Home > POP / PPV / Redirect > Follow-along Campaigns

---

Sweep - Indonesian Geo - ZeroPark Traffic - Aim is to Get it Profitable !!! (55)

---

Hi Everyone,

CURRENT STATS

I paused this campaign 1 month ago (at the end of December 2017).

At the time I was getting ROI
* WIFI TRAFFIC on Selected Zeropark Sites: -27.08% ROI
* CARRIER TRAFFIC on Selected Zeropark Sites: -14.82% ROI

CURRENT BIGGEST ISSUE - infrastructure.

I'm currently running on a Free CloudFlare CDN, My Server is a Box I own hosted in the USA and the Audience is in Indonesia.
The issue is that I can not get all files fully hosted on CDN and so each user page load results in my USA Server being asked for files.

This has resulted in the following issues -
[1] Increasing Load Times - .I think having a better CDN which fully host files (or a server closer to the action in Indonesia) will improve load times and therefore conversions.

[2] Stops Working Under High Volume - It currently can not handle the volume of traffic from a lot of the high volume sites on Zeropark. So I am missing out on some sites which might actually be pretty profitable.

TESTS TO SETUP GOING FORWARD
1). Using a paid CDN service which fully hosts all files (e.g. Amazon) and running tests under load against the current Cloudflare CDN configuration.
2). IF above does not work - Then try getting a fast server closer to Indonesia.

QUESTION FOR EVERYONE
- Does anyone have general advice?

---

For Indonesia I use Amazon S3 in Singapore combination with their CDN and DNS, works great.

Use these guides, really helpful:

1: https://stmforum.com/forum/showthrea...tcost-way-ever

2: https://stmforum.com/forum/showthrea...iliates-Bother

Good luck!

---

Thanks! Will check this out.

---

It's always advisable to host locally, its makes your work easier.

This might be of help.

https://www.cdnplanet.com/geo/singapore-cdn/

---

I have lots of experience running in Indonesia. The tip that @eurosen gave is what you should follow. Point your DNS to Route53 and manage it there, setup an S3/CloudFront bucket and run it host all your files there. It's a challenge to run ID or IN geo's with server(s) to manage the volume potential.

In addition to your CDN setup, I would strongly look at your landers to ensure they are ultra utlra.. ultra lightweight. Most of the time my ID landers have been smaller than 40kb in size and I high suggest that you optimize the lander so that there are as few requests as possible.

Compress images, then convert images into base64 code and update your lander code. Any external css/js files add those directly into your lander code (in most cases atleast, if its tons of code then seperate is fine.) Whether seperate or inclusive, I would be sure to minify your lander code but be sure to test and make sure the lander works correctly after these steps have been taken.

The goal is super lightweight lander with only the need to request the one file.. aka: the lander .html file, nothing else. Sometimes that is not possible but many times it is, even for sweeps.

Tools to help:
https://kraken.io/web-interface
https://compressor.io/
https://www.willpeavy.com/minifier/
https://www.base64-image.de/

Last tips..

Be mindful of the costs accumulated by Voluum if you scale beyond ZP. And ID sweeps are not always able to be profitable, but its a geo/vertical that you can revisit monthly and will find that it works often. I've done 4-figs/day with it.


Cheers and good luck.


Andrew

---

I would suggest to just use a CDN like Amazon Cloudfront, because you'll probably not want to restrict yourself to running only Indonesian campaigns.

Where you host your files would then be irrelevant - as long as Cloudfront can pull those files and distribute them to servers around the world.

eurosen has linked to 2 fantastic guides on how to set up a CDN and DNS. Here's another:

https://stmforum.com/forum/showthrea...277#post330277

Traffic volume won't be a problem with Cloudfront. Looking forward to seeing you get up and running again - you were close to breaking even before!



Amy

---

Thanks Andrew and Amy!

I'll setup the new CDN and DSN and see how it goes.

Follow up question -
1. Wth the super lightweight landers ... at the moment my lander is 73KB BUT takes 19 requests (due to javascript and external images also to be hosted on cdn)
What sort of conversion gain have you seen by getting this below 40kb and putting everything into 1 file (so only 1 request)?
Are we talking about 5% better then before .... or more like 20% ... or more?

---

And another questions/query.

2. According to the chart at the link below, Cloudflare DNS is significantly faster than the paid NSOne or Route53
https://www.dnsperf.com/#!dns-providers,Asia

BUT they don't say whether this is the FREE Cloudflare service (which would be awesome) or if the fast DNS is on one of the higher end Paid Plan (and which plan).
Has anyone done a comparison between Cloudflare Free DNS and NSOne or Route53?

---

Just use CloudFront and Route53, you can explore others later.

As for the lander, I don't have any hard facts on hand but I have tested this in the past. But we optimize landers very well because at times we are doing 5mil+ visits per day. Every bit helps.

73kb isn't bad and 19 its more than I like for a country like ID but you will not make or break the campaign with this improvement alone. Its a small boost usually but at high volume it helps and the sweeps offers are already low payout so maximize it.


Andrew

---

I haven't seen such a comparison.

What WOULD be nice to see, is a comparison of 2 campaigns that are exactly the same in every way except for the DNS provider - to see how their conversion rates differ (if).

My guess though is that the difference in redirection speed (we're talking about 50ms here if the dnsperf.com data is correct) won't have a very major impact on conversion rates, compared to other factors you can optimize - lander loading speed for example.

I would suggest to not do too much more research before picking a DNS provider - when you have a big profitable campaign you can always do a split-test then. (All you'd need to do is register another domain to use with the second DNS provider, copy the lander under the new domain and add it to the tracker, clone the campaign in the tracker and choose the new lander in campaign settings, and clone the campaign on the traffic source and replace the destination url.)

And we have detailed guides on how to set up Route53 and NS1 - so if you're not wanting to go through the additional trouble of figuring out how to set up for another DNS provider, just pick one of these and move onto setting up camps.




Amy

---

TEST RESULTS

The amazon cloudfront CDN did 31.68% better than my free cloudflare CDN !
Definitely worth the money using a paid service !!!

QUESTIONS

I am using Zeropark at the moment to get traffic.
I'm trying to work out how to get more traffic from the profitable 'targets'.
How do I do this? <--

For example in the stats screenshot below, the 1st line - it says 7,580,727 available impressions and I had #1 spot .... but only got 58391 impressions (which is 0.77% of available).
My daily budget is $100 .. and I was only using $4 a day ... so it is not a budget issue.
My setting are in the screenshots below

STATS




CAMPAIGN SETTINGS

---

31.68% better in conversion rate, or?

Either way - congrats and thanks for reporting back!


The BP value is only an estimate. Right now you're only bidding 0.0001/view or 0.10CPM, which is really low. Try increasing the bid for that placement - you may get more traffic.



Amy

---

At what confidence %? For what data? Can't make any sort of meaningful conclusions off $4/day spend.

Have to be careful with jumping to conclusions when testing things and making sure you test to statistical significance and then some.

If a few conversions on a given day swing the balance in favor of one thing or another, then it's not meaningful data to be making decisions on.

---

@vortex, @thedudeabides -- was 31.68% better conversions at 95% confidence (see screenshot below - where 1st line is new amazon cloudfront).



GOING FORWARD

I will be working on increasing the bid and other tricks to see if I can get more traffic from a profitable 'target' on zeropark.

QUESTIONS

1. Do you find that conversion rate can go up if you bid higher.

For example if I initially bid 0.0001 and get an Cost Per Conversion of $0.15
The simple logic says that a bid of 0.0004 will get a Cost Per Conversion of $0.60

BUT do you find that conversion rate might improve with higher bids (e.g. might be 1st popup shown so more likely to click etc).
Is there a rule of thumb ... or do you need to test this on a case by case basis.

---

I don't doubt for a second that having a server or page served from a CDN geographically closer to user will perform better than one that's farther away. Less distance for the connection to travel = faster load times overall

If the page performed 30% better because it was getting 30% more clicks that would be one thing, but it only got about 3% more so it doesn't jump out me as being meaningful yet, despite what the confidence % may say. Its not uncommon to have two near identical pages perform 10-20% differently over short periods of time just by chance.

Regardless I wouldn't spend too much time stressing trying to over-optimize everything. Get a good enough setup in place, then focus on testing more headlines, images, other angles, etc.

---

Bidding higher WILL often get you better CR, even for the same placement - for the reason you stated (i.e. your ad will be shown earlier in a visitor's browsing session, which would make it more likely for them to be patient in reading your ad and going on to convert).



Amy

---

Thanks Amy - makes sense and I'm seeing some interesting results now.

PLANS GOING AHEAD

I'm going to setup 4 campaigns at different bids and run my whitelisted targets on them.
Want to see which one gets the best ROI and enough traffic to make testing easiest.

QUESTIONS

1. I'm trying to figure out what to split test next when it comes to Offers.
If your experience which is the following more likely to happen (when CTR on your landing pages are the same).
-- Some offers are better than other depending on the Android OS Version of the phone
-- Some offers are better than other depending on the Browser Version of the phone
-- Some offers are better than other depending on the Brand of the phone

---

Whitelisting on pop tends not to give good results though. Perhaps a better approach would be to just gradually increase the bid on each good placement to increase daily profits - and when the daily profit decreases after a bid chance, just revert to the previous one and consider the bid optimized for that placement.

And be sure to only go through this type of trouble for the bigger placements, or you'd go past the point of diminishing returns on your time and effort - valuable resources you could be putting towards new camps/tests.


I find that OS is more likely to result in bigger differences in conversion rates. But why the question?



Amy

---

Hi Amy,
Two things.

[1a] Why the questions about what happens with offers ...
Well I found an offer (Win a Samsung S8) that did 2 to 3 times better on "Generic" Phones then on Samsung, Xiaomi and other major brand (see attached below).

I was wondering which is more likely -
* The offer itself might not be technically setup to take samsung,xiaomi,etc traffic for whatever reason. So another Samsung S8 offer might do better?
OR
* People who have Samsung, Xiaomi, etc phone might just not be interested in winning a Samsung S8. So might be better looking for other offers here (e.g. Win an Oppo Phone for people on Oppo Phones etc)?
OR
* I'm pretty sure with the CTR shown that it is not an issue on my lander ... but is there a way to double check (apart from actually looking at them on all the major phones)?





[1b] Why the questions about what happens with offers ...
Well I found an offer (Win a Samsung S8) that did 2 to 3 times better for the "UC Browser Version 12" than on "UC Browser Version 11" and "Android Webview" (see attached below).

I was wondering which is more likely -
* The offer itself might not be technically setup to better accomodate the newer browser version for whatever reason. So another Samsung S8 offer might do better for older browser versions?
OR
* People who have the older browser versions might just not want to win a Samsung S8. So might be better looking for other offers types here.
OR
* I'm pretty sure with the CTR shown that it is not an issue on my lander but is there a way to double check (i.e. emulate UC Browser 11) phones)?

---

Whether it's the OS or browser or device or brand or carrier - that has the most impact - would depend on more than one factor.

The lander you use may also have an impact. For example, when I was running antivirus offers, using a lander that mimicked IOS system alerts or Android system alerts would sometimes make them convert better for the respective OSs.

Another example: Different browsers can load the same lander/offer page at different speeds:

https://lifehacker.com/396048/speed-...t-web-browsers

...and display the same lander/offer page differently:

https://www.browserstack.com/screenshots <--- here is what you can use to see how your lander will be displayed on each device/browser

...which can also have an impact on conversion rates.

Some landers / offer pages have coding that will optimize them for loading/displaying in different browsers, and some don't - and this can make the difference between being able to convert well for all devices and browsers, vs. only converting well for certain devices and browsers (i.e. the ones they load fast and are displayed properly for).

And your point about people who already have samsung phones may not be as interested in winning another samsung phone, is also valid - I have seen that type of trend during the short period I was running sweeps.

Interesting observations on your part for sure!



Amy

---

LESSONS LEARNT SO FAR

[1] Do not Bid on the same target in more than 1 campaign in zeropark.

I went a little crazy and ran 7 campaigns at 0.1, 0.3, 0.4, 0.5, 0.6, 0.8 and 1.0 CPM on zeropark.
Idea was to get an idea of volume at different bids and also relative profitability.

Problem was that I was competing against myself ... ooops
And the lower bid campaigns were being on most unprofitable spots (i.e. 3rd or 4th pop a person sees, instead of the 1st one)

Paused all the campaign apart for the 0.3 CPM one and then adjust bids on the targets in the campaign itself
(so profitable targets got a bid increase and unprofitable got a bid decrease after each day)

Volume for that campaign went up from 8756 to 38533 impression/day
EPV increased more than 50% and went from 0.000087 to 0.00014
Still a way to go - but getting there!

[2] Setting Backspace to Your Own Page

I'm testing out setting up the backspace to go to either
i). My Page (so they hit backspace and they stay on the page)
ii). Go to a smartlink from Monetizer.

At the moment sending a backspace to my own page is winning!
BUT I am still in the process of blacklisting offers which were not working well in monetizer so it might win in the end.

PLAN GOING FORWARD

1. Keep blacklisting daily the badly performing offers in monetizer (where I am sending backspace traffic)

2. Run over the weekend to get more data.

3. See Results on Monday and then setup new tests.

Tentatively going to test
* Samsung S9 offers (against the current Samsung S8 offer)
* Testing sending English Language Traffic to and English Lander and Offer

---

I believe the issue isn't so much that you were competing against yourself - there are a ton of other advertisers competing with you for the same traffic.

In other words, the same placements in lower bid camps would be showing your ad further into the browsing session anyway (i.e. 3rd/4th pop as you said), and whether it's you or a competitor being the 1st ad shown may not be too relevant.

However: Some traffic networks will assign more traffic to higher-bid camps when you have staggered-bid camps like you did (i.e. cloned camps set to different bids). Some networks are worse than others in the traffic distribution. On a couple of networks I've run on, once I set up more than one camp with the same targeting, almost ALL the traffic would shift to the camp with the highest bid. This makes it virtually impossible to test bids by cloning camps and running them at the same time. On those networks, I'd need to test bids by testing one bid a day to collect data, then compare after several days of running at different bids. I believe this phenomenon is worse on networks where there are fewer advertisers.

At any rate: 7 camps with staggered bids is not crazy at all! I think it's an AWESOME way to test bids. I'm always encouraging people to spend more time on testing bids but few people would test more than 2-3 bids. Whereas you bothered to put in the effort and it has paid off which is very nice to see!

Just one advice: Spread out your bids more. At $0.20 CPM apart, you're not really triggering traffic from different placements. In other words, you're probably getting traffic from the same placements for the 0.1 and 0.3 camps for example. Make your bids at LEAST $0.50 apart. I feel that if you would test higher bids than $1, you'd stand much better chance of getting traffic from premium placements you're not accessing at the current lower bids.

If you have a promising offer, don't be afraid to test high bids, especially in geos that are competitive and have a ton of traffic. I've had profitable camps where I was bidding close to $100 CPM. Those are exceptions though. Normally I wouldn't advise testing above $10 CPM. Hopefully that will encourage you to test above $1.

Also: To save on test budget, consider testing bids in sequence instead of simultaneously, and every time you raise the bid, inherit the placement blacklist from the previous bid. This means: Start at a low bid, cut placements, clone the camp with the same placements blacklisted and raise the bid, cut placements, clone the camp again with the same placements blacklisted in the previous 2 camps - etc. etc.


Building a backbutton funnel is an art.

See this post and the ones below it on how far you can take this:

https://stmforum.com/forum/showthrea...l=1#post281535

But really, taking the time to design such a funnel and rotating so many offers would only be worth the time and effort if you have a huge campaign running high traffic volumes. If you can design a generic one for all your campaigns to increase ROI for all of them, it would be a very worthwhile exercise. But trying to come up with a customized backbutton funnel for every small pop camp would be a waste of time. You're doing well by split-testing between monetizer and your own page. Another thing to add to the split-test is send the visitor directly to the offer page. Yet another one to test is send them to an app wall.


Nice!

I'm not sure blacklisting offers in monetizer would be best use of your time though - unless it's a campaign where you're doing high traffic volume. You may be better off spending that time on setting up more camps or testing more offers, or scaling profitable camps to more sources.

But it's a good thing to play with nonetheless. You're an explorer. I can tell you really enjoy setting up tests and observing results. This quality will propel you to success - even if not every test will end up helping you, the ones that do provide valuable insight will give you a leg up over your competition.


Your posts are a joy to read and reply to - thanks for making me think!



Amy

---

@Amy - Thanks! It is a joy to write these ... it actually forces me to stop, think and do proper planning :-)

And of course your suggestions have made a HUGE difference ... so thanks !

**********************************************
LESSONS LEARNT SO FAR
***********************************************

1. Backspace going to Monetizer was not better then just setting backspace to my own lander.

Also kinda did not like giving up control. On Monetizer, I don't know what offers are converting, just the vertical converting.
Setting up multiple backspace sequence ... with different offers and landers is something I am putting on my list of tests.

This will let me use existing traffic to figure out -
- Test same Lander with different SWEEP OFFERS (e.g. supermarket vouchers)
- Test different Landers variation (same offer) with different ANGLES
- Test different Landers variation (same offer) with different LOOK AND FEEL
- Test different OFFERS (e.g. Apps, Make Money etc)


2. Wifi and Carrier traffic convert at different rates.

Some traffic sources might give you more traffic if you ask for all mobile traffic instead of just a segment, but it is not worth testing until I have really improved my conversion rate first.

So seperate all traffic into WIFI and CARRIER.

************************************************
PLAN GOING FORWARD
************************************************

1).INCREASE PROFITABLE TRAFFIC TO INDONESIAN CAMPAIGN -

I need more conversions per day to be able to get to statistical significance.

I did a few calculations using excel stats calculator I got way back --> http://teasley.net/calculator.html
To get to 95% certainty that 1 test is better than the other takes a minumum number of conversions.
And the smaller the change, the more conversions you need to be sure that it is statistically significant.

If Campaign A is only 11% better than Campaign B - you need Campaign A to hit 500 conversions & Campaign B to hit less than 450 = 950 Conversion Total
If Campaign A is only 22% better than Campaign B - you need Campaign A to hit 150 conversions & Campaign B to hit less than 123 = 273 Conversion Total
If Campaign A is only 27% better than Campaign B - you need Campaign A to hit 100 conversions & Campaign B to hit less than 79 = 179 Conversion Total



See screenshot below for more details


So doing a lot of tests and some of them might actually be better than the control, but the difference might be small (e.g. 10%), so need a lot of profitable traffic to get to statistical significance fast.

I'm currently running on PopAds and Zeropark but at relatively low bids.
Going to start
A). Zeropark, PopAds --> Doing higher bid campaigns to get more traffic (making sure it is profitable)
B). Propeller, PopCash --> Start running campaigns on there

2). LANDING PAGE TESTS - once current tests completed

a). Going to look into setting up those backspace tests - may take time to setup.

b). In the meantime - going to be testing ANGLES on my Lander (with winner offer)


3). GEO TESTING - FIND OTHER PROFITABLE GEOs

I've got a fairly good lander working in the Indonesian GEO for Sweeps.
BUT not quite happy with it. Going to run more tests to improve it.

But once I hit a certain CV% - I'm going to test some other GEOs fast in PopAds to see if there are any other Geos I should look at.

Planning (already done)
-- See what Traffic I can get by bidding the 'average' bid in popads for that GEO
-- Shortlist by assuming I can get 30% profit at average bid and investigate Geos with over $x/day.
-- For Shortlisted Geos,
* Find average payment per lead for Sweep Offer in that Geo.
* See what Conversion per View% I would need to hit 30% profit @ average bid price (popads) with average per lead (affiliate networks)
-- Choose Geos based on Traffic Levels and CV% needed to be profitable.

Did the exercise and was surprised!!
Some places like India were expensive ... while some of the place I keep hearing are expensive, seem not to be?
Well it is worth testing :-)

Planned Test Methdology
- Use Proven Sweep Lander Converted to local language
- Run 2 to 5 Offers from Affiliate Networks
- Test for Each Geo will be using Traffic in Popads (top 80% @ average bid price for geo)
* Budget = (10) x (# of Offers) x (Average Payment Per Lead)
* Be Aggressively cutting bad placements
- See which have most promise :-)

************************************************
FOLLOW UP QUESTIONS
***********************************************

1. @Amy - thanks for telling my about some networks pushing traffic to your highest bids ... it makes sense!

I guess zeropark would be one of those.
Do you know whether popads does the same thing or whether you can run a lot of camps at different bids in parallel safely?

I'm going to run the process you mentioned (clone camps with blacklist one at a time) on zeropark, but if popads can be run in parallel safely then I can run a lot of tests quick :-)

2. Any tips/suggestions or comments about current plans going forward?

Maybe a suggestion of something else I should be doing at this stage?

---

What kind of tests?

It would be good to make sure that the small percentages of improvement will have a big enough impact to justify collecting all that extra data.


Can't fault your process, except: Different geos have offers that convert at different conversion rates, so identifying geos that require lower CR to reach 30% profits may or may not give you an advantage. But it's certainly worth a try and is a very well thought-out approach!

One thing to point out that may or may not be relevant, is that bid prices are often driven by the most popular offers that advertisers (i.e. our competition) are running at any given time. So in a sense, high bid prices may not be a bad thing, because if people aren't making profits, they can't continue to pay such high prices. So sometimes it means that are some great offers out. (Not always true though because sometimes it may be the offer owners running traffic to their own offers so can afford to bid higher, or affiliates running exclusive offers that very few people can get access to.)

So in the end, although this type of calculations will help, only actual testing will give a clear picture. Still, narrowing things down using your method definitely has merits.


You can try running parallel on popads - I've done so with good results. Running parallel will give you faster results (i.e. you can get traffic from the largest range of placements at the same time), but it's also the more expensive way because there would be an overlap of placements between camps, so you'd be spending money on cutting some of the same placements multiple times.


Sounds like a plan! On ZP, because it allows you to bid differently for every placement, another alternative would be to just run a single campaign, and tweaking bids for individual placements.

For example, you can increase the bid for placements that are very profitable (e.g. 50%+ ROI), and repeat until daily profits starts to go down, then revert to the last bid (i.e. the most-profitable bid).

For placements that has hope of being profitable (e.g. -30+ ROI), decrease the bid and repeat daily to see if it would become green, and then, as for other profitable placements, when daily profits starts to go down, then revert to the last bid.

Important thing to note here is that when judging the impact of the bid on a placement's performance, you need to do so based on the daily profits it's making, and not the ROI - it's profits you're wanting to maximize and not ROI. This is a common mistake I've made myself in the past.


Nope. Great progress! Once you feel you have a good methodology, it would just be a matter of testing a ton.

Just be careful that you're not perfecting/refining things to the point of diminishing returns - super-smart people are prone to this trap.




Amy

---

@Amy - thanks for the help. And you're right - got stuck in trying to 'perfect' every little thing, when I should have just been testing the BIG CHANGES.

*********************
* UPDATE
********************

1. TRAFFIC - Got 260 conversion per day at about -35% ROI.

Ran Campaigns in Zeropark, Popads and Propeller - and kept the one where I could get them to -40%ROI after 2nd day.

I'm happy with taking the loss during testing.
And I've using Mobile Optimizer with some rules that are improving ROI day to day anyway

2. TESTING METHODOLOGY -
i). Only test big changes (e.g. angles)
ii). Day 1 - Letting Tests run for about 24 hours and get at least 60 conversions per test
iii). Day 1- If the test is beating the control (but not statistically significant yet) then let run another day
iv). After 2nd Day, if the test is above 85% statistically likely to beat (but not 95% yet) let it run another day
v). On day 3, if not 95% statistically significant - Pause, go to new test.




*********************
* QUESTION
********************

1). Need some help with UC Browser Mobile & Android Browser.

3 Browser Versions make up 70% of my traffic and they have very very different conversion rates.

I'm getting really different conversion figures for
UC Browser Mobile 12 -- CV%: 0.18%
UC Browser Mobile 11 -- CV%: 0.07%
Android Browser - CV%: 0.10%

See attachment below


1a). I am assuming that when Voluum refers to "UC Browser Mobile 11" it means version 11.x of UC Mobile Browser. Is that right?

1b). Is there an easy/quick way to test landers in "Android Webview"?


1c). I asked a developer friend to have a look at a lander of mine.
He said it was broken when he viewed it through a Samsung Phone on a Chrome Browser.
Looking at the stats though, samsung is doing well on chrome browser (see attachment below).
And looking at the stats is works well for the chrome versions.
Could it be that the phone model make a difference "Samsung Galaxy s6 edge" ?
Does the same browser work differently in different model phones of the same brand?
Or is this an aberration?

---

I can't fault your test methodology! It will make you more efficient. It's almost synonymous to picking a random winner when multiple split-test candidates are going head-to-head. Would be a waste of time waiting forever just to identify a winner in some cases.

You've been running this camp for a while now - are you (or the optimizer tool) still finding a lot of things to cut? It may be time to start cutting bigger segments, like browsers in your example - cut the worst-ROI segments, run for a couple days to assess the impact, then cut further if necessary.

I would cut UC Browser Mobile 11 for example (if allowed at the traffic source). However, I would first drill down further into other segments as a second level, for example:

UC Browser Mobile 11 > OSs

UC Browser Mobile 11 > OS Versions

Etc.

If, somehow, a segment of UC Browser Mobile 11 traffic is causing the low ROI, you can potentially start a new campaign for UC Browser Mobile 11 with that segment excluded. I won't usually suggest going through this much trouble to optimize a campaign, but if there's enough traffic volume, then the extra work is justified.

That's how I've always interpreted it! You can verify this with Voluum support though.


I tend to be lazy in this area. Going to browserstack to make sure that my lander will display correctly in all major browsers and devices is all I'd do:

https://www.browserstack.com/screenshots

However, if you're willing to go to length of actually testing the lander in all browsers and devices using an emulator, you may be able to find and fix deeper issues - such as functionality-related issues. I think browserstack provides that as a paid service but have never used it so can't really comment on it. If you do look into this type of testing, I'd be really interested in seeing what you'll find. Pop has become so competitive in recent months, that any additional optimization techniques will help.




Amy

---

@Amy - Thanks for reminding me. I was so used to letting Mobile Optimizer take care of the traffic I forgot to take a deeper look myself


**************************
* UPDATE
************************

[1] Traffic - Had a deeper look at all the segment I could turn off and on in the traffic sources

* PopAds:

I look at all the segments (see attachment below) and worked out what the profit would be if I only allowed components that gave ROI above 0%.



The top ones were 'Device Model'
Problem is that the Voluum Device Models do not line up with PopAds device models
Had to redo my popads link to include the [DEVICENAME] tag and let the campaigns run to get data.

PropellerAds

This was not getting better by cutting sites, since it seems there were new sites being added everyday.
There was no one segment which was better than the others.
So I turned it into a whitelist campaign (only siteids that were profitable).


[2] ANGLES - Testing

Testing angles all the time now.
Have about 10 translated and just lined up based on my own brainstorming
Tested 6 so far this week.
4 Failed to beat control ... the current 2 being tested are 90% likely to beat control
Letting it run another 12 hours to be sure.

Also testing out some indonesian copywriters to see if they can come up with better versions.


[3] BROWSER TESTING

Unfortunately browser stack does not do UC Browser - even though it is a large % of the traffic

So I just got my developer to download the current UC Browsers onto his phone and run usability tests on his browser.
Took him 10mins and we figured out that certain UC Browsers break a lot of javascript functionality !
It is called UC Browser Mini and is sold as a 'speed' browser ... uses a lot of caching, minifying and other tricks which sometimes break javascript.
Got a version of the page developed which works and going to test it tomorrow to see if it makes a big difference.

OFFER TESTING - Tested most of the phone sweep offers for my geo. Starting to look at other verticals which are doing well in ID (see if it beats sweeps).

---

Your excel sheet is awesome! However, be aware that segments are often inter-related and can affect each other. For example, the same browser can be making profits for one device but bombing for another device. I wouldn't worry too much about this for smaller campaigns (fine-tuning for small camps just isn't worth it) but for big camps, that may be worth looking into more.

And yeah it's annoying when there's a discrepancy in the targeting between the TS and Voluum - just one of those little details that can have so much impact.


Having new sites pop up to zap large chunks of your budget is super-frustrating. You need to automate things so that any new sites that appear will automatically get paused after x in spend. Does your automation solution allow for you to do that?

Whitelisting often doesn't work too well for pop camps (compared to blacklisting) but we shall see. If you find that traffic dries up for your whitelist camp, let me know.


That's perfect!!


Woot woot!!

So you have a custom lander for UC browsers - that can give you a major advantage when most other affiliates are probably just either excluding that browser from their targeting, or aren't doing as much profits due to the low CR - which will allow you to bid higher to dominate this segment.

Speaking of which: Once you have a good lander that works well, splitting up UC browser into its own camp and bidding higher will likely get you more traffic volume.


Good idea! But have you checked Adplexity mobile to see which vertical(s) is/are doing the most volume in this geo? May save you some time.



I'm extremely impressed by your analytical skills and tenacity in testing out theories. Everything you're doing now will combine to give you a major advantage over your competition when it comes to ID traffic. Once you've done the legwork, you can use the same placement blacklist and the same lander(s) and just keep testing new offers that come out over the next months.

You've spent a lot of time and money on optimization already. It would be appropriate to starting scaling to other networks to recoup your losses and more - either now or soon.

The more good traffic sources you have access to, the faster and bigger you can scale future profitable camps to. There's a lesson on scaling in the 40-day tutorial that you can refer to when you're ready to start scaling. All you need is an offer+lander combo that is doing well - you need a proven funnel in order to test new traffic sources.




Amy

---

Thanks Amy

I'm cutting one segment at a time. So if I cut devices, I'll have to wait to collect more data before cutting another segment (to avoid the issue above)


I'm using mobile.theoptimizer.io for automation and it is working really well for zeropark and popads.
It is also work well for propeller ... but propeller seems to have lot of zones to get through (and hard to get a daily budget of under $30 to stick).
So whitelisted it. 1 day in and doing well - but will keep an eye on it.


Thanks for reminding me. Will start looking at the more seriously.
I used to do google campaigns ... and some of those I'd have running profitably for 2 to 3 years.
Got used to having much longer to collect data before having to make a profit.
(and also just my inner mad scientist wanting to work out exactly what was happening)

Forgot that pops usually has a more limited shelf life lol

---

**************************
* UPDATE
************************


Just launched a new online business last Saturday so I've only been testing angles on this project recently due to lack of time.
Have been using Mobile Optimizer to optimize current traffic by cutting bad placements.

ANGLES - TESTING

3 Sets of Test Done
Overall Improvement of 81% !!!

i). Have done 2 tests of angles I came up with myself and got translated by OneHourTranslation.
1 Beat Control by 32%
1 Failed to Beat Control

ii). Have Hired 2 Copywriters who live in Indonesia and are Native Speakers.
Tested 2 Variations from the 1st Copywriter.
Variation 1: She just made sure it read well in indonesian. She did not change the angle. Beat Control by 13% with 89% certainty
Variation 2: She made her own variation where she did change the angle. Beat Control by 37% with 99% certainty !!!


Currently Testing Variations from the 2nd Copywriter.

LESSONS LEARNT

1). OneHourTranslation are ok ... BUT Much better to hire a local copywriter who can make sure it reads well, can come up with their own angles to test and explain why something may or may not work based on their knowledge of their fellow countrymen.

---

Holy crap! Well done (understatement)!

What kind of ROI is the campaign at now?



Amy

---

Lander fatigue can be a problem as you've described, but it's not a very big issue when compared with banner fatigue.

Somehow, banners burn out a LOT faster than landers do.

If you look at how landers evolve over time, you'd notice that they don't tend to change a whole lot over the course of a year, or even two.

But if you have some landers that are performing around the same, it really wouldn't hurt to keep them in rotation - just so visitors would see "fresh ones" like you said.

As for how long to run a test - statistically speaking, as long as you run the landers during the same time period, and the best one beats the worst one by 90%+ certainty, that would justify cutting the inferior. Unless you're operating in a very small geo, I'm guessing the new lander shouldn't all of a sudden start converting worse than the older lander just because the novelty has worn off - at least not over the course of days or a few weeks.


Absolutely with you on this!

While testing, you need all the traffic you can get - I would only cut the worst budget-drainers. (However, for people that have a small budget, they may want to only use the best-converting traffic to do testing, just to save on budget - the downside of course is the testing would take longer.)

Focusing on improving the funnel so that it can make MORE of the available traffic profitable is smart. Many people make the mistake of being lazy in testing landers and offers, and try to get green by cutting stuff - and would frequently end up with little bits of traffic that make chump change.


Why would you want to figure out the target conversion rate? Say you already have this - how would you use it to optimize the campaign? Cut all placements that don't meet this target?

When you bid differently, you will trigger traffic from different placements, so you'd be dealing with multiple moving parts.

I'll wait for your answer before doing more thinking.



Amy

---

@Amy - Thanks. I was going down the wrong path of thinking with Target Conversion Rate (since as you say, this might vary at higher bids). Now I'm scaling once I'm not making 50% improvement to my Lander in the last week!

******************************
TESTING RESULTS
********************************

[1] Carrier Landing Page - Been testing but have made no improvement over the Control in the last week
[2] Wifi Landing Page - Been testing but have made no improvement over the Control in the last week

Time for me to Scale Carrier Traffic :-)

AIM is $50/Day Profit by 31st March 2018


******************************
PLAN MOVING FORWARD
********************************

[1] Setup so -
- 80% of Traffic going to my Control Landing Page
- 20% of Traffic to continue testing New Landers and Offers

[2] Really Restrict Rule. Going from Wanting Lots of Traffic to Wanting Profitable Traffic
Was
- Pause if 2xSaleAmount Spent and 0 Coversions
- Pause if ROI is below -20% after spent 4xSalesAmount in the last 30 days
- Pause if Lander CTR is below 1% after 200 clicks
- Pause if Lander CTR is above 25% after 40 clicks (bot traffic)


Now
- Pause if 1xSaleAmount Spent and 0 Coversions
- Pause if ROI is below 30% after spent 4xSalesAmount in the last 30 days
- Pause if Lander CTR is below 1% after 200 clicks
- Pause if Lander CTR is above 25% after 40 clicks (bot traffic)
- Pause if Lander CTR is above 50% after 20 clicks (bot traffic)
- Pause if Lander CTR is above 100% after 5 clicks (bot traffic)


[3] Start Making Campaigns with Higher Bids on the Traffic Sources I am Already On
-- PropellerAds
-- PopAds
-- Zeropark (this is tricky since there is a LOT of bot traffic on here)

[4] Go onto other Traffic Sources once got (2) done.


******************************
QUESTIONS
********************************

1. Any tips for working with Zeropark?

I started at the cheapest bid ($0.1CPM) and am getting a LOT of bot traffic.

What I'm currently doing is
i). Running at lowest bid of 0.1CPM at a restricted daily budget - to Build a Black List of Site and Find Profitable Sites
ii). Profitable Sites (more than 3 leads a day) go to a WhiteList Campaign.
iii). After 3 Days (or if 0.1CPM campaign profitable) start running next campaign at higher bid (0.3CPM) using black list
iv). Repeat step i to iii above until stops getting profitable

---

Nice! Have fun scaling!

A few things to point out:


1)I would reconsider about basing lander- or placement-cutting decisions on lander CTR.

The assumption a lot of people have is that higher CTR should translate into higher CR - which is often NOT the case.

I've seen WAY too many loser landers (in terms of CR/ROI) with higher CTR than superior landers, and too many winner landers with lower CTR than inferior landers.

For cutting landers, use a stats calculator, and cut an inferior lander when the "probability of being best" of the superior lander reaches 90%+.

For cutting placements, I would base cutting decisions on how many payouts the placement is in loss by (e.g. 2-4x payout without conversions; in loss by 2-4x payout).

Having said all that - there IS a time when it WOULD make sense to base cutting decisions on CTR.

When a lander's CTR is so low that it would require an impossibly-high CR in order to break even, then it would probably be a good idea to cut that lander without testing it to statistical significance.

When a placement's CTR is a lot higher or lower than the CTR of most of the other placements, it usually means something is up with the placement. If you're running in a geo that has a ton of placements and you're wanting to cut faster, identifying these placements and cutting them can save you quite some money.


2)Regarding "[4] Go onto other Traffic Sources once got (2) done." Are you talking about done cutting placements or landers?

There's no need to wait until you finish cutting placements. Basically once you have a promising offer+lander, you'd be ready to scale to other traffic sources. Main things to do at each traffic source would be a)test bids and b)cut placements.


3)Regarding getting a lot of bot traffic, I would suggest to perform a bot test. It won't cost you any extra, and should not impact on campaign performance. A bot test can help you identify placements with high percentages of bot traffic so you can cut them without having to spend 2-4x payout without seeing conversions.


4)Regarding your bidding and whitelisting approach:

That approach could work! Sometimes though, whitelisting may not work very well on pop sources.

If you're finding that you're not getting traffic volume to your whitelisted camps, try this blacklisting approach instead - which is similar to your approach:

-Run at lowest bid, blacklist a batch of placements.

-Increase the bid to get access to more placements. Run and blacklist another batch of placements.

-Increase the bid again and repeat. When the overall profits of the campaign starts to drop significantly, revert to the previous bid.

-Throughout, don't forget to implement bot test to eliminate placements faster.

Or, you can also try an alternate approach:

-Run at lowest bid, blacklist a batch of placements.

-Clone camp and inherit the blacklisted placements, and increase the bid. Cut another batch of placements.

-Clone the most-recent camp, inheriting placements blacklisted in both camps before this new one, and increase the bid. Cut another batch of placements.

-Etc...until higher-bid camps are no longer hitting new placements.

This last approach doesn't always work. On some traffic sources, when you set up multiple camps with the same targeting but different bids, the camp with the highest bid would get most of the traffic, leaving little traffic for the lower-bid camps. If you see that happening, take the first approach of running a single camp.


Happy scaling! Eager to see how you'll get on!



Amy

---

Hi Amy,

Which was regarding

Sorry for the confusion. Those Rules above are for cutting placements ... not landers.
Said 'Lander CTR' to mean CTR of the placement on the lander ... which yeah .. does sound confusing lol.

For landers, I'm only choosing based on which is the most profitable (not ctr)
Question1 - I'm cutting placements at 1xpayout without conversion. Is that too agressive?

Question 2 - I'm trying to get placement with over 30% ROI. So instead of cutting if they hit a loss of 2-4x payout, I have a rule which says after 4xpayout in spend - the placement must stay about 30% ROI. Is this ok in your experience, or am I missing something by doing this?

I'm currently doing a crude form of that by cutting placements with
- Pause if CTR is above 25% after 40 clicks (bot traffic)
- Pause if CTR is above 50% after 20 clicks (bot traffic)
- Pause if CTR is above 100% after 5 clicks (bot traffic)

Question 3 - Do you think this is ok?

Question 4 - Do you know whether this happens (i.e. higher big campaigns getting traffic) in popads, zeropark or propellerads?

---

Congrats on hitting your first good milestone. That's where it all begins, that first profit day or profitable campaign of $xx-xxx/day.

Keep at it and before long you will be doing big numbers

I haven't read all your thread yet but I can see you learned the importance of testing different offers or even the same offer on different networks. This element has the biggest influence on the campaigns performance, so always start here.

The next major thing is the type of traffic you are sending (assuming you have a half decent lander to work with). I like to test in groups of 3 geos, one source at a time.

Example:

GEO's - AU, IN, TH

Launch all 3 on Zeropark part > see which geo/offers convert. If nothing looks promising after initial test > retest on traffic source #2 > and I will do that for up to 3-4 sources before moving on.

BUT, if only TH does well on ZP > then keep testing on TH but take the other two and test another source.

ALSO, if IN does really bad on ZP but you can clearly see there is one offer doing all the conversions > only use that offer for now when going to testing other sources.


Congrats on your progress!



Andrew

---

Wow! Congratulations on the profits! xxx/day and xxxx/day shouldn't be far off - the process is the same, just more scaling and more extensive and efficient testing to keep finding more profits to scale.

And thanks for keeping this follow-along going! It would be great if you could continue until you reach 3 figures, then 4, and maybe 5...would be so valuable and motivating for other up-and-coming newbies!

I've somehow missed your update from a week ago - apologies! Will play catch-up below.


Ah OK noted!


Question1 - I'm cutting placements at 1xpayout without conversion. Is that too agressive?

Question 2 - I'm trying to get placement with over 30% ROI. So instead of cutting if they hit a loss of 2-4x payout, I have a rule which says after 4xpayout in spend - the placement must stay about 30% ROI. Is this ok in your experience, or am I missing something by doing this?


Where did you get these rules of thumb from?

There are lots of landers I've run where CTR is above 25% or even above 100% (including backbutton triggers) that were big winners. You may cut out a considerable amount of good placements by following these rules.

And if you must cut placements based on CTR, I would suggest to look for a range where most of the placements that have converted, fall under. For example, if your converting placements are mostly in the 30-70% CTR range, then maybe cut placements that have CTRs of <10% and >90% for example. That would make more sense IMO.

But yes - implementing the bot test script will not impact on your campaign performance and can save you money. I'll make a post to do a step-by-step on caurmen's bot test methodology - soon!


AFAIK this doesn't happen as much at these particular networks. It's also not just network-dependent. It's also geo-dependent. I don't have proof, but experience tells me that running in smaller geos with fewer advertisers tends to be more prone to having this issue.


VERY valuable advice!

If you have a really good offer, you don't need an outstanding landing page (but it still needs to be decent!)

Testing the most popular ripped landers and cutting down to a winner will do in the beginning - it's using it to mass-test offers that will be the most value-added part of the testing process.

Having said that: Sometimes, just changing the headline or image on the landing page can increase ROI by 100%+! I've done it before. But I find it easier to achieve big increases in ROI by testing offers - so more often than not I'd test offers first, and THEN lander variations / custom landers.


The first 2 items on your list WILL accelerate your learning and deepen your knowledge, but will not likely increase your profits by huge amounts.

If you're wanting to milk the current campaign, I would suggest to take this opportunity to scale it to as many pop sources as you can find. You'll see profits grow in leaps and bounds - very satisfying!

Either approach would be fine - depending on whether it's learning or immediate profits is your priority. Or, you can do both.


Love this! Good guide.


Such as? What kinds of questions? Is this something you can identify in an online forum in the niche? Could save some time and money.


Wise! MrBraun recently published a post where he suggests testing offers across different verticals. If you have adplexity, you can find out which verticals are getting the most traffic in a given geo:

https://stmforum.com/forum/showthrea...m-your-traffic


And if you're not wanting to cannabalize your profits, assign only a small percentage of traffic to these new offers.


Just keep in mind that unless you know how to keep your landers out of Adplexity and other spy tools, your new landers will be ripped by others as well.

Best way to stay under the radar would be to keep the same look-and-feel but change the angles of the advertising text which would be more subtle. If you create a lander that looks completely different from other landers it will stand out and get ripped even faster.


These have potential to really give you an edge over the competition - because they take more time to set up.

Giving people choices and then building email lists is a VERY smart way of turning broad traffic into targeted traffic. But you'll need to figure out a monetization funnel that will result in profits - that's the hard part.


I think I've answered these above. Lastly, I want to say that if I were you, I would focus on scaling right now and save most of the learning for a bit later. Seeing big numbers will do wonders for motivation - which you WILL need after this campaign dies and you're needing to find new success to replace it. Pop camps and CPA offers usually don't last very long, so milk it to the best of your ability while you can!

Also - in order to test new traffic sources you'd need to have a good offer+lander, so now would be the perfect opportunity to scout out new sources. The more sources you have at your disposal, the faster and bigger you'll be able to scale all your future camps!

This is very exciting - looking forward to your next update!



Amy

---

@MrPayne - Thanks for the Tip. I will Max out Indonesia first and then use that process when going into the next Geos

@Amy - Thanks for the reality check :-) What you say makes total sense, so instead of focusing on increasing ROI this week - I'll focus on milking as much profit out of this as possible.

So going to be focusing on Traffic Sources this week and Getting Good at them
Still going to keep testing to increase ROI, but maybe spend 10% of my time there and 90% of increasing traffic this week.

And yes, definitely planning on keeping this follow-along going !

---

PS: What are the big causes of campaigns dying, and is there anything I can do to stop it from happening or slow it down (so I have more time with a profitable campaigns)?

---

*****************************************
FOLLOW-UP
*****************************************

I had run for a few weeks and when looking through the zoneids in propeller found that most traffic on my current lander were much much less profitable in the were -
* above 100% CTR after more than 5 visits &
* above 50% CTR after more than 20 visits

So I've been using that rule on propellerads to try and cut zoneids very very early
I'll have cut some profitable zoneids - but that is offset by not needing to spend 1x or 2x Payout on those unprofitable zoneids

NB: I saw the same results when looking at old data from zeropark test

Not sure if that will hold on other traffic sources.
Might need to run it without this rule when I start on a new traffic source to get a baseline with my current lander

NOTE: I'll need to revisit this if the Lander CTR get better due to
* Developing a better lander
* Tightening the traffic I'm buying to those who are more likely to want the offer.


*****************************************
PLAN GOING FORWARD - INCREASE PROFIT
*****************************************

[1] PropellerAds
* Continuing to find more profitable zoneids
* For profitable zoneids - testing ways to get more traffic to them
- Adblock Traffic
- Increase Frequency
* Also testing smart bid to see if that increase profit


[2] Zeropark - Focusing Here Right Now

I've ran a Standard RON Popup Campaign a month ago before pausing it.

Planning to
* Restart the 'Standard Ron Campaign' at Low Bid 0.1CPM
* Start a New 'Premium Ron Campaign' at Low Bid 0.1CPM
* Start a New 'Domain Traffic RON Campaign' at Low Bid 0.1CPM

Going to run the same rules I've been running on PropellerAds to cut bad targets.
No plans at the moment to cut sources.

Planning to run each of these campaigns until either profitable or total campaign loss is -50USD.
My gut feeling is that the campaigns will be unprofitable - but I'll be able to identify targets to add to a whitelist.


*****************************************
QUESTION
*****************************************

1. Does the above make sense?
2. I've got a Target Blacklist for Zeropark.
Anyone know how to setup a target blacklist on a New RON Campaign in zeropark?

---

https://stmforum.com/forum/showthrea...formance-Dives


Ah! So you created that rule-of-thumb based on analysis of stats. That's great!

Just be aware that every camp will be different depending on the landers you're using.


Sounds good!


Why not set up staggered-bid camps from the start?

Are you trying to compare the 2 types of traffic to see which one is best?


Sure does!


I don't know of a way to specify a blacklist beforehand while creating a campaign on Zeropark - if you find out please share!

Commercial automatic optimization tools may allow you to do that. You can also write a script to do blacklist those placements as they appear. OR, you could just run a bit of traffic (set a minimum budget), then pause and blacklist those sites before resuming.

Massive progress ahead!



Amy

---

DAYUM))))

No matter! Do NOT be complacent - keep scaling steadily to other traffic sources one at a time and you'll more than surpass $150 - a few dips and rallies later.

That's a simply amazing job!!


ALWAYS a smart thing to do to minimize risks - and help to scale bigger faster!


They don't - unless something has changed between now and when I last ran there.

Actually I thought you were running on another source. For Zeropark, it's not really necessary to run staggered bid camps because you can bid on a placement (or what they call target) level.

So, if a placement is profitable with a good ROI, you can increase the bid to get more traffic, and if a placement is unprofitable but not in excessive loss, you can decrease the bid to spend less money.

When you increase the bid though, be prepared to expect ROI to drop - the idea is to maximize DAILY PROFITS, not ROI.

And when you decrease the bid, be prepared to get less traffic.

So - pros and cons. Basically your goal is to maximize the daily profits of every placement by adjusting the bids.


The way I understand it is that both types of RON campaigns will get you access to different targets. It's just that "Premium" will have better traffic quality (i.e. traffic that converts better) than "Standard". Pricing tends to be higher.


Yup that's what I meant! But as was mentioned above, this is not as crucial for zeropark camps - you can achieve the same by running one camp and adjusting bids on individual placements/targets.


Yup that would be another good bid-testing approach as well! Trying different methods to see which one seems to give you the best results is one of the best things you can do to sharpen the tools in your arsenal (so to speak).


If the geo is small I would forget about the 0.1CPM - you may not get enough traffic. But we'll never know for sure until you test!


We got a reply and confirmation from member duiyao:

https://stmforum.com/forum/showthrea...l=1#post341072

That's clever. I wasn't aware you could do that. Feel free to test that method. I will too next time I run on Zeropark.


Hear you completely. Let's try the method in the other thread first.


All sounds good! This is very encouraging and exciting progress - let's see how big you can scale this!




Amy

---

**************************
WHAT HAPPENED
***************************

Sorry for not updating this for a while.

The offers I was running hit a cap and then were paused.
Then I did something really stupid - I got sucked into another project I was running (ecommerce site).

Switching between 2 or 3 different types of projects/businesses (pop affiliate marketing, ecommerce site, real estate) meant I was not really making any progress on any of them.
So I've put everything else on pause and doing only Pops now.


**************************
WHAT I DID WRONG
***************************

This is what I think I've done wrong with Sweeps.
I've spent too much time on the same project (4 months for 2 sweep projects in only 2 geos)

I focused WAY too much on trying to perfect the project I had running (i.e. to get all the profit I could), instead of what I think I should have been doing
* Going wide into lots of different geos to find the most profitable ones first before spending lots of time on a geo.
* Getting used to Launching Project in different Geos so I can master the process needed for profitable affiliate marketing. As they say, practice makes perfect.

**************************
PLAN GOING FORWARD
***************************

My plan going forward is to get AWESOME at Pop Sweeps Affiliate Marketing.

To do that my goal is to run a LOT of projects in different geos to get good at the whole process (practice makes perfect)
At the moment I'm planning to get at least 15 pop sweep projects launched in the next 6 weeks.


**************************
COMMITMENT
***************************

[1] Going to Get 1 New Sweeps Project Launched by Sunday 29th July in a new Geo

**************************
QUESTIONS
***************************

[1] I've noticed that offers now want the Sweeps Landing Page to be HTTPs. Can this be done in Amazon S3 hosting (anyone done instructions for this)?

[2] My Amazon Web Services Account (for hosting) got hacked and someone ran up a $10k bill using computing resources and the got it suspended!

I'm trying to get the account unsuspended as well as taking that $10k for computing resources out of the bill but their support is really really slow (on the basic plan and taking 2 to 3 days for them to reply to email and I can not upgrade the support plan while the account is suspended).

Does anyone know of another way to contact amazon web services to get this sorted quickly?
NB: Currently have basic support plan and can not upgrade it while account is suspended.

Also anyone got any tips on how to get them to take the $10k out of the bill (i.e. what the people who hacked the account used) - or am I going to need to just pay it if I want to continue working with Amazon Web Services?

---

yes it is quite straight forward, if you are using cloud front cdn , you could just just the cloudfront https distribution address. the downside is the url will be odd: dxxxxxx.clouldfront.net. but for mobile pop it could be a non issue.

to use your own domain, you can take a look at this, it is not difficult either

---

@edgekaos - Thanks !

---

Thanks for coming back to update us on what you've been doing! Real good to hear from you!

I wouldn't say what you did was "stupid". Way too harsh IMO.

You tried to juggle multiple things, found that it wasn't your style, and now you're choosing to focus on one thing instead. Lesson learned. Life is a series of lessons.


On spending too much time on the same project - I agree, but every affiliate has to start slow.

Now that you know what you're doing, you can launch multiple campaigns at the same time - e.g. in several geos, and scale the one(s) that look promising.

As for squeezing all the profits out of the campaign you were running: Focusing on scaling to as many traffic networks as you can is important, but spending time on cutting a bunch of small placements manually may not be worth the time. When it comes to optimizing, keep in mind the point of diminishing returns - don't sweat the small stuff.

Going wide into lots of geos is ultimately a good plan, but focusing on one geo or a few geos at a time would actually be a better idea, because it takes investing money into identifying the best and worst placements in each geo on each traffic network. Once you've made this investment, you can keep testing new offers for cheap. But initially the investment required to cut placements can be substantial (especially for the bigger geos). So be sure that you don't exhaust your budget investing in this way into too many geos before you can make back that investment from any geo.

If you have made this type of investment on, say, 10 geos across 20 traffic networks, then you really have something going. You'll have at least one decent lander in at least one vertical for every geo, and you'll already have a blacklist for every geo on every traffic network. So you'd just need to regularly look for more offers to test for these geos and vertical, and scale the good ones to all traffic networks.


Testing wide is good! But please do mind my warning above and don't spread yourself too thin and run the risk of exhausting your budget.


Wow - really sorry to hear about that! $10k - some people have no morals.

I can't really offer advice as I've never run into this situation myself, but gut feeling tells me that paying $10k for charges you weren't responsible for racking up is not a good idea.

I would sooner put that money towards a basic hosting plan and a CDN other than Amazon Cloudfront, or somehow try to sign up for another amazon account using another email (I don't know whether this is permitted - so be sure to check their TOS thoroughly before attempting to open another account).

Best of luck on resolving this issue soon! Do keep us up to date on your progress.




Amy

---

Hi Amy,
Thanks for the tips !

***********************************
MY CURRENT GOALS
***********************************
- Get Really Good at Sweeps on Pops
- And as a result make money :-)


At the moment I've:
* Got 1 Landing Page for Sweeps which is proven to do well
* Gotten Good at 1 High Volume Traffic Source I spent a lot of time on (PropellerAds)
* Did only OK on the other 6 Traffic Sources I tried since did not devote enough times to each (tried to do them all at the same time).

And I have to STOP myself from spending too much time on only 1 GEO and being a "perfectionist" going for the small gains for too much effort.

**********************************************
SO THE PLAN FOR THE NEXT 3 WEEKS:
***********************************************

[1] AT LEAST 1 NEW PROJECT A WEEK - Do at least 1 New Geo for Sweeps a week using my Best Landing Page and running on the Traffic Source/s I know best (which is only PropellerAds at the moment).

[2] IMPROVE SWEEPS ROI - Keep improving Lander and test different angles and sales system (e.g. email)

[3] SCALE UP TRAFFIC ON PROFITABLE CAMPAIGN - Get good on 1 other Traffic Sources at a time and Keep Testing Other Offers

[4] KEEP PROFITABLE CAMPAIGNS RUNNING - Get Proccesses Developed to keep profitable campaigns making money as long as possible.


The aim is to eventually get into 10 to 15 Geos and be able to keep them running profitably with less and less of my time.
Will start with at least 3 Geos in the next 3 weeks and then re-evaluate the plan.

**********************************
THIS COMING WEEK
**********************************

It looks like some of ID offers have become available again :-)
So will be restarting those this week (while sorting out the Amazon Web Services Issue).

---

Sounds like a plan! How are things coming along?

Have you decided what to do about that Amazon issue?



Amy

---

Hi Amy,

***********************************
LAST WEEK
***********************************

[1] Amazon Issues -
* I got a 2nd account and made sure to set multi-factor authentication on it to reduce the chances of it being hacked. I've been using that account right now to get work done.
* With my 1st account that got hacked: It has been 20 days since I first raised the issue and I've been playing email ping pong with amazon with about 3 days between emails (since I had the free support plan, and could not upgrade while account suspended). The last email said they are sending it to billing to take the changes out ... finally!

[2] Restarted the ID Campaign. Not profitable at the moment since the best offers were capped out.
* Getting ready for when August Caps becoming available
* Using the cheap traffic to test a few Lander Angles/Ideas

***********************************
THIS WEEK
***********************************


[1] Going to Start a new project in a new Geo (have not decided which yet)

[2] Waiting for 'Best' ID Offers to become available again with August caps and then running them (and hopefully scaling it)

[3] Continuing to Test Lander Angles in ID

***********************************
QUESTIONS
***********************************


[1] I use my own lander on pop and I was looking at offers to test for ID last week and I noticed a couple of offers with really different pre-landing pages.
Lets say I run traffic to my own Lander and ...
Lets say Offer 1 has Pre-Landers 1A, 1B and 1C and Offer 2 has Pre-Landers 2A, 2B, 2C and Offer 3 has Pre-Landers 3A, 3B, 3c

To be totally sure, you'd need to test all 9 combinations.
1A, 1B, 1C, 2A, 2B, 2C, 3A, 3B, 3C

Is it better to just go for an approx best by only really run 6 combinations like below
i). Do 1A, 2A, 3A to find best offer first. Let say offer 3 was winner
ii) Do 3A, 3B, 3C

Of course once you get up to 6 different offers with 3 different landers each it gets interesting (18tests versus 9tests) lol

Question: Is the 2nd way the better way?
Question: Do you see much of a difference (i.e. larger than 20%) in sending traffic to different pre-landers for the same offer?

---

[2] Question 2 - At the moment I'm working out the best offers to run based on all traffic from that Geo.
Current - Deciding best offer based off grouped traffic from Traffic Source A (TS A), TS B, TS C etc

If your experience is it worth going more granular?
For example
- Do you find that Traffic Source A might work best with Offer 1 but Traffic Source B works better with Offer 2?
- Do you go even more granular and find best Offers based on each Campaign in each Traffic Source (e.g. Best Offer for PopAds Campaigns at bid Z)
- Or even really granular and find best offer for each very high traffic zoneids at a certain bid (e.g. Zone A on PopAds at bid of 0.5CPM)

---

Well this is embarrassing - have no idea how I missed your updates back in July. I'll blame that on my having to play catch up after AWE - but of course that's not a valid excuse. Profuse apologies for the massive delay in responding!

I just pulled an all-nighter and after the massive delay, the last thing I want to do is to rush through your posts and give a short reply. Give me another day or so to catch up on sleep - will do my best to give you a thorough response.

Thanks so much for your understanding - and especially for not having given up on updating this follow-along!



Amy

---

Hi Amy,
No problem - seriously, I'm really grateful for all the help you've given me in the last few months.

So please take care of yourself (get some sleep!) first - this is not urgent.

Suneel

---

Thanks Suneel for your patience and understanding!

The good news is I turned my sleeping hours around haha! Got up at 4am this morning and got a lot of work done. Will try to keep that schedule.

I'm assuming the different offers are something like different sweepstakes prizes? Like iphone x, samsung galaxy, tesco vouchers etc.?

If that's the case, then each type of "prize" should be tested with multiple offers + multiple landers.

If you only take a single offer + multiple landers, then if that offer turns out to be a dud, you won't get the conversions you need to cut landers down to a winner.

You can certainly test one prize first, then the next prize etc. It's not necessary to test all 3 types of prizes at the same time - although you could. It would depend on how much money you're willing to invest into testing at any one time, and also whether you have time to monitor multiple campaigns. When you have more experience though, it would be good to always have a few test campaigns running - that way you can continually have something promising to optimize and scale.

And can you please clarify the last question regarding the 20% (20% of what?)



For the most part, pop traffic is broad traffic (general audience), and offers that convert on this traffic are broad-intent offers.

Having said that - there are situations where running more granular as you suggested, may make sense.

The big traffic networks broker traffic from one another, and therefore, for practical purposes, we can pretty much assume the audience to be similar.

That, plus the fact that most offers that are suitable for pop are half-shady to fully-shady cap offers that don't have a long lifespan, means that it probably won't be worth it to test offers for every traffic network.

Speed is of essence when it comes to pop. When you see a promising offer+lander, scale out to networks as quickly as possible, quickly cut the ones that you feel would take too much effort to optimize (unless it's your first time running in that geo+network, in which case you'll need to invest into cutting the worst placements before you can accurately gauge potential), quickly optimize the rest to profits, and test more stuff to find more promising combinations - and rinse and repeat.

Speedy scaling of the same offer to more traffic sources will result in more profits than whatever additional profits you can gain by finding the best offer for each traffic source.

As for finding the best offer at each bid - the results would be transient because competition levels are changing constantly, so that the percentage of traffic you can get from different placements at a particular bid will be constantly changing. So you can forget about that.

As for finding the best offer at each high-volume placement/zone: Again, doing all that testing for one zone will likely not make you more money than if you were to dedicate that time towards scaling out to other networks, or even towards testing more offers.

Having said all that - there ARE ways of going granular that may be worth your time. Examples:

1)Finding the best offer for the biggest OSs, carriers, or browsers. While you're testing offers you'll be generating conversions anyways, so there's not much extra effort involved. If you're seeing that a certain offer is converting better for android/ios, then it may be worth it to set up a separate campaign for just that offer and OS.

You can do the same with landers as well.

There are many possible reasons why offers and landers would convert particularly well/badly for a certain OS/carrier/browser. We can attempt to guess the reason, and indeed with more experience you may start to see certain trends, but the most direct way would just be to run traffic and let the stats tell the story.

If this type of granularity is what you're after, try this approach: When testing landers and offers, simply run them past the usual point of statistical significance without cutting landers/offers if possible (perhaps only cutting the ones that are obviously not converting at all or very badly) to collect a good number of conversions, then drill down to the various big traffic segments like this:

OS -> Offers -> Landers
Carrier ->Offers -> Landers
Browser -> Offers -> Landers

....and look to see whether specific offers/landers seem to be converting particularly well for a certain traffic segment (compared to overall performance for the whole campaign). If so, start a campaign with the specific traffic segment - and potentially bid higher to get more traffic.

2)If you can find out the actual site urls for the big placements (there are ways, but you'll need to do more research, because I never bothered to test this), you can seek out offers that you believe would be the best match for the site's visitors. This is similar to your idea above, but instead of blindly testing offers for each placement, which would take too much effort, you're seeking out offers that may be a good fit to each site.

You could even customize your lander for the site by using a similar look-and-feel + a suitable angle - be careful with this though because many site owners may not be too thrilled about this tactic. But doing so can be very good for conversion rate.

Example of this: Selling a VPN service offer on torrent sites.



Hurray!!

Have you tried setting up a CPA campaign on propellerads? Set a target CPA that's lower than your payout to see if you get decent traffic volume.



I haven't run pop camps lately - been focusing on other stuff. Perhaps start a new thread to get feedback?

Here are some relevant threads that may or may not help:

https://stmforum.com/forum/showthrea...-tracking-urls

https://stmforum.com/forum/showthrea...r-to-use-https

https://stmforum.com/forum/showthrea...ome-ssl-update

https://stmforum.com/forum/showthrea...er-with-https-

Hope that helps! Now that I'm all caught up - I look forward to your next update!



Amy

---

Wow - thanks Amy!

Will do some tests based on that information.
Also launched 1 new Pops GEO last week (italy)
Going to launch at least 2 new GEOs this coming week :-)

---

Starting to feel like I'm getting traction.

Learning a heap as I go into new GEOs, but still using ID to run a lot of tests ... and make money!


***********************************
THIS LAST WEEK
*********************************

INDONESIA GEO:

[1] I'm capped out at the moment on the Affiliate Networks I'm running on. Have asked my AMs to increase Caps but not happening.
- So in the process of looking for other Affiliate Networks to Try with decent ID Sweeps (anyone know any decent ones?)

[2] Tests Run

[2a] Retargeting (did not increase ROI)

I used PropellerAds and retargetted people who had been on my prelander and then clicked through to an offer. The traffic from retargetting campaign was really low and after about 2000 conversions the retargetting campaign was only 3% better then the non retargeting campaigns but this was not statistically significant.

[2b] Using Javascript link instead of HTML link to Offer on Prelander (10% better!!!) -

I noticed while running the retargetting campaign that some offers seem to work better with javascript link (with 1 second delay) and not a direct HTML link!
One offer was better with direct HTML Link - 12% better after 900 conversions (95% statistical significance)
Another offer was better with Javascript link with 1 second delay - 10% better after about 1200 conversions (95% statistical significance)

Have no idea why ... but going to test this out on other geos once I get enough volume to make worthwhile.

[3] Tests Running Right Now

(3a) -- Seeing if SSL on the Tracking Link and Landing Page makes a difference overal and just for chrome users.
HTTP Tracking link --> HTTP Lander
HTTP Tracking Link --> HTTPs Lander
HTTPs Tracking Link --> HTTP Lander
HTTPs Tracking Link --> HTTPs Lander

(3b) See if setting up Favicons on the Lander improves ROI significantly


NEW GEOs:
Launched Brazil Last Week :-)


***********************************
PLANS FOR THIS WEEK
*********************************

[1] Try to find more affiliate networks/offer on ID so I can stretch my CAP

[2] Continue Testing on ID

[3] Continue to Optimize New Geos and See If I can make it profitable (Italy, Brazil)

[5] Launch 3 New GEOs (Germany, Austria, Switzerland)

[6] Get Groundwork done to launch at least 3 New Geos next week.

---

Home > POP / PPV / Redirect > Follow-along Campaigns