Home > Mobile > Follow-along Campaigns

---

Young Dentist Drilling Into Pops, "Filling" His Pockets With Gold (21)

---

Hello amazing people,

First of all, let me introduce myself. My name is Lucas and you can read about me here! As the title suggests, I'm currently studying Dentistry at uni, but I seem to have too much time on my hands to be able to pursue AM

I've been active for a couple of months now, reading as much as I can and trying out new things. I ran my first campaigns at Go2Mobi (app installs, content offers, AV pin submits) using the Appetizer method. Unfortunately, these weren't very successful and my fear of losing money got the better of me and I decided to move towards Pops, as they have one less creative to worry about. Pops have been treating me better, but currently I'm a bit stuck, which is where I'm asking for some extra help

Hosting: Vultr.com VPS with CDN (Cloudflare)
Affiliate Networks: Adsimilis, F5, Aragon Advertising, MaxBounty, Advidi, SongoCPA, PeerFly
Tracking Software: Voluum
SpyTool: Adplexity

I want to thank @caurmen, @vortex and @matuloo specifically for their detailed guides, tutorials, and input in follow-alongs. You've been a great help so far and I hope that you can also be of great help in my little follow-along!

--------------------------------------------------------------------------------------------------

So on to the more interesting stuff: the campaign!

Сampaign info:
Source: PopAds
GEO: VN
Vertical: Sweeps
Payout: $0.48

I’ve been running this campaign for 4 days now, so I’ll talk you through my progress thus far.

Day 1 (27-07-2016)
I set this campaign up at PopAds setting the frequency cap at 1/24h, quality to top 60% of websites (5+), PrimeSpot only and popunder (required). I targeted Android and several carriers. I excluded 100 websites according to this blacklist. I setup this campaign using the staggered bids approach. I setup three campaigns:
high: 0,005xx
mid: 0,003xx
low: 0,001xx
The budget was 15x offer payout (15x0,48 = $7.20) per campaign.
I ripped three landers using Adplexity, fixed them up and optimized them as much as I could.
I also setup time-targeting from 10:00-24:00 local time.


Source: PopAds
GEO: VN
Conversions: 23
Cost: ±$20.00
Rev: $11.04
Profit: -8.96

These are the stats for this first day. The low-bid campaign was closest to breakeven, so I continued with that one. Other stats didn't show anything significant.

Day 2 (27-07-2016)

Source: PopAds
GEO: VN
Conversions: 13
Cost: ±$15.00
Rev: $6.24
Profit: -8.76

This day I continued only the low-bid campaign. ROI is worse than the day before, which was a bummer to see. One lander was lagging behind so I cut it after reaching statistical significance according to the method described here.

Day 3 (28-07-2016)
I setup three new campaigns to see if i could improve ROI by changing bids. The average bid (0,0006x) in the PopAds traffic estimator seemed to be much lower than when I initially setup my camps, so i decided to go with
high: 0,001xx (the low-bid of 27-07-2016)
mid: 0,0009x
low: 0,0005x

Question: The bids seem to be changing all the time, depending on the time of day that I check the traffic estimator. What is the logic behind this and how can I use this information to set the bids? Do I need to change bids all day, or should I keep them at a set point, even though my position in the bidding cue goes lower and lower?

The budget was again 15x offer payout (15x0,48 = $7.20) per campaign.


Source: PopAds
GEO: VN
Conversions: 48
Cost: $47.95
Rev: $23.04
Profit: -24.91

First my high bid-camp was depleted, then my mid bid-camp was empty, and my low bid received almost no traffic. I soon realized that overnight the traffic prices had exponentially increased, so that my high-bid was suddenly below average. I decided to pause my low-bid campaign before the budget was spent and continued with the other camps. When the budgets were spent the high bid-camp had the best ROI and most conversions, so I continued running the high-bid with extra budget. I also cut one lander because it performed worse than the last remaining lander (96%vs4%).


At the end of the day, I also cut carriers 2/3 (same targeting setting in PopAds, but different name in Voluum), because it performed worse than the other. In hindsight, the decision wasn't made on statistical significance. I'm still struggling with the fear of losing money, I guess that's why I cut the carrier earlier than I should have.

When drilling down into placements I found 1 placement that I could whitelist (according to this method here), so I made a new campaign that targeted just that placement. This placement still targets the carrier that I cut for the main campaign.

Day 4 (29-07-2016)
I optimized the lander that was still in rotation by changing all the images to sprites and cleaning up the code and by setting up a small test to see if there was a difference between hosting the external libraries (e.g. jQuery) on cdnjs.com or locally. Loading speed is about 0.1-0.2s with about 100kb page size.


Source: PopAds
GEO: VN
Conversions: 84
Cost: $48.90
Rev: $40.32
Profit: -8.58
My test showed that there is little to no difference between hosting locally and using the libraries hosted on cdnjs.com.

This day I could blacklist several placements and found that the whitelist campaign was a bit green. I already cut down to one lander, one carrier, and one OS, so I didn't know how to proceed from this point. I decided to keep pushing it to reach the network's payout-minimum before the deadline (31-07-2016) and blacklist wherever I could.
I also decided to try out a comparable offer from the same advertiser at the same network with a different prize. I copied and changed my landing page and setup a flow, splitting the traffic 50/50 to the two offers.

Day 5 (30-07-2016)

Source: PopAds
GEO: VN
Conversions: 39
Cost: ±$30.00
Rev: $18.72
Profit: -11.28
The new offer performed worse that the original offer, so I decided to cut it again early in the process. The whitelist campaign seems to be holding its profitability up.

Question: what can I do to make the whitelist campaign more profitable?

Question: I don't know how I can use the information I have gathered so far about the main campaign to optimize into profitability. I have already cut down to one lander, an offer that is converting decently and one carrier. What can I do now?

Any help would be highly appreciated! If you have any questions to me, please don't hesitate to ask!
I'm open to networking and an interesting mastermind with like-minded newbies Hit me up on skype: lucschep

Thanks,

Lucas

---

will be following along!
good work so far!

---

Nice & detailed follow-along, I see a few things you could switch up:

In VN all sites with quality above 3 have done OK, so you can try 3+ instead of 5+...
And about VN - they get up at 5 in the morning and go to bed way earlier so your dayparting is a bit off

Your bids are pretty low, it was possible to go break-even/slightly profitable @ VN on PopAds when paying $3 CPM earlier this year running an iPhone 6 offer allowing all carriers, so $2 CPM or $1.5 is worth a try Since PopAds has smart CPM try to bid higher, rarely will you actually pay the actual CPM that you bid - and identify all totally bad placements from your current data and apply that to your high bid campaign. Build a blacklist at low bid then try to up the bid and keep growing the blacklist.

---

You mean vietnamese local time right?

---

Thanks for the responses so far guys! I've implemented Yseke's suggestions regarding the timezones and tried to switch up the bids. I guess I didn't change them high enough, since the CPV didn't get close to my max bid (see Monday 01-08-2016) Did I get outbid here by someone in the same bidding-range?

Sunday 31-07-2016


Source: PopAds
GEO: VN
Conversions: 22
Cost: ±$7.54
Rev: $10.56
Profit: 3.02
Today I had my first day profit!!! I only ran the whitelist campaign though. This turned out to churn out a nice little profit (after cost-correction it was only a couple of bucks, but still ). Still the question remains: how can I get more profitable on this whitelist campaign? Do I need to up my bids to get more traffic? How high should i put my bid? Current bid is 0,0016x. I'm afraid that putting my bid too high makes this a money-losing campaign as well.

Monday 01-08-2016

Source: PopAds
GEO: VN
Conversions: 76
Cost: ±$45.00
Rev: $36.48
Profit: -8.52

Today I started the main campaign again, after blacklisting some placements. I also calculated with Amy's Tweak-bid calculator that if i set a lower bid for this specific placement it could become profitable (Campaign: Whitelist (xxxxxx)). It was amazing to see that this actually happened! Also here the question is: how can I improve the volume for this campaign? (I notice this is a recurring question. Would be amazing to understand it myself )
I also threw in a new lander that turned out to have a low CTR but high CR. Results against my other lander are not significant yet, so I'm excited to see what tomorrow's test will bring.

Tomorrow I will see if I can optimize the new lander even more so the CTR might improve.

If anyone helping me needs additional screenshots of data, please let me know!

Thanks,
Lucas

---

Hey guys,

Time for a small update again!

02-08-2016

Source: PopAds
GEO: VN
Conversions: 67
Cost: ±$26.73
Rev: $32.16
Profit: 5.43

This day I only ran my whitelist campaigns again. A very small profit again, but still no idea how to build on that
Help here would be verrrrry welcome!

03-08-2016

Source: PopAds
GEO: VN
Conversions: 22
Cost: ±$12.49
Rev: $10.56
Profit: -1.93

Again, whitelist campaigns only. Today they didn't turn a profit unfortunately, but kept hovering around breakeven.

04-08-2016

Source: PopAds
GEO: VN
Conversions: 136
Cost: ±$91.51
Rev: $65.28
Profit: -26.23

Now that I have a better converting lander, I decided to go broader again. I opened up to all carriers again (Was this a right thing to do?) and got some more traffic. I wanted to gather data on placements that I could potentially whitelist/blacklist, and see if there was a difference between carriers. I also ran the campaign on a slightly higher bid (0.002xx, it was 0.0016x before). @vortex /Amy's whitelist/blacklist spreadsheet helped to identify 4 placements to whitelist and 1 to blacklist. I hoped to get more whitelist/blacklist data, but unfortunately that was not the case (settings: 80% confidence and a min ROI of 0%).

These are stats on connection:

I don't really know what to do with this data. They are all hovering around the same ROI, and testing for significance doesn't give me any insights either.

These are the stats for the carriers.

I'm targeting 5 carriers n PopAds, but they seem to only give me traffic from these two ISPs. They are both negative ROI and testing for statistical significance gives me 75 vs 25% of being the best, so nothing significant either.

I'm unsure of what to do right now. ROI stays negative and I've tried all my tricks with trying to boost CTR/CR/loading speed. Can I consider this a 'winning lander' already? Or should I start splittesting with completely new/different landers again? Or improve on the one that I have?

The questions are outlined in bold again. If anyone would please shed some light on this, I would be very thankful!

Also, if someone has some ideas on how to structure this follow along better, please tell me!

Regards,
Lucas

---

Hey guys,

Today was data-gathering day again.


Source: PopAds
GEO: VN
Conversions: 136
Cost: ±$146.12
Rev: $135.84
Profit: -10.28

Based on this data I could blacklist around 10 placements, whitelist 6 placements and move 2 placements to the whitelist campaign with lower bid (to see if I could maybe get them profitable this way).

Tomorrow I have setup a smaller, optimized version of the landing page, to see if this will do better. Also I want to measure the effect of the black/whitelisting of today.

Could anyone please give me some guidance along this rocky way? Am I going in the right direction? Any help is welcome

---

All campaign settings sound good - except you definitely should test more landers! I like your approach of using the best traffic in the beginning, although once you're green you may want to open up your targeting more.

Is there any reason why you decided to target just android and several carriers?

The bidding landscape is indeed changing constantly. Unless you have a way to automate bid-testing and adjustments on a regular basis, you'll just need to find other ways to increase ROI such that it will be high enough to withstand the volatility that comes with the everchanging bidding landscape.

As for the decline in conversion rates - this is typical of pop traffic. The first day of a new campaign will often give you better performance. Saturation of the smaller placements could be one reason, but not the main reason. I have my speculations on why that is, but can't discuss them publicly without sounding accusing towards the traffic networks.

I would try increasing my bid and see if that would help (sometimes it does). Failing that, I would try to optimize my camp for a couple more days to see if I could increase performance significantly through my optimizing efforts. If things don't look up I'd just kill the camp.

The decision of whether or not to continue with the camp, will also depend on your test approach, as well as how many things you have yet to optimize. For example if you're testing 10 offers and 10 landers, then even in spite of a decline in conversion rates beyond that first day, chances are you'd still be able to find a good offer+lander combo. In that case I would definitely keep running to see what would happen.

Also, if you're wanting to exhaustively and thoroughly test a vertical+geo, you may very well decide to run at a loss just to go through all the landers and offers you can find. This approach will often uncover gems you wouldn't find by "dappling" in a geo - by just testing 2-3 offers for example.


A couple of points to make here:

1)When dealing with traffic segments - such as OS, carriers, placements - you DO NOT NEED to cut inferior ones! You'd want to keep all the traffic segments that are promising and/or profitable!

Not so with landers or offers - because you can only put a single lander and offer in front of each visitor (not counting backbutton etc.) These you WILL want to cut down to your best one. (There are exceptions, e.g. running multiple landers to delay burnout.)

So, never compare traffic segments and cut the underdog!

2)Having said that, a good approach would be to identify the best-converting traffic segments, and set up separate camps to target them, so you can bid higher. Try not to go crazy with this and start breaking out tiny traffic segments into their own camps - you won't want to be babysitting a bunch of little camps. Only do this for major segments such as a big OS (Android/IOS) or a major carrier. (Or a browser if you're finding that a particular lander is converting really well for that browser, for example.)


Right now, don't worry about whitelisting, or the lack of traffic! You should be focusing on improving your funnel while testing for as cheap as is reasonably possible.

You've already tested bids and running at the bid that's giving you the most bang for your buck. Now would be the time to test more landers and offers - the more you test, the better your chances at finding a combo that will hit green.

Remember that the main reason why you're restricting your traffic by targeting only selected traffic segments, is so you could test for cheaper. Your end goal is to find a lander and offer that will convert enough of your traffic to be making you enough profits, for the camp to be worth your time and effort in continuing to run it. You definitely DON'T want to JUST try to make the little amounts of traffic you're currently getting, to reach green - because that will only net you a few dollars a day in profits at best. You want to improve your funnel so that you can open up your targeting to make more of that traffic profitable.

I've seen many people make the mistake of trying to cut their way into green while restricting themselves to a mediocre offer and lander. Don't fall into the trap of spending a lot of time analyzing placement stats and whitelisting/blacklisting to make/save pennies and dimes. You're here to make xx/xxx/xxxx a day in profits, not babysit a bunch of camps making a couple of dollars a day each.

Instead, use that time to test more landers and offers (ESPECIALLY OFFERS!) Once you find a good combination, cut everything that isn't profitable.

Thanks for a wonderful start to a follow-along - I look forward to accompanying you on your journey to profits!


Amy

---

Thank you for your amazing long response and insights into my endeavours! Much appreciated!! Below I will try to answer your questions/respond to your remarks.

I targeted only android based on a hunch. I think I read somewhere that Android traffic was cheaper or converted better or somthing. Also I was trying to limit my hourly ad spend, to get more even data throughout the day. In hindsight this was probably a dumb thing to do. I did open up some of the whitelist camps to iOS traffic, but ROI dropped immidiately, so I got scared and turned it off again (yes yes I know, I should grow a pair and just bite through the big negative ROI moments).
I targeted only the carriers that were allowed by the offer. No point in buying broad traffic that can't convert on the offer, right?

You're very welcome to send me a PM with your speculations!

How would you suggest increasing my bid? Currently it's running at 0.002xx. Should I setup a staggered bid approach again, with my current bid as 'low' bid? Or should I increase my bid by for example $0.50 CPM and see the effects? Also, I notice that sometimes changing my bid doesn't change my bid position. What's the point of increasing my bid if the bid position doesn't change?

I don't really understand this. I thought you had to kill traffic segments when they were unprofitable (i.e. when your blacklist/whitelist tool tells me to KILL! the placement). I understand that this cuts into the amount of traffic available, but there's no point in keeping unprofitable segments right? It does limit me in any efforts to scale it to high numbers, but if I can get steady mid $xx/day for a while it would pay for my bills and would be an amazing way to learn the basics of running poptraffic/optimizing/the business side of AM. This is basically the approach that @matuloo described here: http://stmforum.com/forum/showthread...nd-you-can-too
I feel like focussing on this approach gives me a better foundation than trying to scale this campaign to $xxxx/day for which I 1.) dont have the experience and 2.) don't have the cashflow (small budget, need to grind my way up. I've been put on bi-weeklies already to help out with this cashflow-problem).
What are your thoughts/advice on this approach for this campaign?

This is what I've been doing, but I've been doing this on whitelist-campaigns. Because of my targeting settings I was already down to one big OS.

These are today's stats so far by carriers. I don't know if anything here is promising enough to put into its own campaign and change the bidding or how that would help the campaign.

The testing multiple landers and multiple offers is a concept I find hard to understand.
For example if this is an iphone 6 sweep: should I get as many iphone 6 offers and link them to iphone 6-customized landers? Or should I also get as many samsung galaxy s6 offers and link them up to customized samsung galaxy s6 landers? And should I then also test samsung galaxy S7 offers, by throwing them in the same flow linking them up with galaxy S7-customized landers. That would give me 3 paths under the same campaign:
several iphone 6 landers -> several iphone 6 offers
several galaxy s6 landers -> several galaxy s6 offers
several galaxy s7 landers -> several galaxy s7 offers
It seems like that would require a pretty big testing budget, considering the number of combinations I could get. Some explanation on how I can best go about this 'testing multiple landers and offers' thing would be highly appreciated!

Would you say this is a mediocre offer and lander? Since I'm pretty new to the scene, I don't really have any idea of what would be mediocre numbers or 'amazing winning offer' numbers. I lack experience or reference in this aspect.

Thank you again for your reply and help so far! Have you also seen my update posts underneath my original post? They also contain some stats that you might comment on

---

Ah so it's a carrier-billing offer that only accepts certain carriers! Nice!

Nothing wrong with testing just Android traffic first, as long as you remember to test IOS later. In my experience a vast majority of my profits were made on Android traffic as well actually...


Haha! They're not going to help anybody, so I'll keep the negativity to myself.

Basically I would suggest to just try different ways to make pop work. I mean I could go on about all the ways in which pop sucks, but it won't help us make money from it, and yes it's still possible to make quite a bit of money from pop! Having said that though, I do recommend for people (including myself) to branch out to other traffic types as soon as they have experience.


You could do it either way. Basically the idea of increasing your bid when you see a sudden and drastic drop in CR, is to see whether the drop is caused by being outbid. As for exactly how much to increase by - that would depend on the situation. I may even just bid at max just to see if I get ANY conversions. If you don't start seeing conversions even at max bid - I'd say don't bother running the camp further unless there's other stuff you haven't optimized yet.


That's not what I meant at all.

Killing traffic segments that are UNPROFITABLE or UNPROMISING is GOOD!

But that wasn't what you said. You said:

"I also cut carriers 2/3 (same targeting setting in PopAds, but different name in Voluum), because it performed worse than the other."

As long as you understand the difference between cutting something because it's not profitable, vs. cutting something just because it's worse than the others, then all is well. The 2 stats calculators are designed for these different usages, i.e. the Bayesian split-test calculator at peakconversion is for cutting inferior candidates to identify a winner, whereas the calculator spreadsheet based on binomial confidence levels is for cutting stuff that's unlikely to reach profits under current conditions. Looks like we probably just misunderstood each other. I wanted to clarify things because I've seen a LOT of people make that mistake, thereby limiting their traffic volume unnecessarily.

And I have absolutely no problem with your approach of shooting for xx at first - it's encouraged! That thread you referenced, if you scroll down you'll see my posts where I've posted a screenshot of a really small geo of 1.3 million population in total, from which I was making $40/day in profits from just one carrier on one traffic source. (I hadn't reached $40/day yet when I posted that screenshot - but very shortly after, I cut more placements to increase profits.) That camp ended up making money on autopilot for many weeks until I recently tried to tweak a campaign setting and the CR went to shit. I tried to revert to the original settings but the profits didn't come back - just one of those quirky things that some traffic sources will do to you.


What I would do is break out carriers 1 and 2 each into a separate camp, and test staggered bids for each.

Don't be afraid to test MAX bid! You're bidding quite low for carrier traffic actually. And don't forget that, by bidding higher you'll often get higher CR as well.


I HAVE set camps up in the past exactly like you've laid out, i.e. have different paths for different "prizes".

While that kind of setup is great for a quick and dirty initial test (I was testing a lot of stuff and wanted to speed things up), it's not normally recommended because having all those stats merging together can cause great confusion. I basically set up that camp to see which offers had potential, before I broke out the promising ones into their own camps.

In short though, it would be much better to just target one type of prize in each campaign.


Oh when I said that I wasn't referring to your offers or landers specifically!

I was just making a general comment. The key point I was emphasizing, is that it's important to try your reasonable best to find the best offer and lander first, before doing finer tweaks like cutting placements.

If your ROI is really negative, it would be difficult to get to green by cutting placements unless you have several big bad placements you can cut and STILL be left with enough traffic to bother. In addition, a placement can only be as good as the offer and lander you're running.


I sure did read through all your posts! It's just that I didn't have anything new to say that I hadn't already said in response to your first post.


Appreciate the update - please keep them coming! Seeing green is exciting!



Amy

---

Will definitely keep that in mind! First got to get a profitable funnel going on Android and then i'll test other untested segments!

Indeed we misunderstood each other haha. One of the carriers was unprofitable, whereas the other was actually profitable. In the interest of losing as little money as possible while testing I cut the carrier, but in hindsight this turned out to not be statistically significant. Since that realisation I opened it up again, and have gotten many conversions on it (close to breakeven)

So I followed your advice by setting up 3 camps (low-mid-high bid) for the 2 carriers. I also took added in some landing pages and the same offer from different networks. I decided to test by prize first. I set a budget of 15x average offer payout to test the waters. That budget ran out, so it's time to post some data (not much green unfortunately )

Question: I've read several times that in order to compare campaigns, they should run at the same time. However, every time I setup a staggered bid-approach, first the entire high-bid budget is depleted, then the mid-bid campaign starts, and then the low-bid budget will be slowly drained. Can I compare this data? What can I do to counter this problem?

Data by campaign:


Data by lander:


Data by offer:


If you need me to drill down in any specific way, please let me know and I'll show screenshots!

Question: what puzzled me from this data, is that both Low-bid campaigns (that had the same bid as the main-campaign that made about 500 conversions while hovering around breakeven) have ZERO conversions.. It's the exact same bid at a time that the main campaign at other days still pulled in conversions. Does this have to do with my small testing budget per campaign, or is this just a normal occurrence in pops? Or is there another explanation? I know that it converts nicely on the lowest bid, since I've been running the campaign at that bid for several days and it was around breakeven during that time. Based on this data it looks as if the low bid doesn't work, so I should cut it, but I just KNOW it converts good enough for the lowest bid.

Please let me know how I can approach this data. Was the testing budget big enough, or should it have been bigger in order to account for the multiple combinations of lander+offer? How can I best drill-down into this data?

Also, when analysing the lander-stats, should I use the peakconversions calculator on the aggregated data in the screenshot above, or should I drill down into the seperate campaigns and analyse it like that?

My AM from the network i ran most traffic to for the past week asked me to switch some ad copy around on my landing pages and then offered me a paybump of $0.02 if I would change the ad copy (so my total payout would be $0.50). Should I try to get more out of it, since other networks are pushing this same offer for $0.56? No doubt the $0.02 helps, but more would of course always be better.


Thanks for the help so far!

---

This is the strangest phenomenon! I've never seen it before, but lately I've seen more than one person report this. Perhaps this is because my account's on VIP status so camps start getting traffic almost instantly?

If you can't run them at the same time, then just run them at different times and choose the bid that's giving you the best ROI anyway. Try not to run any of those at typically bad-converting hours - around 12am-9am.


Could you please do offer -> lander -> OS for each camp?


Yeah that puzzles me as well! Before you broke out the individual carriers into separate camps you were breaking-even/in profit for those carriers. But after you did, the CR plummeted...

OK wait a second! I just re-read your previous post - were the carriers only profitable in a whitelist camp? You were also targeting just a single OS correct? If so, do you have the exact same targeting for these staggered bids camps?

Also - did you also add new landers to these staggered bid camps that weren't being run in the original camp? If so then that could have contributed to the drop in CR as well!

Let's see those offer -> lander -> OS screenshots for each camp and see whether we see ANY green first.

If you DO see some green - try to leave the best 2 camps running, i.e. high-bid camp for carrier 1 and mid-bid camp for carrier 2. Collect more data so we can decide whether the ROI is close enough to breakeven for you to cut your way to green.

If you don't see any green at all, then I would highly recommend that you start back at ground zero - by opening up all your targeting and mass-testing offers and landers.


I personally don't see a big problem with using aggregated data, unless you're wanting to be so accurate in your optimization that you'd be willing to cut landers for each carrier and bid separately.

Having said that - it would be nice to collect enough data so that you can see a sort of trend, e.g. lander 3 is in the lead for both carriers at multiple bids. I always like to look for trends to reinforce my decision - although the largest factor will always be analyses from stats tools.


I highly doubt a paybump of just $0.02 will make/break your camp.

Your focus should be on increasing ROI in BIG ways, i.e. testing offers and landers. You should aim to monetize a large enough part of the total traffic to be making more than peanuts in profits, and THEN cut everything that isn't profitable for your best lander and offer (placements, hours of day, etc.)

I would sooner test that same offer from the multiple networks before I'd test that tweaked lander to go after that 2 cent increase in payout, as they can convert VERY differently (I've seen before, differences of over twice the CR for the same offer from multiple aff networks) - not that you can't do both though.



Amy

---

I ran the test around 5PM-8PM. Going back into the data it turned out that these were not the best hours for the main campaign, but it still pulled in some conversions.
Takeaway: Next time I test something I will pay closer attention to the local time and how the camp performed in the past.

That's indeed what I've been doing, but what I can't understand is that the test-data tells me to drop the low-bid campaign, whereas that campaign had been breakeven at $100/day revenue for several days.

The carriers were only profitable (a few dollars) in the whitelist campaign, but were very close to breakeven in the main campaign at $100 revenue/day (at around 150-200 convs). I used the exact same targeting for the new campaigns as I used for the main campaign, so only 1 OS (Android), primespot only, site quality 3+. The only thing that was different was that I didn't have the same placements blacklisted. I only blacklisted the top 100 that I mentioned in the first post of this follow-along. Maybe the bad placements had been eating the entire budget? This would also be a weird conclusion, since most of the bad placements that I cut actually got a lot of conversions, but just couldn't get profitable. So getting 0 conversions is just crazy!

I did add in 2 new landers and 2 new offers (same offer on different networks), but all combinations got conversions. Just not on the original bid. I understand adding new landers/offers can increase/decrease the CR, but to have a drop to 0 conversions on the same bid is just crazy. Was the 15x offer payout per campaign too small of a budget for all these different combinations?

IMPORTANT: when I just checked to take the screenshots both low-bid campaigns had 2 conversions. Apparently they were delayed by quite a bid, since I wrote the post a few hours after the tests had stopped. Conversions came in from all the 3 networks, so it's not an issue specifically with 1 network. This invalidates all my complaining about the low bid campaigns above, since I suppose these drops in CR can be explained by the new combination of landers+offers+time of day.

I did not include the OS, since I only targeted Android in this test-run. I figured including it anyway would only clutter the already cluttered images.
I could not order them in a logical way, so pay attention to the labels/names when interpreting the results.

Overall - New situation:


Carrier 1- High bid


Carrier 1- Mid bid


Carrier 1 - Low bid


Carrier 2 - High bid


Carrier 2 - Mid bid


Carrier 2 - Low bid


All this data gives me some kind of information overload: I don't know where to start, what stats to pay attention to, or what could have promise. I would love to get some guidance here!

So there is actually some green at some points, but I doubt that's very significant. More data here is needed.

What I don't understand though, is that I already had a campaign (at low bid) doing bigger volume that was doing breakeven. Why throw away all that data and try to optimize my way from -50 ROI to profits, when I could be optimizing my way from -5% ROI to profits? Is this a faster/more cost-effective/efficient way to get to green rather than following @Yseke's approach that he mentioned in his post where he said 'build a blacklist campaign and increase bid, while blacklisting/whitelisting placements?

Wouldn't it have made more sense to keep testing landing pages at the breakeven bid, while blacklisting really bad placements and whitelisting the good ones? Maybe this wouldn't be the best approach to get to $xxxx/day, but I'd suppose with bigger volume it could have the potential to go to mid-high $xx/day?

I lack experience, so I don't know what the potential could be. Maybe I've also grown attached to this campaign a bit, which might cloud my judgement.

I guess I wasn't entirely clear in my last post. It was not so much a request as it was a compliance issue. I HAVE to change my landing page copy before going live again, but I would get a small paybump because of my quality/volume I guess.

Once again, thanks for your detailed reply!

---

It's really difficult to say what the cause could be for the lack of conversions. Pop is an especially volatile type of traffic - even the same camp could perform very well one day and completely tank the next day with no apparent reason.

My approach is to just do the best I can in testing bids, offers and landers, and to identify the best traffic segments to target - and then if I can't get something to continue performing well, I'd just pause the camp. Always be scaling and always be setting up new camps to test new stuff so you'd have new profits coming in to replace dying camps.

Regarding that most-blacklisted-placements list: I've observed the exact same thing you did - that they're not actually bad placements. With the exception of placements that are heavily infested with bots, most placements can only convert as well as your offer and lander will allow, meaning some people must be making them work but not others depending on how good their offers and landers are. This leads me to suspect that most of the placements on that "bad" list only made it there because they're high-profile, high-trafficked sites - the bigger a placement is, the more people would have received traffic from it, and therefore the more likely it would be for it to be blacklisted. This is why I stopped using that list and just based my exclusion on my own placement stats.


Not sure. Personally I'd run 2x payout multiplied by the number of lander+offer combinations. So if I was running 3 offers and 10 landers and my average payout is $1, then I'd run $60 to get a rough idea.

If I was also testing 3 staggered bids then I'd run $20 to each bid.


Aha! Nice to know! So the camp isn't hopeless after all!

Thanks for all the screenshots! There is some green although not a lot. Worth running further are: Carrier 1 either high or mid bid camp, and carrier 2 high bid camp.

Actually, what would be a better plan, would be to just keep these camps on pause for now, and use your original camp to test more offers and landers first until you find a really good combination.


Good question!

The answer is that when you're targeting each carrier in a separate camp, you'd be able to adjust your bid to find the optimal bid for each carrier. What I didn't/couldn't foresee was that the performance ended up being so different (which may be due to the new landers and offers and running different placements).

You're right - you could use the original camp to continue testing until you have a winning lander and offer, and THEN break out the carriers into separate camps. That actually would have been better - like you said, you'd want to make use of existing/previous data as much as possible. It's just that I saw the camps were green which indicated it was the right time to break out the carriers and potentially be able to bid higher and get more traffic. I forgot you were still in the middle of split-testing.


Oh! I would suggest to just keep gathering data to see if that particular lander would even win. If another lander ends up being the winner, then you wouldn't need to change the text, right? Unless you have similar text across all landers.

Question for you: How many landers and offers have you tested so far IN TOTAL? Ideally you should test all the usual landers you see on adplexity for example, and then use that to test lots of offers. VN is a big geo. If you can find an offer that converts well on wifi, and scale it fully, you could be looking at at least 4 figures/day in revenue. Carrier traffic is great and I love it, but there's not much volume and it's expensive, so I try to only run high-converting 1-click offers on carrier traffic.

I'm not suggesting for you to give up on the current carrier camps you're running. A few bucks/day could be very encouraging for a new person, and will keep you motivated to go after bigger profits next. But don't restrict yourself to just running carrier traffic. Keep testing offers and include wifi in your targeting - because that's where all the volume is.



Amy

---

How do I recognize a very good combination? I thought that breaking even with an offer was a pretty good start, if I didn't do much optimization next to it. These were the settings my main camp have been running on so far, which caused it to breakeven with around $100/day revenue for several days. Targeting that restrict certain types/segments of traffic have been written in bold text:
Quality: 3+
Frequency cap: 1/24hr
Primespot: Primespot Only
Adv type: popunder (required)
Max bid: $0.002xx
Throttling: disabled
Categories: Allow all
Country Targeting: Vietnam
Society targeting: Allow all
Operating Systems: Android (All)
Browsers: Allow all
Screen resolutions: Allow all
Form factors: Smartphones
Devices: Allow all
Connection types: Allow all
Connection speeds: Allow all
ISPs: 5 ISPs that are allowed by the offer
Time targeting: 09:00-24:00 local time
Website targeting: exclude: STM PopAds top 100, several placements that I could kill according to the blacklist-whitelist tool (some of these were big placements that I broke out into a whitelist campaign with a lower bid).

The traffic estimator tells me I could get 38k impressions/day at current bid ($0.002xx). At top bid this would be 189k impressions/day.

As you can see from the settings above, I target all connections (including Cable/xDSL and Cellular/Carrier traffic). Most conversions came from broadband and xDSL:
Data is from last 7 days.



This is the data per campaign


Where am I going wrong?

In total I have, in this campaign, tested 4 offers (3 galaxy s6 offers, 1 galaxy s7 offer, that did worse), and 5 lander-themes. I tweaked some landers if they performed well enough, so that's another 5 variations on a theme:
- Facebook questionnaire
- Orange spinning wheel
- Facebook spinning wheel
- Whatsapp questionnaire
- Google questionnaire
These were the themes that kept recurring in AdPlexity. Did I miss some?

Does this campaign with this targeting have the potential to get to mid-$xx/day, or is it better to start 'fresh' with a RON and multiple offers/landers from the get-go?

Thanks again!

Lucas

---

Ah! So you're getting a lot of volume AND you're practically breaking even on wifi traffic! Why are we wasting time breaking out carriers into separate camps then lol!

You're really close to making profits. Let's not focus on carriers for now, nor the whitelist camp. Your original camp has a LOT of potential.

OK here's what I would suggest:

1)Once you cut down to the last offer and lander, mass test offers! Make a list of all offers you can find on all the affiliate networks you're with, for the same offer type and geo. Again, this is the most critical task, and the best thing you can do to potentially increase ROI by large amounts.

2)Once you're down to the best offer, cut everything that's still not profitable then (including placements, time of day, etc.) That would also be a good time to open up your targeting to test stuff you haven't already tested, or stuff you've cut temporarily while you were running inferior landers/offers. (Such as sites with <3 quality.)

3)You may have done this already, but if you haven't: Clone this camp to test staggered bids.

The reason why I keep urging for people to test more offers is because it's THAT important! The success of the entire camp hinges on it. Don't think that testing 10-20+ offers will require a huge budget either! Most often what would happen is that you'll see a couple or a few offers making a bunch of conversions early on, which will allow you to cut a lot of the offers before long. It sounds like you have a lander that works. So let's find a better offer and then cut unprofitable placements. Having a better offer will enable you to scale more effectively as well. So start testing offers!


Amy

---

So guys, this thread is not dead. Unfortunately I had to learn the hard way that affiliate marketing is a business of ups and downs. Basically all the offers in the geo I was running/testing went down due to regulations. Bad news for people running sweeps in VN I guess: hard times ahead.

Regarding the previous post by Amy:

This campaign had Samsung Galaxy S6 sweepstakes as 'offer type'. Does this mean that from this data I can only conclude this campaign has a LOT of potential for Galaxy S6 sweepstakes? Or could I update my winning lander to Galaxy S7/iPhone 6s sweepstakes 'offer types' too?
I think my main question here is: what makes this campaign have a LOT of potential? The geo-lander-offer combination? The geo-lander combination with offers still to be tested to find the winner? How does the bid relate to the profit-potential of this specific campaign?

On this specific campaign I tested about 6 landers with variations for some of those landers. On the backend I rotated 5 offers at 3 different networks, with 1 'winning' offer that I expanded upon. I lost less than $150 on this campaign on around $1100 spend, with a little under $1000 in revenue (so -12% ROI). I got experience building relationships with networks, negotiate payment terms and paybumps.

Throughout this Follow-Along and you guys' input I learned how to do things better from the get-go for the next campaign I start. I will update this follow-along when I have something running again.

This is the data on this campaign.


Does anyone have specific tips/things I can do right now/ask of my affiliate managers?

Thanks to everyone reading this follow-along. ESPECIALLY THANKS to the people contributing to this follow-along. You guys' suggestions definitely helped me and changed my view on certain things! Pure gold and I really appreciate it!

Any questions/remarks, post them here or hit me up on skype!

Lucas

---

Shit! Regulations problems again? They were really going after advertisers and aff networks and affiliates a few months back, and then things got a bit better, but I guess we're back on the radar with all our aggressive marketing...

Sorry to hear! But there are a lot of other geos to test, so I wouldn't dwell on that one camp too much.


I would start by focusing on just S6 offers.

And of course you can test other "prizes" as well, because when you see something converting well, it means both the offer AND the traffic have to be good - so you can test different offers on the same good traffic to see what works best.


Great question!

Without looking at detailed stats - I made the "a LOT of potential" comment just based on the fact that your high-bid camp was breaking even (at -12% ROI), and you had more stuff to cut yet (placements etc.) to optimize the camp further - and the fact that VN has a ton of traffic so you can cut aggressively and still have enough left over.

To gauge how promising a camp is, I would test staggered bids of course. Then I would run at the best bid for a bit, then drill down to offer+lander plus major traffic segments such as OSs - and the more green combos I see, the more promising the camp will be to me.

It's too bad that the regulations are being enforced so strict again. Otherwise, if you could mass-test offers using your best lander, there's potential to turn this into a 3-4 figures/day camp.

The good thing is that, as you said, you've gained good experience from this. Offers can go down, but the experience will benefit you in future camps.



I'm subscribed to this thread - so if you do update it with a new camp I'll know to check it out.

Again - thanks so much for posting! And I do look forward to seeing future updates from you.



Amy

---

Thanks for your amazing feedback again! Definitely something I can work with next campaign!

I decided to target a different geo because of the regulations in VN. This geo has considerably less traffic, and the offers I found (all the same prize) have different targeting settings. Since it's a bit more difficult to map the right carrier to the right isp (which I can target), and considering the fact that I found that the bulk of my conversions came from wifi, I'd like to just target broad (ie: no restrictions on ISPs in PopAds) and reroute the traffic based on rules in Voluum.
However, I have difficulty understanding precisely how to setup this flow. What's the best setup?

Currently I have it setup like something like this:

Default: Don't Know
Rule 1: Mobile Carrier IS NOT 'carrier1' --> landers + offer1
Rule 2: Mobile Carrier IS NOT 'carrier2' AND OS IS NOT 'iOS' --> landers + offer2
Rule 3: Mobile Carrier IS NOT 'carrier1' OR 'carrier2' --> landers + offer3

Offer information:
Offer1 specifically does not allow carrier1
Offer2 specifically does not allow carrier2 and iOS
Offer3 specifically does not allow carrier1 and carrier2, but does say 'carrier4 only'

I based the rules on 'IS NOT' logic, since Wifi traffic could potentially convert too, and if writing rules like Mobile Carrier IS xxx, no Wifi traffic would end up at that offer.
However I feel like there are some issues with these rules. I don't want to send traffic from certain ISPs to offers where they can't convert anyway. I also don't want to cut too many ISPs in PopAds since I could be left with too little traffic from the start.
How would you guys go about this?
And how would you set the Default rule?

Thinking of it now:
Would it be better to do it like this:

Default rule: landers + all offers
Rule 1: Mobile Carrier IS 'carrier1' --> landers + offer1
Rule 2: Mobile Carrier IS 'carrier2' AND OS IS NOT 'iOS' --> landers + offer2
Rule 3: Mobile Carrier IS 'carrier1' OR 'carrier2' --> landers + offer3

Or are there issues with this setup too?

Any insights would be greatly appreciated!

---

EDIT: I made a mistake about the order in which Voluum evaluates rules. I've set up testing to confirm that Voluum will actually evaluate the TOP rule FIRST, and work its way down to the last rule. Any traffic that doesn't match any of the rules will go to the default path. Apologies about the mistake!

Haha yeah voluum rules always takes me a bit of time to figure out!

The main thing to note is that the last rule is evaluated and enforced first, and only if the visitor does not match the condition(s), will they be compared to the second-last rule and so on - and of course if they're not a match to ANY rule then the default path is executed.

So I see rules as being an upside-down sieve - where every rule filters out and redirects some of the traffic, before sending the rest to the next rule above it.

You've got it partly figured out! Let's go over the 2 setups you have - and then I'll propose a third.

So what will happen here? To keep things simple let's just say you were only buying carrier traffic for Carrier1&2&4. (We seemed to have skipped Carrier3 but no matter...)

At Rule 3, Carrier4 traffic will get "caught" and redirected to Offer3, but carrier1 and carrier2 traffic will be allowed to continue through to Rule2.

At Rule 2, Carrier1+non-ios traffic will be caught and redirected. Allowed through will be: Carrier1+ios & Carrier2.

At Rule 1, Carrier2 will be redirected, leaving Carrier1+ios to go onto the default path.

So for carrier traffic, this setup will redirect everything fine. But what if you send wifi to this setup?

Since it's neither Carrier1 nor Carrier2, it would get caught and redirected by Rule 3 to JUST Offer3, which is not what you want.

What you could do is add a Rule 4: Mobile Carrier IS 'Other' --> all offers that accept wifi. This way all the wifi traffic will get redirected first, leaving all the carrier traffic to go on to Rule 3. I've never done it this way before though.



With this setup, all Carrier1 and Carrier2 traffic will be sent to Offer3 - which is not what you want, because Offer3 does NOT accept these carriers.



This is what I would suggest:

Default: All Offers
Rule 1: Mobile Carrier IS Carrier2 --> Offer1
Rule 2: Mobile Carrier IS Carrier1 AND OS IS iOS --> Afflow/YTZ
Rule 3: Mobile Carrier IS Carrier1 AND OS IS NOT iOS --> Offer2
Rule 4: Mobile Carrier IS Carrier4 --> Offer3
Rule 5: Mobile Carrier IS NOT Mobile -> [all offers that accept wifi]
(Alternate setup for Rule 5: Mobile Carrier IS Others -> [all offers that accept wifi]. I haven't tried this one though.)

So the only traffic that will go to default will be ALL carriers other than Carriers 1&2&4.

I would send Carrier1+IOS traffic to a monetizing service such as Afflow/YTZ, because none of the offers will accept it. After you've gathered enough stats to decide, you may want to set up separate camps to target 1)IOS, and 2)Android and other OSs. That way, you can exclude Carrier1 in your IOS camp.



ALSO:

I would actually suggest to target wifi and carrier traffic in separate campaigns, because carrier traffic is typically a lot more expensive than wifi traffic so you'll need to bid higher.

I know it can be tricky to figure out which popads isp corresponds to which carrier, and that sometimes it's not possible to separate wifi and carrier traffic for a certain ISP. But figuring out what you can and cannot target would be a first step - I've written about this topic here:

http://stmforum.com/forum/showthread...esting!-Part-1



Have fun!


Amy

---

ALERT: I made a mistake regarding how Voluum evaluates rules - please see the "EDIT" in the beginning of the previous post.

All the logic is still correct - except the ORDER of the rules should be REVERSED. I've set up an actual test to confirm this.

I think I've been lucky that I never had rules set up where the order mattered (i.e. when rules criteria are mutually exclusive / don't overlap). Well now I know!

Many apologies for the wrong information!



Amy

---

Home > Mobile > Follow-along Campaigns