Home > Mobile > Follow-along Campaigns

---

One profitable mobile pops campaign by the end of the year (31)

---

Hi,

I have been a member of stm for a month already and I am really excited to see so much information and so many exerts in one place. Really glad I found this forum. I have spent time reading through as much as I could in this month and finally decided to get to work.
My goal is to get to one profitable campaign by the end of this year. I don't have huge expectations by this deadline, just to make it on the green.

After reading the forum, I came to the same conclusion like everyone else here that I should concentrate on promoting sweepstakes through mobile pops.

Here it is where I got so far:

I have setup the following tools:

- Voluum for tracking
- AWS + Cloudlflare CDN (Free)
- Hidemyass to test my LPs and the offers
- I have access to 10 different traffic sources for mobile pops
- I have registered on 5 different affiliate networks, but for start I decided concentrate on offers from Avazu. The reason I am doing this is because they were the fastest to respond to any of my questions.
- I have found a few nice sweepstakes offers on Avazu.
- I tried to test the conversion flow to see if the conversions are reported correctly.
There was an issue at this last step. The conversions were not shown in the Avazu dashboard.

How I have tried to test them:
- Go to the affiliate link provided by Avazu and completed the conversion flow (SOI in this case) with real user data from the selected geo. I am using hidemyass to get the ip from the right geo.
- Checked to see if the conversion is shown in Avazu. Here I can see the clicks, but no conversions, even though I have completed the flow myself manually.
- Asked the AM on avazu and got no answer to this issue.

My question is how do you test to see if the conversions are properly tracked in the affiliate network?
What do you recommend to do next? - My Plan is to test other offers from other networks and see if I can see the conversions in their dashboards.

Thanks.

---

Nice preparation! Sounds like you're ready for battle!

Many aff networks are using systems that can detect when a visit is from a VPN, and won't record the conversion - at least that's what I've been told.

Your AM will probably be able to help you to effect a test conversion - just ask them.

Looking forward to your next update!


Amy

---

Thank you for your help Amy.

Here is my update:

I finally got hold of the AM from Avazu and some of the conversions were tracked at a later time ( 1hour after) and some were not. I have tried hidemyass as well as tor and hoola but some of the conversions were not tracked. I will continue with the offers that tracked the conversions properly and see where I am getting to.

---

Once you setup tracking a few times, you are pretty sure you can set it up for every offer. And when it doesn't work, you can update it later in most cases.

I wanted to briefly mention that there's something else that could happen sometimes. Some affiliate networks, or some advertisers do what is called scrubbing, aka not reporting all conversions.

When it comes to leadgen, double registrations (existing emails) sometimes don't get reported. Sometimes the scrubbing is not as justified as that but just so you know, it does happen, networks rarely, if ever talk about it and just about all affiliates know it happens.

---

By the way, are you Romanian?

---

Hi,

Yes I am Romanian.

---

Where from?

---

Hi Manu,

Didn't get back to you earlier as I didn't see your question here. I am from Cluj.

---

After setting up AWS + Cloudfront +Voluum , finding some sweepstakes offers and ripping some pages from adplexity I managed to run 3 campaigns.

Even though I have read that it is way better to test your landing pages in tier3 countries I have decided that I will invest the translation money to get traffic and in this way will save some time as well.

My campaigns were as follows:

1. Win iPhone7 NZ - SOI
Landing pages: 2
Offers: 2
Mobile ads networks: PopCash; Propellerads
Views: 4000
Conversions: 2
Spent: 15 USD
Revenue: 2.4 USD

2. FREELOTTO - Canada - SOI
Landing pages: 5
Offers: 2
Mobile ads networks: PopCash; Propellerads
Views: 2780
Conversions: 0
Spent: 9.91 USD
Revenue:0 USD
The 5 landing pages were 5 landing pages I have used for the iphone sweepstake and other 5 pages I created only for this offer.

3. Win iPhone7 Ireland - SMS confirmation
Landing pages: 10
Offers: 1
Mobile ads networks: PopCash; Propellerads
Views: 4330
Conversions: 0
Spent: 8.1 USD
Revenue: 0
Here I have tested the top 10 landing pages I found on adplexity in all the English speaking territories for the iPhone offer. The reason this failed might be related to the offer, not necesarily to the landig pages. The offer reqired the users to pay 8 euro via sms. This might be one of the reasons there were no conversions.

My next steps:
- I have decided to retest the landing pages in other geos with offers that don't require sms confirmation. I will restart the campaign in New Zealand and will use the LPs that got most of the clicks in Ireland.
- I will download all the landers I can find for iphone7 from adplexity. In this way I will be sure that I will cover all the angles that were already covered by others and will add my LPs to cover some extra angles.
- I have spend one week cleaning and optimizing the landers I already have (under 3s to load completly). I will spend a few more days optimizing the new landers so they will load in under 1.5s.
The landers I already have are all the landers one cand find on adplexity for uk, us, ca, nz, aus.
- Get all the win iphone7 offers from all the affiliate networks I am in for the English speaking countries and test them with all the landers. In this way I will cover every possible offer I can get my hands on.

If this experiment doesn't get any conversions will switch to promote somethign else and will drop my subscriptions to adplexity as the LPs there don't seem to work.

If you see any flaws in my judgement please shed some light and let me know what I am doing wrong.

Thanks.

---

Nice progress - kudos for taking massive action!

Here are some suggestions to your approach that will to test more efficiently, and improve your chances of finding good offers:

-You don't need to do any translating at all in the beginning unless you really want to. If you rip enough landers in the target language and test them, you will most probably find at least one that will work well. After that you can always get the best lander "properly" translated.

-You can test so much more efficiently if you'd pick a single prize (e.g. iphone 7) and a single language (e.g. English like what you picked; although there are other languages that are spoken in geos with less competition, that may make better choices while you're learning the ropes), and just prepare one set of landers in that language. That way you can test multiple geos without having to spend time making landers for any one geo.

So for example - you could use those 10 landers for Ireland to test NZ offers as well!

Also consider making a compliant set of landers based on the set you originally ripped and fixed up. Most of what you see on adplexity are aggressive. If you take out the aggressive ad text (e.g. by replacing "you've won" with "you have a chance to win") and the brand logos, you can submit the compliant landers to your AM for approval. That way you can use the aggressive set for offers that don't require lander approval, and use the compliant set for offers that do require lander approval.

-In the initial testing stage, it would be good to ask AMs for 2-3 proven offers that have converted for other affiliates, and use them to cut as many different landers as you can rip and fix up. Having 10 like in your Ireland campaign is great - and if you had included 2-3 proven offers instead of just one, you would have increased your chances. During this stage, you don't need to have the best offer. You only need at least one that converts well enough to help you cut to a winning lander.

Then, after you have a winning lander, you can use it to test as many offers as you can find for this geo and offer-type (e.g. iphone 7 offers for Ireland).

Great! Except: Don't judge landers by the number of clicks! I can put a piece of code on any lander and bump the CTR up to 90%, but it doesn't mean it will convert!

I've seen some of the best-converting landers having lower CTRs than many of the other landers in the same test.

When cutting landers - please check for statistical significance: http://stmforum.com/forum/showthread...Banners-Part-1

Another suggestion: I understand that you've spent a lot of time prepping English landers, but for next time, it may be better to choose another language to target. English-speaking geos tend to be competitive because - well - many affiliate marketers already speak that language to some degree. That's the same reason why you picked it yourself, no?


Adplexity is a great tool, and there are lots of good landers to be found. You've only tested 5 offers! Keep in mind that not even the best lander will convert a dud offer. (Not saying your offers are duds - just listing that as a possibility.)

One of our members, whose follow-along I've been following, has just reached $1.5k/day in revenue - he talks about the importance of testing lots of stuff in this post:

http://stmforum.com/forum/showthread...l=1#post294217


Looking forward to your next update!




Amy

---

Hi Amy,

Thank you so much for your help.

Really opened my eyes here. I was spending way too much time selecting the landers I considered to be the best when I should have tested more landers.

I will retry to promote the iphone offer in tier2 and tier3 countries and will try to rip all the landers I can find for those countries and reuse the landers I already have for IE and NZ.

I will spend the next few days to find the offers and set everything up and will update this followallong after I will run the next campaigns.

Thanks,
Vlad

---

Oh, missed this as well, haha.

I have a high school friend who lives in Cluj now. I'm Romanian too by the way, that's why I asked.

Will be checking out your follow along every now and then and chime in with some advice when I can, and @vortex doesn't beat me to it already

Good luck!

---

Hi Manu,

Nice to see other Romanians around. I will message you in private if that's ok.

---

Update:

Since my last post I took Amy's advice and I have concentrated on one offer, one language (even though I went for English as it was easier and faster for me at this point). I went with the iphone7 campaign because I had the landers ready to be launched.

This time I tried to get traffic from UK and I did at 1,16 CPM from PopCash and PropellerAds. After running traffic to the landers for one day (21.22 USD on PopCAsh and 11.43 on PropellerAds) I stopped as I got only 3 conversions in total and something seemed to be wrong. I realised that the Cloudfront CDN was not configured properly and the loading time was too long (more than 4s).

I spend another day setting up CloudFront and optimising the loading speed of the landing pages. Now any landing page loads in under 1.5s.

I have set everything up in Voluum now and I just started the same campaigns with the same traffic sources.

Will post the stats here when I will get enough traffic to make it statistical significant.

---

Great work on decreasing lander load times! At 4s they really didn't stand a chance.

1.5s is OK, although not great. But you can always optimize further after finding a good offer.

UK is pretty competitive. Let's see how you get on first, but for your next geo I would recommend choosing a less-competitive geo.

There are less-competitive geos that also speak English, but generally speaking, it would be best to start with a non-english-speaking geo to avoid the worst competition. Too many people know English - that's the main reason YOU are wanting to target English geos yourself, right? Well, same goes for a million other affiliates.

Latin Spanish will allow you to target almost all of the LATAM geos (except Brazil which speaks Portuguese), and Arabic is spoken in a lot of African and Middle-Eastern geos. So if you want to save time on preparing landers by using one set for multiple geos, either of these languages would make a good choice.

Some of the Asian geos would also make great choices - e.g. TH and ID have good traffic volume, although it's more difficult to find offers for ID.

Targeting the above lower-tier geos has another benefit as well: offer payouts tend to be lower, which will allow you to get more conversions, to allow offers and landers to reach statistical significance with less traffic.

Looking forward to seeing some stats!



Amy

---

When it comes to affiliate marketing, these would be my top 3 languages, based on how "scalable" they are:

1. Spanish
2. Arabic
3. English

If you have your creatives in those languages, you can test more geos that you can count. And I mean in that order because Spanish and Arabic speaking countries respectively are the same tier, so have many of the same main verticals. English is less homogenous in that sense, so you would have to jump the vertical a bit more for what works best in each country.

---

Hi

Thanks for the tips guys. I will translate my landers into Spanish.

At the moment I am putting together an automated system to improve lander performance right from adplexity. I managed to get most of my landers to load in under 1s, some in 0.5. All the image files are included as hex code directly in css, so there will be only 3 files loading: html, js and css. All of these are automatically minified and compressed with gzip.

The CDN works perfectly now and the loading time with Voluum 302 redirect is under 1.5 s.

Here is an update from the campaigns.

I ran the campaigns with a bit more budget, to get more traffic because I was inpatient to see if the improved loading time actually changed anything, it didn't. I guess this is because it is a competitive market as Amy said.

Here are the stats from my first campaigns.

I had 5 offers, 11 landers and 2 traffic sources.

Here are the stats

Campaign 1
Visits: 6594
Conversions: 2

Campaign 2
Visits: 14906
Conversions: 1



You can see that campaign 2 had almost double the visists campaign 1 got and only 1 conversion. This is the 2nd traffic source.

Campaign 1 had 2 conversions with half the visits.

Here are the stats for Campaign 1

Landers



Offers



I will not post the stats for the first traffic source as I think they are irrelevant since I only got 1 conversion for 12k+ visits. I won't use that traffic source for this campaign.

I am thinking of setting up a new campaign and only use the 2 landers that converted with the 2 offers that converted. I don't know if this is such a good idea since I only got 2 conversions. Should I move forward with this?

As I said I will also translate the landers into Spanish and try the LATAM offers.

Regarding the latest campaign, with improved page loading time, it didn't change much and I won't post any stats as I only got one conversion there.

---

You have not sent enough traffic to any of your campaigns to make any solid decisions.

Your average offer payout is $2 and you have 11 landers you are testing. Use this formula to get an idea of the minimum amount you should be spending on your tests...

(avg offer payout) x (number of offers) x (number of landers) x (3-5x) = Test Budget

Which is... $2 x 2 offers x 11 landers x 5 = $220

You may question and say that is too much but for the number of items you are testing, that is not too much. If you wan to spend less, then I recommend removing a few of the landers at the beginning.

My general rule for testing is to test 2-3 offers with around 5 landers at the beginning. If I get some conversions, I will then test a bunch of landers and narrow it down to 1-2 good landers.. and then test as many relevant offers I can find and narrow it down to the best.

If you are not atleast -25% or better at that point, then it's probably time to move on.

But you are getting conversions you just haven't purchased enough data to make a clear decision on this campaign.




Andrew

---

I wouldn't recommend doing it this way - the campaign is looking promising at this point, so it would be better to cut as things reach statistical significance.

It's not like you're giving the camp a last-ditch effort by picking only stuff that has converted before giving up.

It's a good approach to run a little bit traffic to both traffic sources to see if one performs better than the other, and pick the better one for the rest of your testing. But it's generally not a good thing to keep both camps running, as you'd be spending twice the money on cutting landers and offers.

And improving page load times is always a worthwhile exercise! Nicely done!



Amy

---

Wow, now I know where I went wrong.

I just finished reading Amy's guide on how to cut landers and I am reading Andrew's followalong.

I have started a new campaign now, with your recommendations: Only 2 offers (tested and the conversion is working ok) and 5 landers(properly tested and with loading time <1.5s).

I calculated my budget with the formula recommended by mrpayne:

(avg offer payout) x (number of offers) x (number of landers) x (3-5x) = Test Budget


I set the campaign on only one source, the one that converted before.


In order to identify the best lander I will use the split test calculator recommended by Amy in her post.

https://www.peakconversion.com/2012/...al-calculator/

I will provide the stats of the campaign here in my next posts.

Thank you so much for your help. Joining STM was the best decision.

---

It's only just the beginning

---

Here is an update on the campaigns so far.


On this campaign, I started to test 11 landers, didn't get enough traffic because I stopped the campaign way too early and decided to retest.

I reduced the number of landers to 5 and chose 2 offers to be able to test properly and with a reasonable budget.

My budget for this campaign was calculated to be 104 USD.

I have ran the campaign and used Amy's calculator to see which landers I have to keep or kill.

Here are the screenshots from the calculator.

First test with the calculator




After this I killed the landers that were marked as KILL in the verdict column

Second test with the calculator




After this test I killed the other landers that were marked as KILL in the verdict

And here is the last test with the calculator



This basically leaves me with zero landers, but I didn't spend the allocated budget. Should I leave the campaign running until I consume the budget or stop it and move on?

Thanks,
Vlad

---

Aiyaiyai! You've used the wrong calculator for cutting landers! It's easy to mix them up - let's sort out the confusion.

The calculator you should be using to evaluate and cut landers, is the one described here:

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

This calculator will tell you which split-test candidate is better, and by how much "probability of being best". It does not evaluate how good each candidate is, ie. they can both suck big time, and one could still be better than the other. With this calculator, I would recommend cutting the inferior when the "probability of being best" for the inferior candidate falls under 10% (actually, under 5% or even at 0% would be better).

The spreadsheet calculator you were using, will tell you how likely it would be for a candidate to end up being profitable, assuming present conditions remain the same. This is usually used for placements and banners. So for pop, assuming you've already identified a good offer+lander+targeting combo and are running at positive ROI, but want to optimize the camp by cutting placements. You would use this spreadsheet to cut placements that will probably not end up being profitable.

(This spreadsheet will give accurate results, but to pay for that degree of accuracy you'd need to spend a lot of money to collect data before statistical significance can be reached. For pop camps which are typically short-lived, you wouldn't normally have the luxury of spending that kind of time. I would recommend using rules of thumb to cut placements instead:

http://stmforum.com/forum/showthread...Rules-of-Thumb)

This is a direct result of having used the wrong calculator. But it's OK - let me make suggestions on what you can do next.

1)Turn the campaign back on and keep running it. You've only gotten 2 conversions so far. It's looking hopeful at this stage, but you'll need to gather more data.

2)Check lander stats regularly using the peakconversion calculator, and cut inferior ones as they reach "probability of being best" of >10% (or better, >5%).

3)Keep cutting until you're left with a single lander. (Or, if multiple landers are performing the same and keep going head-to-head with the number of conversions, then either just pick a winner randomly (See point #3 in caurmen's post here for details), or leave them both running.)

For future reference, it would be better to test more landers in the beginning. Yes doing so may take up more budget to cut landers, but that way you can maximize your chances of finding a good lander. In your first round of testing, your aim shouldn't be to reach profits (although it would be nice if you did of course!), but to find the best lander. When you have a good lander, you can mass-test offers in the second round. If you restrict the number of landers you test in the first round, you'll restrict your success later on. Also, if you only test a portion of your landers in the first round, if you don't see conversions you'd be left wondering "should I test the other landers?" Whereas if you test in the first round all the landers you've ripped, and don't get conversions, you'd be fairly confident that it's NOT because your landers suck - in which case you'd have the option of testing another batch of offers, or moving onto another geo/vertical.

You're making great progress!



Amy

---

Update:

I have restarted the campaign and got more conversions. I have cut the landers that erached under 10% chances of beeing succesfull using the peakconversion calculator.

Here are the stats for the landing pages.

Visits - Conversions
6,236 - 3
5,054 - 2
3,247 - 0
3,122 - 0
2,610 - 0
2,609 - 0

After cutting the landing pages I left with the the first two landing pages and continued with these as the confidence levels for these were pretty close.

I have created a new campaign with the same traffic source, these two landing pages and I have added a few more offers from the same geo to see which one converts better. Is this a good step forward or should I stick with the offers I already tested?

Should I test other traffic sources as well, or should I start cutting placement in the traffic source I am already using?

I have read this post http://stmforum.com/forum/showthread...icient-Testing and realised that I my approach is wrong. I went only with one traffic source from the beginning instead of testing multiple traffic sources and see which one is suitable form the sweepstakes niche.

Is there a list of good pop traffic sources for the sweepstakes niche? So far I have only tested PropellerAds.

---

For future reference: It wasn't necessary to create a new campaign. That way you have data split up across multiple campaigns, which will make them harder to optimize.

Since you're down to the best-converting (or in your case, the ONLY-converting) landers, testing more offers would be appropriate. Just be careful though that when you're testing several offers with varied payouts, you can't really cut landers accurately - because the peakconversion calculator assumes all conversions to be worth the same.

What I would suggest doing, is cut offers in this round. Then, when you have a winning offer, drill down to lander stats for JUST that offer, and evaluate them using the calculator to see if there's a winner.

All the while, it would be good to keep checking the campaign to see if it still has the potential to be profitable. If you're already testing all the offers you can find for that offer type and geo, and the campaign isn't looking like it could end up anywhere near profitable, then there would be little point to continue.

On the other hand, if there are additional offers available that you haven't tested yet, then it may be worth it to run at loss until you get a winner offer in this round, so you can start a next round of offer testing by rotating-in the remaining offers. (Tip: It actually takes LESS money to test all offers in the same batch, instead of splitting them up into separate batches.)

So how would you check if a camp still has profits potential? Drill down into offer > lander > [major traffic segments] and look for green. The more green you see, the more promising the camp would be. All you need to end up with a profitable campaign is a single offer + lander + major traffic segment combination.

A "major traffic segment" is a segment you can target, that will give you enough traffic volume for it to be worth your time and effort in targeting it exclusively. Usually this would be a major OS - Android/IOS/Windows. Or a big mobile carrier if you're targeting carrier traffic. Or even a major browser like chrome.

So - if you drill down into offer > lander > [major traffic segments] and they're all in loss by more than say 2 offer payouts, I would just stop the camp. (Unless, as mentioned before, you have additional offers to rotate-in and test.)


Once you have a winning offer+lander combo, by all means test it on other traffic sources. (That is, assuming your camp looks promising enough for you to run it until you cut down to winners.)

Pop camps typically don't stay profitable for very long, so you don't have the luxury of spending a long time to cut placements to get to green. For this reason, it would be best to test a lot of offers and landers until you have a kickass combination that converts very well and has a good ROI, before cutting placements.

Once you've cut bad placements for a geo+targeting on a given traffic source, any subsequent camps you run there will be cheaper. Basically you'll get a head-start because you'll have already spent money cutting placements once. So if you find that a geo+targeting on a traffic source has traffic that has good volume and quality, then you can run there again and again to reap the benefits.

Hope that helps!



Amy

---

Hi Amy,

Thanks again for your help and your fast reply. It is really helpful.

I just realised that some of the offers I was testing are not working anymore and I don't know for how long this happened. I didn't get any notifications from the AM. I will setup a monitoring tool to monitor the uptime of the offer pages and will restart testing. This time with the winning landing pages and the offers that are still working.

Vlad

---

Oh gee...

Yeah I HATE it when that happens!

This guide by iamattila may help you:

http://stmforum.com/forum/showthread...s-with-Pingdom

Another way to do this: Some networks will send email alerts from a designated email. For those you could use IFTTT to send SMS / Pushbullet notifications to your phone whenever an alert email is received. I finally implemented this after losing $700 in a few hours when an offer went down.




Amy

---

I found a pretty good solution for this issue with the offers going down.

I have setup a trial account on https://www.site24x7.com. Here you can monitor your offers pages and look for specific keywords on the offer page.

I have setup the monitoring tool to look if the keywords "iPhone" or "Samsung" are present on the offer page. If the word is not found I get an alert on slack.
You could get sms alerts as well, but you have to buy credits for that, maybe here you could use ifttt to automate it further and receive sms.

A nice feature of this monitoring tool is that you could choose the location from where you want to access the offer URLS, so you can monitor it from the geo you are promoting in.

This is how I am planning to use the monitoring tool:

1. Before I start promoting any offers I am adding them to the monitoring tool to see if they have down times during the day. If they do have down times more than 25% of the time I am not planning to promote the offer as it is unreliable and will just waste money on traffic.

2. After I chose the offers and start the initial, testing phase I am adding the offers to the monitoring tool and forwarding the alerts on slack

3. If I will ever find a winning offer and a winning lander I am planing to automate the monitoring of the entire user journey from my landing page to the offer url and as far as I can go on the conversion funnel. The monitoring tool allows you to record the user journey and in case anything goes wrong in that path you get an alert.

I hope this helps.

---

Upate on the campaign:

As I have described above I have found a landing page that was the best landing page from all the ones I have tested in a previous campaign.

I used this landing page to create a new campaign where I started to test all the offers I could find in my chosen GEO.

Unfortunately, I started the new campaign before reading Amy's advice to not separate the campaigns.

I the end after running the second campaign which included 6 offers until it reach statistical significance I didn't get any combination of Offer >> Lander >> Major traffic segment to get to green. The best ROI I got from this was -46%.

At this point I decided to test 2 new landers. I continued to test with the same campaign, but got the same result.

I decided to drop this geo for now and concentrate on tier 2 geos.

I am planing to start a new campaign this week after I will have enough offers from tier 2/3 countries.

I have a question regarding traffic sources. Is there any place where I can find some good information about propellerads, zeropark, popads, popcash and adcash?

I have used propellerads so far and just realised that most of the offers I could find there for UK were simillar with the one I was promoting.

Will come back with some stats on the new campaign by the end of the week.

---

Thanks for recommending https://www.site24x7.com/ - I'm going to look into that myself later! Pricing is pretty OK too - basically $1/item/month.

Nice test! And yup it's about testing more offers in more geos.

They're pretty much littered across a lot of threads. Perhaps one of these days I'll do some thread-crawling and collect them and post them in one place. For now, please feel free to ask as many questions as you need to and I'll do my best to answer them.

Yup propeller is one of my favorite pop sources - popads is the other one.

Looking forward to seeing your new stats!



Amy

---

Hey Vlad! Sorry to hear that you've lost some money, but I would say it was money well-spent, as you've gained a lot of insights in exchange for that ad spend.

As usual, I'm going to read through your post again and comment wherever I can provide feedback.

I'm curious: What was the percentage downtime like? I'm also thinking that if the "unreliable" ones convert significantly better than the "reliable" ones, that they may still end up being winners in spite of the downtime. But if you prefer not to test those, it wouldn't be a bad choice either, (i.e. results could go either way).


Actually, unless you already have a proven lander for that geo, or one that you've seen perform well across multiple geos in the vertical, testing lots of landers would be recommended.

It would often actually be CHEAPER to test all landers in the same split-test, rather than dividing them across multiple rounds. Reasons are:

1)When you do the next round of testing, you include the winner from the previous round, thus testing the same lander twice.

2)The more landers you include in one test, the greater the performance will be between the winner and the losers, thus the faster statistical significance can be reached.


I wouldn't worry about saturation. Sweeps is a big vertical, and we're looking at the most popular pop sources. You'll just be one more advertiser amongst hundreds, if not thousands.


Actually, testing more than one traffic source is not a bad idea!

However, it really wasn't necessary to run THAT much money to each source. You were basically spending several times the money on cutting the same offers and landers.

For future reference, I would suggest to either just do your offer+lander testing on a single source, or set up camps on 2 sources and run limited traffic to each, then see which one seems to be giving you more bang for your buck and then continue your testing on that source (pausing the camp on the other source).


Good call! Once you have a winning lander though, it may be worth it to retest offers you've cut prematurely.


For future reference, only choose offers and landers that belong to the same type of offer, for example only "win an iphone 7" offers with "win an iphone 7" landers.

Otherwise, you'd need to treat each offer+lander combination as a single split-test candidate - which means you'd need to cut landers for EACH offer separately to find the best lander for that offer (instead of the much cheaper way of cutting based on TOTAL lander stats across all offers).


Great! It's an easy mistake to make. Everyone has done it. Next time you'll be more careful.


A couple of questions here:

-Were you down to your winning offer and winning lander before you did this? Changing too many things at the same time (i.e. cutting landers and offers while changing placements around) may not be such a good idea.

-Why would you want to add placements that have merely converted into another campaign? Are they profitable placements? The idea of sorting placements into different camps, is because they convert at different rates, so you want to set up camps at different bids and sort each placement into the bid where they would make the most profit.


It looks like the landers are performing around the same. This doesn't happen very often. Often there would be a clear winner or two.


Very good observations! Although as I've mentioned before, testing many landers was actually a good idea.


1. MY is OK as a geo, but traffic isn't really THAT cheap, and it's gotten more competitive than before. Try to venture into geos where affiliates can't target using English if you want less competition.

2. I prefer the second approach. I would rip as many different-looking landers from adplexity as I can find, use 2-3 AM-recommended offers with similar payouts to cut down to a winning lander, then use that winning lander to test every offer I can find for that geo and offer type. However, spending a lot of time to perfect landers first MAY be justified if the potential rewards are large. For example, if you're prepping a set of arabic landers to test sweeps offers in 20 arabic-speaking geos, then spending more time working on landers will save you enough money for the extra effort to be worth your while.

3. I don't know how other people do it. Personally I would just pick the traffic source that has a lot of volume for that geo. ZP, PopAds, Propeller for example will all tell you how much traffic they have for a given geo. PopAds and Propeller will even give you an estimate of traffic volumes for the targeting you specify. And of course if I've run in a geo before and have scaled a camp to other sources, I would know which source gave me the best results last time, and would just start at that traffic source the next time I have offers to test for that geo.

4. Nice! I was just about to get around to that!

You've no doubt collected a lot of data. Let's drill down and see if this camp has a chance. Could you please provide screenshots of the following:

Offer > Lander

Offer > Lander > OS

Offer > Lander > Mobile Carrier

Offer > Lander > ISP

Offer > Lander > Browser

Let's start with these and we'll go from there!




Amy

---

Home > Mobile > Follow-along Campaigns