The diary of two hustlers - Mobile

[Vitamin A]

Yo, wassup folks. Fresh into affiliate marketing, we decided to launch a Follow Allong, so that we can keep track of our progress and failures and help each other on the way of success!

We are 2 guys from Germany, while one of us is located in Thailand and the other one is taking care of the company back home.

What we did so far:

• Created a company to get cheaper traffic in european trafficsources
• Created company website, so we have a better chance to get accepted at networks
• Setup a VPS + CDN at Amazon aws
• Got Voluum pro membership to have perfect tracking right from the start
• Applied to Yeahmobi, ClickDealer, F5, AppFlood, Avazu
• Started our 1st campaign

So yesterday we went online with 2 CPI from all the different networks to splittest. As a trafficsource we choosed to do popunder. We launched 5 campaigns in 5 different geos and results seem to be good for the start, but unfortunately we have problems with optimizing the camapigns, since both landers seem to have the same CTR and CR, haha.

Our gameplan was to optimize like that:

1. Splittest offers and cut the worst performing
2. Splittest landers and choose the winner
3. Optimize OS, Carriers and handsets
4. Blacklist poor perfoming site IDs

Rule of thumb for optimization:

• 1000 imperssions, no conversion - Cut it (if spent 3x payout)
• 1000 impressions, ROI a bit below -50% - Cut it and make a note of the placement
• 1000 impressions, ROI better than -50% - Let it run to gather more significant data
• 5000 impressions, ROI still negative - Cut it, but make a note that its converting at least

As I mentioned before, we encountered a problem with the 2 Landingpages and offers, since they seemed to convert the same, So we changed the CTA for both landers, which really changes the whole impression of the lander. Just uploaded them, so waiting for traffic to come in. About the offers, there was 1 with a lower payout converting way better than the other with the higher payout, also It showed a bit more in revenue, so we removed the higher payout offer. As I am writing this, I realize the ROI of the lower on dropped, so we start splittesting both again.

Campaigns after 1 day

Landers of the most promising campaign

Offers to splittest of this promising campaign

What we did to optimize today:

• create 2 landers with different CTA
• stopped 2 countries because of low trafffic/too high bidding
• raised bid for 3 other countries

We also wanted to optimize Handsets, but unfortunately this option is not possible at the trafficsource, so I guess we have to deal with it. It has like 7k impressions, but only 3 conversions, so the ROI is -90%. Any suggestions?

We also would be really happy, if you could give us any feedback on our gameplan. There is so much out there we have to learn, but we are really excited!

 

[Vitamin A]

Voluum - How did you handle it in the beginning?

Another topic we want to talk about is Voluum. We signed up for the pro membership, which includes 1 million events, but after 1 day we already realized, that this is by far not enough.

So now we are at 150k events, and by the end of June this might be already 10 million+

We are aware, that this is the cause of running popunder traffic, but I want to ask how did you guys handle this in the beginning?

Lets assume we run breakeven for 10 million pops next month, but have to pay $1000 for voluum. It seems to be hard to grind that money back with running CPI offers.

Maybe we should use Prosper202 or something else that is cheaper for the beginning? We just want to make sure, that we dont blow our whole budget just for using Voluum.

What kind of ROI should we aim for to make a decent profit after all expenses?

 

[Rian Hafiz]

Cool man, im watching the thread.
Currently im using adsbridge for tracking you can check that out
Btw do you mind share your traffic source?

 

[captain47]

Hey guys,

Some green on Day 1. Any AM's wet dream. Nice work.

And I see that you are using my guideline as the thumb for optimization. I feel famous. haha!

I have a question, regarding this: "Created a company to get cheaper traffic in european trafficsources". How does creating a company help? Your traffic seems to be quite cheap. I know asking for the name would be too much, but if you do not want to share it online, please PM me? Its cool if you do not want to though.

Regarding your game plan, it sounds solid. Based on the screenshots though, I think you should test more landers. 4 is not a good number in the initial test when you are buying 100,000 impressions. You could have tested at least 50 or so landers. Start with 10 for now.

Regarding Landers again, as a comment to this "create 2 landers with different CTA". The hierarchy for testing Landers goes something like this:

Lander Design > Heading - Images > Sales Text - Colors - CTA

Which means, first you should split test Lander design with the angles you are working on, then on the winning design, split test Headings, then Images, then the rest of the elements.

Regarding "Voluum - How did you handle it in the beginning?". The Answer: Direct Linking or use a self hosted tracker. IF you are running Pop Traffic, once you optimize your campaign and know its profitable, you HAVE to direct link anyway to save the load time. Fast load time is very important for Pop traffic esp. Pop ups.

Good luck man you two!

 

[InterGMediaGreg]

iMobiTrax is a self hosted tracker many people use(d) before Voluum. you need a decent server but the monthly $ is way less than Voluum.

 

[InterGMediaGreg]

Captain47 - when you portray yourself as an ad media company and have a nicely designed website and Linkedin profile, Ad networks tend to take you more seriously and sometimes offer you pricing that regular affiliates might not get.

 

[Vitamin A]

Yesterday we tested some changes in the LP, increased the bid for placements to get more traffic. Unfortuntaley the ROIs dropped on all campaigns. Once because we increased bids and also the changes on the landingpages werent really effective, but now we know at least what doesnt work.

The only thing we cant figure it, is the bidding strategy. Even if we have a profitable campaign and want to scale it, we have to bid higher to get more volume, but then our ROI will drop. So which ROI should we have first, to survive with a high bidding? Does higher bidding mean better converting quality traffic too? Or is it just about volume?

Another mistake we made, was testing the landingpages for too long. We kinda just put in more trafic, altough we have enough data. At how many impressions can we make sure, that the LPs are definitely worse than the others. What is the statistical significant number of impressions?

@captain47 Thanks for your reply. With the cheaper traffic I mean european trafficsources like zeropark, exoclick, adcash etc. who will charge you 20% more, if you dont provide a VAT ID. By the way, we started out with 2 completely different lander designs. We werent aware of where to start optimizing landingpages. So the next step would be to keep the both top perfoming landers and change the headings + images, pick a winner and then rinse and repeat with the salaes text? OR should we even make competely new lander designs? THey have both different design, but the angle is pretty much the same. What would you suggest to do?

@Rian Hafiz Unfortunately I dont want to share the trafficsource that openly, but I can say that its a pretty common popundertrafficsource.

@InterGMediaGreg Thanks for recommending iMobiTrax, we will have a look at it.

Mistakes and takeaways from yesterday

• Dont bid way too high in Bigger geos with high competition, you will get crushed. Better let the data roll in slowly and try learn and optimize in a less expensive way
• Dont let too many impressions run in, if you have already statistically relevant data. The rest is just a waste of many, if you are not close to breakeven
• Optimize Landers in this order: Design, angles, heading/image, Adcopy, CTA. Always change 1 thing, choose a winner, rinse and repeat with the next section.

Next steps for today

Now we will try to optimize the campaigns a bit with cutting the worst performing landers and offers. I saw a lot of potential when drilling down reports, that certain landers convert better with other offer, so what we will do now is just send traffic to the best performing lander and direct them to the best perfoming offer, hoping to increase the ROI a bit. Also we will try to tweak around the landingpages to make an even better one.

 

[captain47]

Nice going guys.

If it was me, I would:

- Stick to the first and the second landers (The ones with -21 and -30 ROI. Assuming they are the same).
- Brainstorm NEW Angles. Find 5 images per angle and test them out.

This could get massively green very soon. I feel it.

 

[Vitamin A]

Well, today was kinda a setback and we couldnt really get the things done we wanted to do. First of all, we realized our costtracking was quite wrong in voluum, cause we actually spent a little bit less for the campaigns, than showing. Unfortunately we have no idea how track costs properly, since voluum doesnt work with CPM or auto, so we adjusted the CPC ourself, but obviously it is really sketchy. How do you guys track your costs when running pops via CPM?

Second of all, we realized that there is something wrong with our tracking. We pass all the parameters for the trafficsource, but unfortunately 70% of siteIDs are showing just as the parameter. So we cant really build a blacklist. We had a friend cooming over, who checked all the stuff and it seems to be all right, so we sent a mail to the support and still waiting for a response. So unfortunately 260k of the 350k impressions are useless data. Really want to know what issue is, and if its on our backend or theirs. Whatever it is, I am pretty sure its not possible to get that data back somehow, so we will just move on, fix it and give it a shot again.

 

[ChestOfCash]

@Vitamin A very inspiring Follow along. Your CPM is really low and converting well, that's must be a really good traffic source or unsaturated geo, 3th tier perhaps? with the traffic you get I would keep RON running and keep testing aggressively, gathering profitable siteids.

@captain47 how much landers you suggest testing with CPS offers? I do manual spying and just keep watching the same angles all over the place, how can I innovate?

Regarding "Voluum - How did you handle it in the beginning?". The Answer: Direct Linking or use a self hosted tracker. IF you are running Pop Traffic, once you optimize your campaign and know its profitable, you HAVE to direct link anyway to save the load time. Fast load time is very important for Pop traffic esp. Pop ups.

On high volume pop traffic, running CPI offers, you don't use landers but just direct linking? I would be glad if you can clarify what you meant.

Thanks!

 

[koalahood]

@Vitamin A very inspiring Follow along. Your CPM is really low and converting well, that's must be a really good traffic source or unsaturated geo, 3th tier perhaps? with the traffic you get I would keep RON running and keep testing aggressively, gathering profitable siteids.

@captain47 how much landers you suggest testing with CPS offers? I do manual spying and just keep watching the same angles all over the place, how can I innovate?



On high volume pop traffic, running CPI offers, you don't use landers but just direct linking? I would be glad if you can clarify what you meant.

Thanks!

By Direct linking, he meant to not use a tracker, just link to your landers from your traffic source, and in your landing pages, use your offer link instead of the clickthrough link from your tracker.

P.S. Make sure you have at least a VPS as server or something equivalent

 

[ChestOfCash]

@koalahood it makes sense, but if you have redirect rules you have to apply php scripts with DBs to perform it, I prefer voluum's infrastructure upon VPN+CDN which sends requests to your VPS and it would just make it slower.

If you don't need any smart redirects, maybe it is the way to go. @captain47 is that what you meant? thanks!

 

[captain47]

Well, today was kinda a setback and we couldnt really get the things done we wanted to do. First of all, we realized our costtracking was quite wrong in voluum, cause we actually spent a little bit less for the campaigns, than showing. Unfortunately we have no idea how track costs properly, since voluum doesnt work with CPM or auto, so we adjusted the CPC ourself, but obviously it is really sketchy. How do you guys track your costs when running pops via CPM?

Second of all, we realized that there is something wrong with our tracking. We pass all the parameters for the trafficsource, but unfortunately 70% of siteIDs are showing just as the parameter. So we cant really build a blacklist. We had a friend cooming over, who checked all the stuff and it seems to be all right, so we sent a mail to the support and still waiting for a response. So unfortunately 260k of the 350k impressions are useless data. Really want to know what issue is, and if its on our backend or theirs. Whatever it is, I am pretty sure its not possible to get that data back somehow, so we will just move on, fix it and give it a shot again.

Cost tracking works best when the traffic source sends back the "cost" parameter to your tracker. Some traffic sources do that. Apparently yours doesnt. Even when they send back the cost, there will always be discrepancies. Your best bet is to use the data on Voluum only as a reference and use an Excel sheet for detailed analysis.

Voluum recently introduced manual cost updating, but they pulled it back.

 

[captain47]

@captain47 how much landers you suggest testing with CPS offers? I do manual spying and just keep watching the same angles all over the place, how can I innovate?

On high volume pop traffic, running CPI offers, you don't use landers but just direct linking? I would be glad if you can clarify what you meant.

Thanks!

As many landers as you can afford. By afford, what I mean is that, you need to have enough money to send at least 1000 visitors to a lander for display traffic and 2500-5000 for Pops. So if you have the testing budget to buy 20,000 visits on Pop traffic, then you can easily test 5-6 landers.

Coming up with angles is your own creativity, this might help: Back to Basics: Mastering Angles - CharlesNgo.com

And for this: On high volume pop traffic, running CPI offers, you don't use landers but just direct linking? I would be glad if you can clarify what you meant.

Answer is what @koalahood said. Direct link once you find a lander that works and you are done collecting data. In the initial phase of a campaign, you NEED a tracker like Voluum.

Hope that helps.

 

[ChestOfCash]

@captain47 thanks for answering! I understand that if you sales funnel is full optimized and performing well, you just let it run directly and if you see decrease on ROI you always can get it back to be tracked.

Did you see in the past campaigns that their performance increased dramatically duo to making this move? I guess you already checked it out and have data. Thanks!

Another question if I may, on the initial run you start blacklisting underperforming siteid while testing landers? That's how I am operating but maybe you have a better approach to do this. Thanks!

 

[captain47]

@captain47 thanks for answering! I understand that if you sales funnel is full optimized and performing well, you just let it run directly and if you see decrease on ROI you always can get it back to be tracked.

Did you see in the past campaigns that their performance increased dramatically duo to making this move? I guess you already checked it out and have data. Thanks!

Another question if I may, on the initial run you start blacklisting underperforming siteid while testing landers? That's how I am operating but maybe you have a better approach to do this. Thanks!

Yes. you start blacklisting but after you have some significant data. 100 visits and no conversions isn't significant. The rule of thumb @Vitamin A has up there is pretty cool:

• 1000 imperssions, no conversion - Cut it (if spent 3x payout)
• 1000 impressions, ROI a bit below -50% - Cut it and make a note of the placement
• 1000 impressions, ROI better than -50% - Let it run to gather more significant data
• 5000 impressions, ROI still negative - Cut it, but make a note that its converting at least

Btw, we are hijacking his thread. Lets give this one to him

 

[Vitamin A]

Small update before heading off to bed. Kinda spend like around 12 hours with affiliate marketing, but didnt really get that much forward campaignwise. Since yesterday we had the problem with the SiteID tracking, which we couldnt figure out till a couple hours ago. Apparently we did some really big mistake, so our data of 350k impressions is kinda really mixed up and not really useful anymore. Dont even ask me to explain it, I dont know how we could fuck it up like that, haha.

What we did today:

• Searched for other trafficsources
• Sent an application to adsimilis
• started to read Finchsells premiumposts 2015, highly recommend to check out his blog.
• Made a campaignjournal for every campaign we launched, wrote everything down from A to Z. Unfortunately I cant share this here so openly free, since it contains all infos about the campaigns.
• Registred at a new trafficsource, setup the tracking and started an initial testrun. Conversions are coming in really slowly, traffic seems more expensive, but lets wait for the whole data of $20 to roll in till evening or so

Takeaways from today:

• Dont rush into things, do better research before starting a campaign. This includes spying, searching trafficsources and preparing of the landers. Dont put your lander on, as soon as u received your translation and copy paste it. Take your time, think about how you could make the LP better, check all scripts and tracking setups.
• Plan your campaigns the day before and set goals/ tasks u want to get done. I often catch myself being "busy" with affiliatemarketing stuff, but actually I am just dumpling around, because I feel so overwhelmed of the problems I face and dont know what to do. Write things down, ask yourself why its important to do it and think about a strategy how to remove that from your list.

Gameplan for tomorrow:

• Analyze data of the new campaign on new Trafficsource B
• Cut the losing LP, cut placements spending offerpayout 3x and make 1 new Variation of winning lander. Simple A/B splittesting approach. This is all we will do tomorrow, since we are both busy with our real jobs, but we try to put in as much time as possible. Stay tuned my friends.

 

[Vitamin A]

Okay, so we got setup on the new trafficsource and tracking is working fine now. The cost is a little bit scetchy, since we run CPM again, but at least it gives us a rough idea of the ROI if we drill down reports. As you can see, both landingpages are almost exactly as good, so we have double the work to splittest, but also this is not a bad thing, because they are completely different, so landingpage blindness shouldnt be a big issue anymore.

We are restructuring our serverstuff at the moment, the campaigns are on hold so far. We made up some new variations and gonna put them online as soon as we are done with the server technical stuff. We will splittest different images and headlines, and increase the daily budget for the campaign to $50, cause the $20 lasted only for like 8 hours.

Another thing, that is on my mind is the offercap. Its set for 1000/daily. I guess this is usually network wide, which is pretty bad to scale. Other networks have the same offer without cap, but the payout is less lower. Plan is to try to omtimize the campaign so good, that even the lower payout ones will have a solid roi. We need to gather more data to cut placements anyways.

 

[Vitamin A]

Quick update about the small changes before going to bed. Somekind more and more conversions were rolling in, altought the campaigns were stopped for like at least 12 hours. Even visits and clisk increase and we dont know why, anybody an idea?

Anyways, as more conversions came in, we can say with a 75% chance, that Lander A is better than B, so we just will put lander B back and try all the optimization with the better performing one. We topped up the campaign wit $50, hoping to get enough data to finally cut bad placements and see if we could optimize the LP. In case the lander dies out, we have already Lander B which is also positive.

We also applied for other offers, sent creatives for approval. Hopefully everything will go fine and we can launch our next campaign soon!

Takeaways from today:

• Spent a long time reading today about splittesting, landers in gerneral etc. Many ideas popped up in my head and we will for sure watch out for different offers and not only stick to CPIs
• Make the most out of the data and visits you buy. Write down evey step you do, so always can look back, find mistakes, ways to optimize and improve.
• Redirect all unwanted traffic to an international offer. Trafficsources still send visitors from other geos, OS, Models you dont target. Dont waste that traffic, yo paid moeny for it! Just redirect it to an international offer, it might not convert super sick, but hey, even if you make $5 out of it, its a freeroll - The other traffic wouldnt even be able to convert. This might be a really important and underrrated thing if you go high volume.

Gameplan for tomorrow:

• Choose winning image, exchange Headlines and put them in rotation
• Cut site IDs with 3x cost offer payout
• Blacklist cut siteIDs

 

[Vitamin A]

Fuark, that shit is exhausting man. Woke up really excited how the optimized landers were doing and checked stats and I was really surprised. Thy were all breakeven to slightly negative. The surprising thing tho, all the new landers were better than the controll lander having 80% roi the day before, was at like -25%. We are really confused by this and cant find an explanation. Maybe because we took out the other lander with a completely differnt design? That may have resulted that people dont really pay attention to the same ad anymore? Anyways, we also have a huge conversion lag and things got a bit into green but if costs would be correct it would be a bit in the red. (15 conversions not showing in voluum tho, anybody know why and how to fix it?)

The problem we have, is that the originial LP was performing really poor today, but if we pick all the data together, it has stell an okayish ROI we can work with. Really no clue what to do now, but I guess we just go with the new one, just because it performed better the day before.

This are the stats for the whole campaign on the new trafficsource. Currently everything is paused, till we analyzed everything into detail and put the new translated headlines on the server and into the tracking. Cost are a little bit wrong tho, we spent around $79, so back to even steven.

This game is really not that easy as we thought, but really fucking hooked to it and we will make it happen no matter what. We were also talking with the AMs, got some nice offers to test. Mobile subscription and CPI, altough the CPI is for US traffic, which seems tempting at first, but I dont really know how to feel about. Our AM told us to give it a shot, the offer is sick etc, but everone else told us to stay away from US traffic at the beginning. I guess we will just get fucking smashed by the competition. Will think about it, but more tended to leave it first.

What we did today:

• Translated 6 more different headlines
• Created 1 new Landingpage for the new offer
• Spyied a bit on competition to get an Idea what offer they run + ideas for landers
• Made 2 more landingpages for the new offer
• Waited till enough data rolled in for the old campaign and started to block placements. We blocked in total 6 placements with the rule of thumb was posted above. Thanks @captain47

Insights of the day:

• Shit takes longer than we thought
• Dont block placements to quick. Rather wait a couple days and collect solid data till you can make a statistical significant test and you can say to 80%, that a placement wont be positive anymore. Use JavaStat -- Binomial and Poisson Confidence Intervals to process the data. Just type in your Impressions in Denominator (N) and your conversions in NUmerator (x). This will tell you the minimum CV and the maximum CV. If you use it for Impressions and Clicks it will tell you your CTR. Dont forget to set the confidence level in the bottom of the page. You can find a videototurial on When to Kill and Ad - walter.reade's library .And yes, the music sucks.

Plan for tomorrow:

• Pick the best performing lander
• Put the translated headlines on the server and setup tracking
• Make 2 variations of the Landers for the new offer so we have 6 to start

Tomorrow we will be also busy with private stuff, will go out tomorrow and get hammered, so possibly no update for tomorrow - maybe a little quicky if we are fancy, haha