolute performance (such as ROI). Simply use the peakconversion calculator to crunch your numbers (by identifying the current-best offer and comparing it to each of the other ones, and cutting if the "probability of being best" is less than 10%; if you want to be extra sure, cut when this value is below 5% - see here for details on how to use the split-test calculator).

Here's an example on some offers I assessed and cut yesterday - these are all in the same geo, rotating in one camp:





As you can see, the best offer has a "probability of being best" of 95%+ when compared to all other offers. I was using the same lander that was customized to match each offer, in such a way that it would be as fair as possible. Needless to say, I've paused all except the best offer.

Note: At this point you shouldn't really care about ROI! Even if the best offer isn't giving you a positive ROI, if you've tested most/all of the best offers in your vertical for that geo, then you have optimized the "offer" variable of your system for now! Next you'll want to use this offer to test and optimize landers etc. The point is by using the best offer you'd be minimizing your ad spend during lander testing.

(To choose a good geo to use for your testing: talk to your AMs and/or scour STM posts to find one that performs consistently well for your chosen vertical. One that has lots of offers to play with. This way you'll be sure to find an offer that converts well so you don't lose your shirt in the testing stage. Also - if you choose a geo that speaks the same language as you, you wouldn't need to get stuff translated and would be able to test out new lander ideas on the fly.)


OMG!! $400??

As Greg has pointed out above - use Fiverr if you have a lot of text to translate! Sort by ratings and pick translators that have very high ratings (4.8+ out of 5) and lots of reviews and you can't go wrong! They do anywhere from 100-1000 words per $5 gig.

More importantly - I have a question to ask: Why are you mass-testing landers for several geos? Have you already identified really high-performing offers in these geos?

Casting a wide net is great, but the trick is to expend minimal effort (and cost!) possible before finding promising offers and geos that warrant further investment. I wouldn't even start testing other geos until I've done a few rounds of testing using my initial geo, so that I have a pretty solid lander! After that, I would translate JUST that lander and use it to mass-test offers in different geos.

Right now I have around 40 offers in 15 geos being tested using 1 lander. Of those 15 geos I'm seeing promise with around 5 offers across 3-4 geos. NOW would be the right time for me to once again mass-test landers for ONLY these 3-4 geos - because they have convinced me they would be worthy of me spending further time and money to hit them out of the park!


Actually you only need 1-2 high traffic sources to do your testing. Finding traffic sources that have less competition is GREAT if ROI is what you're after, but for testing purposes you want lots of relatively cheap traffic - and zeropark is the largest self-serve pop network I know (for test campaigns where I need to pause and unpause etc. on the fly, I don't want to go through a managed network where I need to wait for a rep to do it for me, so managed networks are out). If anyone knows of a better network to do massive testing at please let us know!


Whew! Another extremely long-winded post - hope I haven't bored you and I'll be watching this space - thanks for doing such an interesting follow-along OJay!


Amy

---

Nice suggestion! Here's the volume and average bids chart letsgosingapore was talking about:

https://zeropark.com/volume/



Amy

---

It's been a productive week. I may be at a point where I can start optimizing landers to get my campaign profitable.

Here are some responses to comments. As always, I'm including headlines so you can scan for what you're interested in and skip everything else.

Translations
Thanks for the suggestions on getting cheaper translations. I ended up translating about 10 landers (726 words) using Google Translate, then paying someone on Elance $15 to proofread and edit them. Translating these landers using OneHourTranslation would have cost $69 for a basic translator and $105 for an expert translator. My Elance translator was a professional. I'm happy with the results.

Coding
Thanks, Amy, for the great coding suggestions. I'm still trying to avoid having to improve my coding, but if I do have to brush up then I'll definitely start with your suggestions. They sound great.

For my 10 landers, I ended up scrapping the ones that would have required more coding than I could do easily. I spent some more time spying and found ten landers that I could rip that didn't require more than basic coding. I just updated their links & text, copied and pasted mods (sound, vibrate, etc.), and made some other basic changes.

Offer Testing Method

This was an absolutely huge lesson for me. I really can't say enough about the gigantic hole in my understanding of affiliate marketing that this filled in. It seems obvious in hindsight, but it would have taken me forever to figure this out on my own (and a lot of money). Thanks, Amy!

Again, this was an eye-opening insight. I put two of my geos aside and focused on creating 10 landers for just one geo. I selected 3 top offers in that geo, and I started testing.

One Geo, Ten Landers, and Three Offers – Results So Far
Immediately one lander started converting, while the other nine did not. I had made a note during my spying that this one lander was likely a top performer. But I didn't have enough data to prove that this one lander was better than the rest. Regardless, I made a gut-decsion to pause the other nine and see if this one lander could help me cut offers quickly. It did.

Two offers quickly ran to two conversions each. The third offer didn't pull any conversions. I ran the numbers in peakconversion calculator, and with 97% confidence the bad offer was outperformed by each of the converting offers. So I cut it.

Now I have two offers that look the same so far. Here's where things currrently stand:

Offer: Antivirus
Payout: $9
Geo: Small (population ~ 15 million people)
Traffic Source: PopAds
Impressions: 7,405
CPM: $12.29
Conversions: 4
Total Spend: $91
Revenue: $36
Profit: -$55
ROI: -60%

Here are the numbers for the one lander running the two converting offers:

Impressions: 4,118
CPM: $12.29
Conversions: 4
Total Spend: $50.61
Revenue: $36
Profit: -$14.61
ROI: -29%

Next Steps & Questions
Here are the next steps I'm leaning towards:

1. Get new landers ready – start creating variations of the lander that's doing well, so I can mass-test landers on the converting offers.
2. Launch new traffic sources – start launching this campaign on some new traffic sources and see if I can get similar numbers to the above Popads campaign.

My thinking here is that I want to be ready to optimize this lander to profitability and then start moving it to new traffic sources. It's going to take a little time to learn those new traffic sources, so perhaps I should start now.

QUESTION #1: Does anyone have an opinion on these as next steps? Is there anything I'm overlooking?

QUESTION #2: My CPM on Popads is $12.29, which seems higher than what many STMers bid for pops. But if I bid less than this, the traffic comes in very slowly. It hardly seems worth it to save a little $$ at the expense of having to wait several days for data to come in. Does anyone have any suggestions on getting cheaper/faster pops traffic for testing offers?

ZeroPark
I tested out Zeropark, like everyone suggested. I set my starting bid at about 50% of the average bid, but I barely got any data so I raised the bid to 67% of the avg bid. Still nothing, so I raised my bid to the avg bid.... still barely any data. I kept raising it until I was at 4x the top bid ($.04 vs. top bid of $.01). And even then I still could only spend $8/day. So I paused Zeropark and went back to PopAds.

Even worse than the lack of traffic on Zeropark, the traffic I did get was 45% tablet. I'm running mobile phone offers, so 45% of my traffic was basically useless. This happened even though I had my targeting set to carrier-only.

QUESTION #3: Any insights into why I'm not getting traffic on Zeropark? I've seen tons of other STMers say the same thing, but I haven't found any STM posts on how to crack this barrier on Zeropark. If anyone knows of any good STM posts on this topic, please point me to them.

I'm excited to move into the next phase and hope that the data keeps looking good.

---

OJay! -29% ROI with minimal spend! This is TREMENDOUS progress! It's guys like you that make our posting efforts rewarding and worthwhile.

Some thoughts / replies:

-Making that gut-decision to use what looked like was the best lander to test and cut offers - beautiful move! However, now that you've identified the best offers, you may want to turn the landers back on and let them run to statistical significance.

-I would wait until the best landers are identified before testing variations. As usual, try to test factors that have the biggest effect, i.e. completely different landers, before tweaking the smaller elements. However, if you have a really good feeling about this lander, by all means go ahead with your testing of variations.

-Does this offer require lander approval from the aff network or the advertiser? If not, I'd recommend focusing on making your lander as scary as possible. Some of them will get rejected by some traffic sources, but that's just about the worst thing that could happen and that wouldn't be too bad now would it?

-With an ROI like -29%, you MAY be able to see green on other traffic sources. Another benefit of testing other traffic sources now, is that you may be able to find one that can give you more traffic than popads or zeropark. Do try mediahub and selfadvertiser amongst others.

-Do you have an account manager at zeropark you could ask on how to increase the traffic you're getting? Another way to combat the lack of traffic would be to use a bigger geo to test landers.

You're really close to seeing green! Looking forward to your next update...


Amy

---

Yeah, I think I'll do that.

It does not, so I'll aim for panic-striking scariness

Interesting, I set up an account with SelfAdvertiser, but then I saw this STM post which is very, very harsh on it, and I got turned off. Do you have any suggestions for making SelfAdvertiser work, given it's problems with bot traffic?

The person at ZP who I contacted just suggested trying domain traffic...

---

That will put you ahead of 99% of the players in this game.

---

It's hit-and-miss I guess, but one of my friends recently scaled a campaign to this network and got profitable right away. I've also set up a couple of small test camps there and results have been OK. Have yet to thoroughly test out this network properly - it's definitely on my to-do list though.

When you find something profitable give PropellerAds and Gunggo a try. Their traffic quality is solid. The only "downsides" are 1)the bigg'ish deposits (i.e. USD$1k) and 2)the fact that they're managed so there will be a delay in starting/stopping camps because you'd need to ask your account manager to do it for you.


Amy

EDIT: Wow I never did read that entire SelfAdvertiser thread from start to end before now! Some of that feedback is brutal. Makes me want to stay away now too!
Seriously though - thanks for directing me to that thread. I'll think twice before recommending it to people before I test it out further. Mediahub though we've had lots of success with, so definitely try that one!

---

Just a quick update. Things are starting to get interesting with the lander tests. On Amy's advice, I restarted all of the landers that I had paused during offer testing. I am now letting them run to statistical significance.

Amy, thanks for the tips on traffic sources. I just signed up for Mediahub by contacting them through STM.

Lander Update
I’ve cut down from 10 landers to 4 by using vortex’s method for testing for statistical significance. Here are the current leaders:



It’s interesting to note that “Lander A” is the one that I used to cut offers, because it had the most initial conversions. Now it’s in danger of getting cut.

It's also interesting that “Lander B” is tied for the lead right now, even though it has the lowest CTR of the final four. It ranks 7th out of all 10 landers for CTR.

I’m still waiting on more data, but I hope that these double-digit green numbers stay nicely in view.

---

Nice progress!

Is lander B a new lander you added to the split-testing after you used lander A to find the best offer? Or was lander A performing better before and suddenly the trend reversed?

When deciding which lander to cut, CTR isn't particularly important - I'd focus on CR instead because that's what will directly affect ROI.

Having said that - a lander's CTR can give you insight into how you can potentially improve your lander. e.g. if CTR is high and CR is low, it usually means you're either making promises to the visitor that the offer page doesn't uphold, or the message conveyed on the lander is not consistent with that conveyed on the offer page. Also, if you can figure out a way to increase CTR without decreasing the "conversions per offer lander view", that would be ideal because it would mean an increase in the overall CR.

And of course seeing green is always a nice balm to the battered affiliate marketer's soul....


Amy

---

More progress today...

Campaign Recap
I'll recap the steps that I've taken up to this point on the current campaign, to clarify where things stand (and answer Amy's question).

1. Launched 10 landers with 3 offers in 1 geo; initial goal was to test for the best performing offer
2. 1 lander got 2 conversions right away, so I paused the other 9 landers hoping that this 1 lander would help identify the best-converting offer quickly
3. 2 offers ended up performing equally; the 3rd offer never converted, and I cut it once significance was reached using peakconversion.com (the two I didn't cut are still performing equally -- 7 conversions each)
4. I restarted the other 9 landers that I had paused, since they had never reached statistical significance; new goal to identify the best performing lander out of the 10
5. Using Amy's post and peakconversion.com, I was able to cut the bottom 6 performing landers, leaving 4 landers
6. TODAY: Another lander fell to a 10% chance of being best when compared head-to-head with the top lander, so I cut it and am now down to the top 3 landers

The lander that I cut today was the lander that had the 2 quick conversions that I used to test offers. So it was definitely a good decision to restart the other 9 landers and let them run to significance!

Lander Optimization - Next Steps
I'm down to 3 landers now. In the peakconversion calculator, the top one has about a 67% chance of being best when I compare it to each of the other two separately. From looking at how these 3 have been performing day-to-day, I wouldn't be surprised if it takes a really long time for the top one to win based on statistical significance. It may never even reach significance. All three of them may perform roughly equally.

I searched the forum for some ideas of what to do in this situation, and I found this quote from Amy in this post:

QUESTION: I was thinking of creating 1-3 variations of each of the remaining 3 landers to try and improve on their CR's. Does this seem like the next best step for my campaign?

New Landers
I'm also thinking about creating 3-5 brand new landers using what I've learned works from the last round of lander tests.

The Final 3 -- By The Numbers
Current Leader -- 3 conversions & 1550 impressions; ROI wavering between +20% and +50%
Other Two -- 2 conversions each & roughly 1,600 impressions each; ROI wavering between -30% and +50%

New Traffic Sources
I have launched on some new traffic sources and am waiting for data to come in. My goal is to start collecting as much data as possible so I can figure out if one of the final 3 landers will emerge the victor.

Help Wanted!
Here's my best guess of what to do next. Any thoughts?

1. Drive more traffic using new sources to collect more data on the final 3 landers
2. Add variations of the final 3 landers to try and improve CR
3. Add some brand new landers based on what I've learned works

---

OJay I don't think you need any more help - all your ideas are great! I would suggest going ahead with everything you mentioned.

The reason why you wouldn't want to cut down to the last lander, is because when you're running really different landers, you're not comparing apples to apples because they may be different in more ways than one. With further optimizations of the best ones, there's no telling which one will be the big winner!

Whereas, if you were testing a single element on the same lander - say 10 different headlines with everything else staying the same - then you would probably want to keep cutting until you're down to the lander with the best headline.

Having said that - you're not obligated to wait until your lander split-testing is over, before you can scale! Every traffic source is different, so if you already have a lander that's doing consistently positive ROI, I'd suggest trying it on at LEAST 2-3 other traffic sources to see how it will do (preferrably networks you have experience with and know the traffic quality is OK), and go from there. Lander testing should never stop - it will keep you ahead of the pack so to speak, and will also allow you to give yourself your own "pay bumps" by adding more ROI to already profitable camps across all sources. So there's no need to wait until lander testing is "complete" before moving onto scaling. The lander + offer will probably not work on every single traffic source, so you run the winners and pause the losers. And when you have a better lander you can rotate that into all your camps, plus unpause the previous "losers" that were close to breakeven to see if they'd do better.

Another way to speed up your lander testing: Since you already have a lander doing positive ROI consistently, you could pause the other two for now and just optimize this one. Then in the future when you have new angles or new copy or whatever else, you can "upgrade" the other 2 landers and test them again if you like.

And making your own landers is always a good idea! Ripping and modding is a great approach and is the fastest and easiest way to optimize in the beginning, but innovative ideas will unfailingly find you after a while (this happens with everything we put focus on for long enough). New lander themes, completely new angles, etc. Really eager to see what kinds of damage your new landers can do!


This is so much fun!


Amy

---

Thanks, Amy. It's good to know that my direction is vortex-approved

The recap that I just posted is pretty evolved from where I started. I have learned SO much from this experience. Soon I will start replicating this process to launch new campaigns.

I also hope that others who are just starting out in pops can see this thread and follow the process that I've followed to get to this point. I'll keep updating it as I make progress.

I'm going to keep testing landers and traffic sources for now.

There was one point in your message that I wasn't clear on... it's where you said:

Did you mean that it's best to keep the final 3 landers as they are for now and not test any variations? And, if that's the case, it sounds like you're suggesting only adding new landers that are completely new ideas? That way, one of the ideas will eventually reveal itself to be the best one, and I can start optimizing variations then...?

Also, I really liked your idea of just using the one profitable lander for now. I'll definitely consider that.

---

That comment was really just food for thought more than a suggestion. As long as you're always clear on what your strategy is and what your goal is for every test, I'd say keep doing what you're doing. You can test completely different landers first, and then pick one or more landers to tweak elements and see which ones will boost CR. Or you can just randomly pick a lander and split-test elements first to see which ones will increase CR, and then apply the same elements to a variety of different landers. Chicken first or egg first - it really doesn't matter that much - because lander testing will continue probably even after you've scaled the campaign. Meaning, your current round of testing probably won't be the last time you'll be testing different-style landers or split-testing lander elements. So the order matters little.

So if you do pick one lander to split-test elements with, once you know which elements work you can always add those to the other promising landers and re-test them if you like.


Amy

---

Over the past week I did the following:

Cut 1 UNprofitable Carrier, Then Watched My ROI Go DOWN
I discovered that one carrier was keeping my campaign from being profitable. Here are the numbers:



I tested these carriers head-to-head using peakconversion, and the one with the negative ROI lost easily. So I cut it.

After I cut the unprofitable carrier, the profitable carriers started performing worse and worse. They went from +50% ROI to break-even and then to -36% ROI, which is where they are now.


Crunched Placement Numbers
1. The "profitable" carrier gets most of its profits from one placement.
2. The unprofitable carrier may be profitable on one placement.

I need more data before I take action on these findings.


Added New English-Speaking Geo
I took my best performing lander and launched it in ZA. The ZA data has been similar to my other geo. There are a couple of placements that look like winners. I need more data.

Whether or not I get ZA profitable now, the big benefit is that I can start testing landers on the fly in ZA. I won't have to translate them.


Launched Unsuccessful New Traffic Sources
I launched on MediaHub, which didn't result in any conversions after 3,000+ impressions. MediaHub also delivered 45% tablet traffic, even though I set the targeting to mobile phone-only. I am trying to get in touch with them about this.

I also set up a campaign on AdCash, but they rejected it. I am not sure why, since there are tons of aggressive antivirus campaigns on AdCash.

After this experience, I stopped launching on new traffic sources and decided to focus on improving my landers.


Landers Landers Landers Landers Landers Landers Landers Landers Landers
I've spent about 80% of my time on this campaign working on landers – spying, ripping, modifying, and generating ideas. I have about 20 new ideas for landers that I'm working on right now.

My goal right now is to get a profitable lander with +20-50% ROI.

---

Hi @Vortex & OJay,

I like your thread guys. lots of wisdom.

I am the Director of SelfAdvertiser and wanted to say one thing:
Online advertising game is all about results. Usually bad results (which happen to everyone) raises faster at first to the surface.

Vortex, as you said, you saw potential, and your friend did to. That should be your focus.
Some comments are directed to be bad, and can derive from many reasons which not all are true (competitors, banned advertisers etc.)

Cheers

---

Like to check.

How can I set set hourly throttle for popads campaign? There's manual and automatic, and how to set it properly?

Any help? Thanks.

---

Very nice progress OJay! Your follow along is really intriguing to me as well as the advice that Amy provided! Don't mean to hijack your thread, but I have a question about initial testing phases:

How do you know what to test first? Landing Pages vs. Offers?

Let's say you have 10 landing pages and 1 offer. Is that a good initial test vs another test like 10 landing pages and 5 offers?

---

Simon_89, I'm still learning, but here's my approach.

I'm following Amy's advice from this thread, so I test landers and offers at the same time at the beginning. I'll test 5-10 landers with 2-3 high-performing offers that are recommended by my AM or are top performing offers in my affiliate networks. Once I see which offer is peforming best, I pause the others and continue testing the landers until one wins.

In your example, if you're testing 10 landers, why wouldn't you test 2-3 high-performing offers at the same time? You're spending money gathering data anyways, so you may as well use that money to test offers at the same time. You'll get data supporting two separate tests for the same spend.

If you have any more questions about this, feel free to PM me. If anyone else has thoughts, feel free to share.

---

I like OJay's answer.

Basically you're looking at 2 variables that can be improved in order to increase ROI. Ideally, you'd want to optimize the variable that will take the least amount of time and money and yield the biggest ROI increase first. However, this is usually not possible - i.e. you're not certain how good your offers OR landers are. In this case it would be good to test both at the same time (2 separate camps would be best). Or you can arbitrarily pick one to optimize one first, then the other etc. (do a round of lander testing, cut bad ones, do a round of offers testing, cut bad ones, do another round of lander testing etc.)

However, if you have a proven offer (e.g. that someone you trust has told you is converting very well), then it would be better to skip the offer testing and focus 100% on improving landers. Or, if you have a proven lander that you've found success with for other offer(s) in the same vertical, you can just use that to mass-test offers.

In summary, perform actions that have the biggest impact on ROI (i.e. improve on your "weakest link") FIRST. But as long as you're always improving on something, I'd say you're on the right track!


Amy

---

Ojay,

How did you ever solve what happened to your ROI when you cut the unprofitable carriers?

Have you been able to nail down the mistake ?

---

Home > Mobile > Follow-along Campaigns