Home > Mobile > Follow-along Campaigns

Road to become a super affiliate: First mobile campaign (sweeps) (24)

11-26-2015 05:58 AM #1 traveltotokyo (Member)

You don't need to buy an android phone. You download Opera emulator and use a VPN to simulate android phones.

11-26-2015 06:39 AM #2 Mr Green (Administrator)

Hey mate,

To answer your questions;

Whats your opinion about my Voluum setup? splitting the campaigns based on connection type and device type? Can i split my campaign on other categories? It's fine for now.

When do i know an offer is working or not? You should of done this research before launching. Is the offer proven? Did you see the offer being run a lot when spying? Did the network or advertiser mention it is doing volume on pops?

Can i do anything about the click loss? Click loss is part of the game. Make sure your landers load fast. Then just focus on the more important components of the campaign like the angles, targeting, offer, optimisation etc.

Any advice how to go on?

1. Find out if the offer is proven.
2. Cut this angle - (own) Boost your sex appeal - facebook theme
3. Test Ireland on popads

11-26-2015 03:49 PM #3 vortex (Senior Moderator)

Nice detailed start to a great follow-along! And of course thank-you for the kind words about my threads on stat sig. Glad they're helping someone!

Several suggestions:

1)If possible, split-test offers as well in order to minimize chances of having a dud offer. i.e. try to find other iphone 6s offers for the same geo.

2)Your bid is too high at 0.04/pop (you probably thought this was the CPM price - when it's actually the cost per pop). I just looked up SA on the popads inventory chart and the bids are as follows:

Avg Bid = 0.00121
Max Bid = 0.02000
Effective Max Bid = 0.00510

What you can do, is keep your current camp running if you like, but also duplicate your campaign several times and set them to lower bids. I would set bids to 0.00052, 0.00222, and 0.00422. Then I would run traffic for 1-2 days (in your case it may take a few more days because there's not much SA traffic on popads) to see which campaign seems to give me the best conversion rate as well as a healthy dose of traffic, then pause the others and use just that campaign for subsequent lander and offer testing.

While you're testing landers, you may be able to get away without using the spreadsheet calculator, and just keep reviewing the placements with the most spend and cut based on "rules of thumb". You can retest those again later.

Once you have an offer+lander combo that's doing at least close to breaking even, you can choose to unpause those other camps you set up earlier with the other bids, and use the spreadsheet calculator to cut unprofitable placements for each camp. Of course if any of the camps are losing excessive money you may want to just pause it again instead of continue cutting placements. At this time you may also want to retest previously blacklisted placements, as was mentioned above.

2)Splitting camps by Mobile/Wifi & Smartphone/Tablet is one way to go. Doing so will give you more accurate results in terms of which landing pages will work better for what. The downside though is that it will take you a lot longer for every round of lander testing to be finished. However, in your specific case, you're only down to 2 camps now which is manageable.

3)When split-testing landers and deciding which to cut, do not use minimum viable conversion rate. Instead, use the peakconversion split-test calculator - please see part 1 of "my stat guide" which you've kindly referred to already. What you want to do is identify the best lander of the bunch, and use it as the "control" for the next round of testing.

However, since you're only in your first round of testing, and all your landers are vastly different, you may not want to cut all the way down to just one last lander. A better approach would be to use the split-test calculator to cut the ones that obviously aren't promising, until you're left with the ones that show promise (this decision will be a subjective one unfortunately). Say you have 2 landers left over that look like they may be profitable with additional improvements, you would then make variations from both and split-test those. If none of the landers look promising enough, cut down to the last one, and use it as the control for the next round of landers where you'll want to rip and test more landers. For something like sweeps on pop, I would recommend just mass-testing ripped landers until you're semi-close to profit, and then improve the winning lander to give it that final boost.

4)Also try mediahub and zeropark - they have more volume for SA than popads does. However, if you're getting enough traffic from popads for testing (i.e. the testing speed is not too slow for you) then by all means stick with popads until you have a good offer+lander combo, and then scale out to other sources.

Hope that helps!

Amy

11-26-2015 04:02 PM #4 cbrughmans (Member)

IMO becoming a superaffiliate is not only about testing & optimizing but also about scaling up (fast). Many affiliates focus too much on solely the testing & optimization phase and look only at percentages instead of looking at total $$$ value. That being said, all the tips here are fantastic input and feedback for any affiliate, regardless of their size.

11-26-2015 06:32 PM #5 whtang (Member)

Originally Posted by traveltotokyo

You don't need to buy an android phone. You download Opera emulator and use a VPN to simulate android phones.

@traveltotokyo:
Thank you for the tip. i should try this before getting an android phone.

Originally Posted by Mr Green

@Mr Green:
Thank you for your feedback :-)

I ask my AM manager for a top offer list sorted by volume. I pick the offer with the lowest payout and did some research with spy tools and find out that people are running this offer. I will also improve the loading speed by using the awesome tutorial of Caurmen http://stmforum.com/forum/showthread...ight=load+test

Thank you for the advice to cut the angle "(own) Boost your sex appeal - facebook theme". Can you explain to me why? is it because the 3-5x payout without conversion rule?

And i will definitely have a look testing Ireland on PopAds :-)

Originally Posted by vortex

Yes im definitely doing that.

Originally Posted by vortex

2)Your bid is too high at 0.04/pop (you probably thought this was the CPM price - when it's actually the cost per pop). I just looked up SA on the popads inventory chart and the bids are as follows:

Avg Bid = 0.00121
Max Bid = 0.02000
Effective Max Bid = 0.00510

What you can do, is keep your current camp running if you like, but also duplicate your campaign several times and set them to lower bids. I would set bids to 0.00052, 0.00222, and 0.00422. Then I would run traffic for 1-2 days (in your case it may take a few more days because there's not much SA traffic on popads) to see which campaign seems to give me the best conversion rate as well as a healthy dose of traffic, then pause the others and use just that campaign for subsequent lander and offer testing.

While you're testing landers, you may be able to get away without using the spreadsheet calculator, and just keep reviewing the placements with the most spend and cut based on "rules of thumb". You can retest those again later.

Once you have an offer+lander combo that's doing at least close to breaking even, you can choose to unpause those other camps you set up earlier with the other bids, and use the spreadsheet calculator to cut unprofitable placements for each camp. Of course if any of the camps are losing excessive money you may want to just pause it again instead of continue cutting placements. At this time you may also want to retest previously blacklisted placements, as was mentioned above.

This is pure GOLD. Im going to apply this the next test run. But why these numbers: 0.00052, 0.00222, and 0.00422 ? Are those just random numbers?

PopAds will give you an estimate how many impressions you get. And those number wont give my any impressions. But lets find out.

Originally Posted by vortex

2)Splitting camps by Mobile/Wifi & Smartphone/Tablet is one way to go. Doing so will give you more accurate results in terms of which landing pages will work better for what. The downside though is that it will take you a lot longer for every round of lander testing to be finished. However, in your specific case, you're only down to 2 camps now which is manageable.

Yes thats the trade off. im trying different things to find out what works.

Originally Posted by vortex

3)When split-testing landers and deciding which to cut, do not use minimum viable conversion rate. Instead, use the peakconversion split-test calculator - please see part 1 of "my stat guide" which you've kindly referred to already.

What you want to do is identify the best lander of the bunch, and use it as the "control" for the next round of testing.

However, since you're only in your first round of testing, and all your landers are vastly different, you may not want to cut all the way down to just one last lander. A better approach would be to use the split-test calculator to cut the ones that obviously aren't promising, until you're left with the ones that show promise (this decision will be a subjective one unfortunately). Say you have 2 landers left over that look like they may be profitable with additional improvements, you would then make variations from both and split-test those. If none of the landers look promising enough, cut down to the last one, and use it as the control for the next round of landers where you'll want to rip and test more landers. For something like sweeps on pop, I would recommend just mass-testing ripped landers until you're semi-close to profit, and then improve the winning lander to give it that final boost.

Another great advice :-).

Can you explain why i shouldn't use the minimum viable conversion rate?

Im going to follow your advice using "Part 1" and using a split test calculator.

Originally Posted by vortex

4)Also try mediahub and zeropark - they have more volume for SA than popads does. However, if you're getting enough traffic from popads for testing (i.e. the testing speed is not too slow for you) then by all means stick with popads until you have a good offer+lander combo, and then scale out to other sources.

Hope that helps!

Amy

Totally awesome. Thank you for all the advice.

Originally Posted by cbrughmans

Absolutely true and will definitely keep that in mind

11-27-2015 10:37 AM #6 cbrughmans (Member)

The clickloss is normal as its less than 10%. Different systems track different stats.

I think it's going to be very, very hard to get this profitable. I would cut my losses and focus on the next offer.

11-27-2015 11:52 AM #7 whtang (Member)

Originally Posted by cbrughmans

Yeah you re probably right. I should stop this campaign and find other offers. Im currently running this campaign with the bid strategy described by Amy. Lets see what will happen :-)

11-28-2015 10:53 AM #8 whtang (Member)

*update 20151127
I paused all my campaigns and need to evaluate. Its problably best to cut the losses and move on to the next offer.

11-29-2015 10:58 PM #9 vortex (Senior Moderator)

But why these numbers: 0.00052, 0.00222, and 0.00422 ? Are those just random numbers?

PopAds will give you an estimate how many impressions you get. And those number wont give my any impressions. But lets find out.

Those bids are kind of random but not exactly. I usually include the average bid, then bid as low as I can but still get good traffic, then I'd also bid a little above the maximum bid. Other than these 3 "standard" bids, I also like to bid between the lowest and average bids, and between the max and average bids. That's a total of 5 bids.

Although popads' estimator isn't accurate, you'd usually want to aim for a good level of traffic even for the lowest bid. In your case, if my suggested bid amounts aren't resulting in healthy traffic volume shown in the estimator, please adjust upwards accordingly.

(As for the 0.xxxx2 - I like to bid a bit above the usual "round" values like 0.001, and many people have already caught onto that trick by bidding 0.0011, so I like to end my bids with a trailing 2.

)

Can you explain why i shouldn't use the minimum viable conversion rate?

Im going to follow your advice using "Part 1" and using a split test calculator.

The minimum viable conversion rate is used in calculations where you decide to cut/whitelist anything that will be profitable/unprofitable in the long run. This is normally applied to banners or placements - stuff that you can include more than one of (i.e. you want to run multiple banners that are profitable; same goes for placements; you don't want to cut these down to the best one).

The split-test calculator is used to identify the best candidate(s) of a bunch of stuff. This is usually used for offers and landers - stuff you usually want to run the best ONE of. (Having said that, it's sometimes good to keep multiple offers and landers running even after a campaign is in profit, typically to reduce burnout, and to detect lead shaving/scrubbing. But you get the idea.)

The Daily budget for campaign “sa/cellular/smartphones” is used around 4-5 PM

You may want to do this the other way around: Start your campaign around 5PM (local time in your target geo) because for many offers, conversion rates are usually better in the evenings when people have more time to spend on their cell phones.

Whats your opinion about the above stats? the ROI doesnt look good after spending $75. What to do with this information?

Sweeps is a very big vertical with lots of offers in many geos. It will take more than one attempt to crack this, but when (if) you do, the rewards can be huge. It will require testing multiple offers and landing pages, cutting based on statistical significance, and several rounds of offer and lander testing to see the potential of a specific geo on a specific traffic source.

Actually, the stats you have aren't terrible for a start! What you need to do is pick a geo that has more traffic (or like I've suggested, pick a traffic source that has more traffic for your target geo), rip at least 5-10 landers you see through spying (or do this from spy tools - I recommend Adplexity or AdsXposed) that are very different, compile a list of similar offers for your geo, running enough traffic to reach statistical significance for both your offers and your landers, then see how close you are to breaking even, and go from there. Pop traffic has become more than a bit saturated from the influx of newbies in recent months, so if you want to make profits, you'll need to be prepared to spend money on testing, period.

Any time you split-test something and not run the test to statistical significance, the data you've collected is rendered useless and you've thrown money away. This may sound harsh, but STM member after STM member has made this common mistake (including myself many times), some over and over, before they realized how important it is to run tests to statistical significance. Just earlier tonight I was chatting with a friend on skype - an intelligent and persistent affiliate - and he tells me he's finally starting to realize that the biggest reason why he still hasn't hit on profits after several years and $40k+ spend, is because he's always stopping/changing his testing prematurely.

To give you an idea of what a sample optimization process may look like:

-Test 2 offers and 10 ripped landers. Offer1 and Lander6 win. ROI is now at -50%.
-Test 3 more offers along with Offer1 (the test control), using just Lander6. Offer4 wins. ROI is now at -20%.
-Test 10 variations of Lander6 along with the original Lander6, using just Offer4. Lander6e wins. ROI is now at 0%.
-Use the calculator spreadsheet ("Part 2" of my stats series) to cut anything that doesn't meet 30% minimum ROI - placements, devices, browsers, carriers, whatever.

You can also start cutting placements/devices/browser/carriers/etc. before the last stage if you want to save on test budget. Just keep in mind that once you have a better offer and lander, you may want to retest some of the stuff you cut before, as they may be profitable now.

That's it in a nutshell! Good luck with your next campaign! Even the pros need to test multiple campaigns (sometimes 10 or more) before finding a profitable one, so be sure to test extensively, and once you find something profitable, scale it as big as fast as possible.

Amy

11-30-2015 09:05 AM #10 whtang (Member)

Originally Posted by vortex

)

The minimum viable conversion rate is used in calculations where you decide to cut/whitelist anything that will be profitable/unprofitable in the long run. This is normally applied to banners or placements - stuff that you can include more than one of (i.e. you want to run multiple banners that are profitable; same goes for placements; you don't want to cut these down to the best one).

The split-test calculator is used to identify the best candidate(s) of a bunch of stuff. This is usually used for offers and landers - stuff you usually want to run the best ONE of. (Having said that, it's sometimes good to keep multiple offers and landers running even after a campaign is in profit, typically to reduce burnout, and to detect lead shaving/scrubbing. But you get the idea.)

You may want to do this the other way around: Start your campaign around 5PM (local time in your target geo) because for many offers, conversion rates are usually better in the evenings when people have more time to spend on their cell phones.

Sweeps is a very big vertical with lots of offers in many geos. It will take more than one attempt to crack this, but when (if) you do, the rewards can be huge. It will require testing multiple offers and landing pages, cutting based on statistical significance, and several rounds of offer and lander testing to see the potential of a specific geo on a specific traffic source.

Actually, the stats you have aren't terrible for a start! What you need to do is pick a geo that has more traffic (or like I've suggested, pick a traffic source that has more traffic for your target geo), rip at least 5-10 landers you see through spying (or do this from spy tools - I recommend Adplexity or AdsXposed) that are very different, compile a list of similar offers for your geo, running enough traffic to reach statistical significance for both your offers and your landers, then see how close you are to breaking even, and go from there. Pop traffic has become more than a bit saturated from the influx of newbies in recent months, so if you want to make profits, you'll need to be prepared to spend money on testing, period.

Any time you split-test something and not run the test to statistical significance, the data you've collected is rendered useless and you've thrown money away. This may sound harsh, but STM member after STM member has made this common mistake (including myself many times), some over and over, before they realized how important it is to run tests to statistical significance. Just earlier tonight I was chatting with a friend on skype - an intelligent and persistent affiliate - and he tells me he's finally starting to realize that the biggest reason why he still hasn't hit on profits after several years and $40k+ spend, is because he's always stopping/changing his testing prematurely.

To give you an idea of what a sample optimization process may look like:

-Test 2 offers and 10 ripped landers. Offer1 and Lander6 win. ROI is now at -50%.
-Test 3 more offers along with Offer1 (the test control), using just Lander6. Offer4 wins. ROI is now at -20%.
-Test 10 variations of Lander6 along with the original Lander6, using just Offer4. Lander6e wins. ROI is now at 0%.
-Use the calculator spreadsheet ("Part 2" of my stats series) to cut anything that doesn't meet 30% minimum ROI - placements, devices, browsers, carriers, whatever.

You can also start cutting placements/devices/browser/carriers/etc. before the last stage if you want to save on test budget. Just keep in mind that once you have a better offer and lander, you may want to retest some of the stuff you cut before, as they may be profitable now.

That's it in a nutshell! Good luck with your next campaign! Even the pros need to test multiple campaigns (sometimes 10 or more) before finding a profitable one, so be sure to test extensively, and once you find something profitable, scale it as big as fast as possible.

Amy

Wow Amy. This is gold. Thank you for this amazing feedback.

Im going to follow your advice
- To find other traffic sources that has traffic for the GEO im targeting.
- To test multiple offers (from different aff networks) from different GEO

Basically i need to repriortize my testing order. (Traffic / Order > Angle > Landing page type > landing page components > Banners)

Again thank you for your input.

01-18-2016 06:50 PM #11 matuloo (Legendary Moderator)

Hello,

I dont know what kinda offer you picked but NL is usually one of the most expensive geos, which means high payout offers and a big testing budget. Why did you choose this particular GEO?

I see you are targeting smartphones and tablets in the same campaign, you should split them into 2 campaigns. First of all tablets usually convert differently from smartphones and also the bids usually reflect this.

Now to your questions :

1. you should test as many sources as you can handle. Every source MUST go into a different campaign, so it cant skew your data. Treat every source as a standalone campaign.
2. this depends on where your problem is - is it the angle or the offer? If its the offer, better angle will not work. If its the angle, then coming up with a better one is exactly what you need.
3. when starting out, its kinda the only way to go to be honest. I know I said in question 1 that you should test as many sources as possible, but ONLY when you can handle it. Start with limited amount of campaigns and increase as you feel fits your knowledge level. As for testing new things, you need to stick to one change at a time. If the change is too big, you should consider starting a new campaign.
4. dont give up and dont waste time by sticking to something that cant work. If the traffic source really sucks, move to another. If the offer simply wont convert, move to another or change verticals.

Good luck

01-18-2016 07:08 PM #12 whtang (Member)

Originally Posted by matuloo

and dont waste time by sticking to something that cant work. If the traffic source really sucks, move to another. If the offer simply wont convert, move to another or change verticals.

Good luck

Hi Matuloo,

Thnx for the response :-)

im aware that NL is one of the most expensive GEO. The reason i picked this is:
1) i speak the language
2) i was trying to find similar offer from different networks
3) find out that these offers are runned by multiple affiliates

In my first campaign i did separate devices in different campaigns. this time i use custom flow which separate devices and connection type. when i find out that for example tablets arent profitable then i can turn off that part in Voluum. I what to find out which way works best for me.

my responses to your answers of my questions:
1) i will try other traffic sources separated in different campaigns
2) My testing order is traffic > offer > angles > landers type > landers components > others
3) Thanks for the advice
4) I wont give up. thats part of life :-)

01-18-2016 07:50 PM #13 matuloo (Legendary Moderator)

Originally Posted by whtang

Oki dokie

So when youre sticking to NL, try BE too, you know the language is the same there in a significant part of the country, right?

Custom flow in Voluum wont help you, since you have the same bid set in the traffic source which means you are most likely overpaying for either smartphones or tablets.

01-19-2016 02:59 AM #14 simon_89 (Member)

Originally Posted by whtang

In my first campaign i did separate devices in different campaigns. this time i use custom flow which separate devices and connection type. when i find out that for example tablets arent profitable then i can turn off that part in Voluum. I what to find out which way works best for me.

I think what Matuloo is trying to refer to is that if you're creating separate campaigns targeting different devices then there's no need to start a custom flow. However, if you're doing one single campaign and targeting multiple devices then a custom flow would make more sense as you'd probably tailor your devices to certain landing pages and/or banners.

It is important to divide up your campaigns into different campaigns if you're targeting device types because each device has a different bid range. For example, if in your one campaign that you're putting all the devices in and you're bidding .0080 in that campaign when tablet traffic highest bid is .0030, you're paying .0050 more for tablet traffic. But, if you were to split tablet traffic within its own campaign then you wouldn't lose that much money. But, I think splitting campaigns in terms of devices is too much, instead I'd suggest splitting them into desktop/mobile. At the end of the day, it depends on the traffic source you're bidding on as well. Some traffic sources go by the method I just mentioned. Let me know if you need more clarification.

01-19-2016 09:59 AM #15 matuloo (Legendary Moderator)

Originally Posted by simon_89

It is important to divide up your campaigns into different campaigns if you're targeting device types because each device has a different bid range. For example, if in your one campaign that you're putting all the devices in and you're bidding .0080 in that campaign when tablet traffic highest bid is .0030, you're paying .0050 more for tablet traffic. But, if you were to split tablet traffic within its own campaign then you wouldn't lose that much money.

This is what I meant by overpaying and thats the reason why you should split campaigns based on device type. Unless of course you are using a traffic network that uses smart bidding = adjusting the actual bid charged, to only a fraction above the closest competitor.

So yes, in straight bidding networks, you should create separate campaigns based on device type and for easy optimization, also create separate campaigns for them in Voluum, otherwise you are about to face some serious mess

01-20-2016 05:36 PM #16 whtang (Member)

Originally Posted by matuloo

Im not sure you really need to bid at that many levels. It creates too many campaigns and splits your budget between them. I think a high and low bid should be more than enough.

1 - I dont use Voluum flows all that much, Im using paths and add rules where necessary. It means more campaigns in Voluum, but I got used to it and to me its easier to follow.
2 - you gotta accept the conditions of any traffic source, it either works regardless or not. In some cases all you can optimize is the zone. Good news is, that it should be the same for everyone and the prices should reflect it.
3 - depends on the particular situation. When optimizing zones for example, some are so shitty that you will see it after a few hits - when they dont click thru the LP for example, in some cases you have to wait longer. If I spend about 2-3 times the minimum, maybe 5 with low payout offers and dont see a conversion, I start to look for possible reasons. Im talking about the whole campaign, not a single zone!
4 - facebook is more regulated and the barrier of entry is higher because of that, if you can make it work, there is a TON of volume, only adwords can beat them. So yes, if you can run on FB, definitely do it.

Thanks for the response.

What do you consider low payout offers? everything below $3-$5? in my case. The pay out of the 3 offers are $2,20, $2,25 and $2,25 so should i consider looking to other offers? although i havent that many clicks. maybe i should run more traffics for 1-2 more days.

01-20-2016 07:03 PM #17 matuloo (Legendary Moderator)

Originally Posted by whtang

By low payout offers I mean those around $1-2 or lower, the lower the payout the more tolerant I am towards the X-times spend rules

You changed the bidding strategy now, and as you said you didnt blacklist anything too. So maybe run for some more time and then look at the data again.

01-21-2016 07:53 PM #18 whtang (Member)

Originally Posted by matuloo

By low payout offers I mean those around $1-2 or lower, the lower the payout the more tolerant I am towards the X-times spend rules

You changed the bidding strategy now, and as you said you didnt blacklist anything too. So maybe run for some more time and then look at the data again.

Thnx for the explanation :-). I use Amy's excellent whitelist kill placement sheets for blacklisting.

03-29-2016 09:07 PM #19 whtang (Member)

*** update 20160329 ***
- i think i need to adjust my bids for the VN campaign. The amount of traffic i got is pretty low. Maybe its because im targeting for the following carriers MobiFone, Vietnam Mobile, Viettel.
- Not sure its worth the price but im going to research the bids again
- maybe i should finding offers with a broader targeting restrictions
- maybe i should i should remove some landers and test those later.

03-30-2016 08:58 PM #20 vortex (Senior Moderator)

- i think i need to adjust my bids for the VN campaign. The amount of traffic i got is pretty low. Maybe its because im targeting for the following carriers MobiFone, Vietnam Mobile, Viettel.

I've looked up the volumes for the carriers you listed...

MobiFone:
ISP Details
ID: 796618
Related to ISP: 1216
Name: CMC Telecom Infrastructure Company
Organization: Mobifone Global JSC
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 46,366 impressions

Vietnamobile (Under HT Mobile):
ISP Details
ID: 7296
Name: HT Mobile Communication Center
Organization: Ha Noi Telecom Mobile Center
Country: VNVietnam
Connection Type: Cellular (3)
Connection Speed: Cellular/Carrier (4)
Approx. daily traffic: 2,670 impressions

Viettel:
ID: 175
Name: Viettel Corporation
Organization: Viettel Corporation
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 1,095,363 impressions

Vinaphone (under VNPT):
ID: 166
Name: VDC
Organization: Vietnam Posts and Telecommunications (VNPT)
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 1,383,517 impressions

So:

-Vietnamobile's volume really isn't worth going after.

-MobiFone does have SOME volume, but do know that around 60% of that will be WIFI traffic, and only around 40% will be carrier traffic. This is true even if you were to set "connection types" to "cellular" and "connection speeds" to "cellular/carrier". AFAIK there's no way to get MobiFone carrier traffic only.

-Viettel has a lot of volume, but again, I don't know of a way to get JUST carrier traffic. Based on my stats, I'm getting 80% wifi and 20% carrier.

-Vinaphone is under VNPT, and if you target VNPT you'll get around 40% of Vinaphone carrier traffic and 60% of VNPT wifi traffic.

So how is any of that relevant to us? Since it's apparent that most of this traffic is WIFI, and that there's no way to just get carrier traffic, we need to find an offer that allows AND converts well for wifi traffic!

Yup - I put that in a bigger font, because it's THAT important!

Now go test more offers!

Amy

P.S. In case you're wondering how I found out all that info on the relationships of carriers and isp providers and what you can and cannot target on popads with regards to ISPs, this is what I did:

-Did a search in google to find out some info on which carriers there are in VN and how they're related to ISPs:

https://en.wikipedia.org/wiki/Teleco...ons_in_Vietnam

Viettel Mobile (directly under Viettel Corp.)
MobiFone (directly under VMS)
VinaPhone (directly under VNPT)
S-Fone (directly under CDMA S-Telecom)
Vietnamobile (ex-HT Mobile)
Beeline (directly under G-Tel)

-I then set up 2 test camps on popads, one with "connection types" set to "cellular" and "connection speeds" set to "cellular/carrier", the other camp with "connection types" set to everything EXCEPT "cellular" and "connection speeds" set to everything except "cellular/carrier". Ran a little bit of money to each, then went into Voluum stats and drilled down to ISPID -> mobile carriers. That way I could find out which carriers I can get JUST carrier traffic for, and which carriers I can't (i.e. when I target certain carriers I need to be prepared to get wifi traffic along with it as well).

This kind of detective work is always a good step when running a new geo on popads, especially for a big geo like VN.

03-30-2016 09:17 PM #21 whtang (Member)

Originally Posted by vortex

Wow - must have been a lot of work posting those screenshots! I've looked over them, but unfortunately because you only had a few conversions and most were for YTZ, there's not enough data to go on. I feel really bad that you went to all the trouble and then have me come back to tell you we can't use the data. We'll need to take another look when you have enough conversions for your main offers.

I've looked up the volumes for the carriers you listed...

MobiFone:
ISP Details
ID: 796618
Related to ISP: 1216
Name: CMC Telecom Infrastructure Company
Organization: Mobifone Global JSC
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 46,366 impressions

Vietnamobile (Under HT Mobile):
ISP Details
ID: 7296
Name: HT Mobile Communication Center
Organization: Ha Noi Telecom Mobile Center
Country: VNVietnam
Connection Type: Cellular (3)
Connection Speed: Cellular/Carrier (4)
Approx. daily traffic: 2,670 impressions

Viettel:
ID: 175
Name: Viettel Corporation
Organization: Viettel Corporation
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 1,095,363 impressions

Vinaphone (under VNPT):
ID: 166
Name: VDC
Organization: Vietnam Posts and Telecommunications (VNPT)
Country: VNVietnam
Connection Type: Residential (10)
Connection Speed: Cable/DSL (2)
Approx. daily traffic: 1,383,517 impressions

So:

-Vietnamobile's volume really isn't worth going after.

-MobiFone does have SOME volume, but do know that around 60% of that will be WIFI traffic, and only around 40% will be carrier traffic. This is true even if you were to set "connection types" to "cellular" and "connection speeds" to "cellular/carrier". AFAIK there's no way to get MobiFone carrier traffic only.

-Viettel has a lot of volume, but again, I don't know of a way to get JUST carrier traffic. Based on my stats, I'm getting 80% wifi and 20% carrier.

-Vinaphone is under VNPT, and if you target VNPT you'll get around 40% of Vinaphone carrier traffic and 60% of VNPT wifi traffic.

So how is any of that relevant to us? Since it's apparent that most of this traffic is WIFI, and that there's no way to just get carrier traffic, we need to find an offer that allows AND converts well for wifi traffic!

Yup - I put that in a bigger font, because it's THAT important!

Now go test more offers!

Wow another gold tip. thank you Amy.

One question: How do you know the approx daily traffic? are those your daily stats or did you get those from pop ads advance inventory?

*** update 20160330 ***
- Spend: $15,97
- ROI: -100%
- impression: 4721
- tracked clicks: 636
- traffic source clicks: 235
- landing page ctr %: 2,706% (not sure if calculated correctly: formula is tracked clicks divided by traffic source clicks (636 / 235)
- Cut 4 landers and 1 offer based on stats from march 21th . i cut those landers based on data across all campaigns
- apparently there are some carriers issue. didnt get any notice from a couple of affliate networks so how do you deal with this?
- change some targeting settings based on Amy's post. i also include wifi and what to see if the traffic improve. I still have to separate carrier traffic from wifi traffic (going to do it tomorrow)

03-30-2016 11:00 PM #22 vortex (Senior Moderator)

setup question
Is it better to create different popads campaigns with different bids and targeting setup and just a couple of Voluum campaigns separated by device type and connection type? In this case
Popads
- campaign with avg bid targeting smartphones with carrier traffic
- campaign with high bid targeting smartphones with carrier traffic
- campaign with avg bid targeting tablets with carrier traffic
- campaign with high bid targeting tablets with carrier traffic

Voluum
- campaign with avg bid targeting smartphones / tablets with carrier traffic
- campaign with high bid targeting smartphones / tablets with carrier traffic

Determine how many landers / offers to test at once question
How many landers / offers should you test at once? testing too many landers / offers at once have some impact on budget and amount of time to get enough data. is there a sweetspot / guidance / best practices / rule this? How do you guys do this?

Why would you set up 2 voluum campaigns and use them to set up 4 popads campaigns? I'm not understanding how that would work. Why not just set up 4 voluum camps and 4 popads camps?

You can for sure get really granular with the way you test, and I'm all for getting granular. Just keep in mind though that the more campaigns you have, the more data you'll need to reach statistical significance. It's the whole accuracy vs. efficiency thing I've discussed with various members on various threads in the past. The more granular you get, the more accurate your results will be (for example you can test offers for Android+Carrier1+Tablets+Browser1), but the more time and test budget you will need to spend.

Everyone has a different style of testing. Personally I would sometimes set up camps for each major OS+Carrier, then cut stuff like tablets and browsers only if they convert very badly or after I've optimized my funnel.

Amy

03-30-2016 11:10 PM #23 vortex (Senior Moderator)

One question: How do you know the approx daily traffic? are those your daily stats or did you get those from pop ads advance inventory?

This tool will let you look up info on any ISPID:

https://www.popads.net/traffics/token_details

*** update 20160330 ***
- Spend: $15,97
- ROI: -100%
- impression: 4721
- tracked clicks: 636
- traffic source clicks: 235
- landing page ctr %: 2,706% (not sure if calculated correctly: formula is tracked clicks divided by traffic source clicks (636 / 235)
- Cut 4 landers and 1 offer based on stats from march 21th . i cut those landers based on data across all campaigns
- apparently there are some carriers issue. didnt get any notice from a couple of affliate networks so how do you deal with this?
- change some targeting settings based on Amy's post. i also include wifi and what to see if the traffic improve. I still have to separate carrier traffic from wifi traffic (going to do it tomorrow)

$16 spend without a single conversion? Something must not be right. You need to start testing more offers to find something that will convert well for wifi traffic.

Carriers issue and not gotten any notice from affiliate networks? What's that about? Please provide more details.

There's no way to separate carrier and wifi traffic for most carriers for VN on popads, so don't even try. This is exactly why you need to find offers that convert well for wifi - because there's no way to isolate the carrier traffic.

Amy

03-31-2016 08:59 PM #24 whtang (Member)

Originally Posted by vortex

Why would you set up 2 Voluum campaigns and use them to set up 4 popads campaigns? I'm not understanding how that would work. Why not just set up 4 Voluum camps and 4 popads camps?

You can for sure get really granular with the way you test, and I'm all for getting granular. Just keep in mind though that the more campaigns you have, the more data you'll need to reach statistical significance. It's the whole accuracy vs. efficiency thing I've discussed with various members on various threads in the past. The more granular you get, the more accurate your results will be (for example you can test offers for Android+Carrier1+Tablets+Browser1), but the more time and test budget you will need to spend.

Everyone has a different style of testing. Personally I would sometimes set up camps for each major OS+Carrier, then cut stuff like tablets and browsers only if they convert very badly or after I've optimized my funnel.

Amy

You are right. i should separate thinks when possible. im also thinking way too hard :-(.

Originally Posted by vortex

This tool will let you look up info on any ISPID:

https://www.popads.net/traffics/token_details

$16 spend without a single conversion? Something must not be right. You need to start testing more offers to find something that will convert well for wifi traffic.

Carriers issue and not gotten any notice from affiliate networks? What's that about? Please provide more details.

There's no way to separate carrier and wifi traffic for most carriers for VN on popads, so don't even try. This is exactly why you need to find offers that convert well for wifi - because there's no way to isolate the carrier traffic.

Amy

I think the issue are my campaigns Voluum rule setup. In my stats i see “WI-FI” as Mobile carrier. All those traffic are redirect to my global offers because i didn’t set this up in Voluum. Actually you can’t set up “WI-FI” as a Mobile carrier. So i havent truly tested my offers. The only thing i can do is removing the mobile carrier rules in Voluum for all my campaigns since the offer restrictions applies to all the offer im currently testing.

and thanks for the advice to find offer that converting well for WI-FI. im working on it :-)

I have to change my ways of thinking and doing because i havent achieve anything yet :-(. I know its isn’t easy so i will move on :-)

Some questions:
Cutting campaigns:
1) Is there a way to figure out what conversion rate i need to back out? How do you calculate it when you only know the payout and the possible traffic cost? Could you give an example?

2) how do to figure out how large my sample size must be? of course the required sample size is dependent on the conversion rate you need to back out. Could you give an example?

Campaigns setup:
3) its quite interesting that you setup your camps for each major OS+Carrier. what about device type and connection type (thats what im doing right now)? What i learned form different threads is to separate campaigns for variables with highest impact.

Bids:
4) in previous post you mentioned that you don’t change your bids that often. could you tell me when you change your bids? like when you noticing decrease of traffic and scaling?

Again i really appreciate your help and really learn a lot :-)

Home > Mobile > Follow-along Campaigns