Home > Paid Traffic Sources > Mobile

---

PropellerAds Optimization (8)

---

Hi,

I have a question regarding optimization. Today i have started to promote an offer with PropellerAds from Mobidea. I used Vortex guide to make all the setup, and i am promoting an offer with direct linking. I have created several campaigns to promote the offer, to be able to have a lot more control over the cost. Now i have an issue in trying to understand what statistical significance means to pop-under ads.

There is a post on this forum regarding statistical significance, but is more related to banners, and i am more interested in statistical significance for pop-under ads from PropellerAds.

Basically, what i want to find out is...
- when is ok to cut placements, if they don't have any conversions?
- How many impressions should the placement make, before considering to negative it out?
- i have 2 carriers (from 4 total) that made over 2k impressions and 0 conversions. Should i pause them? - i already paused some placements in those 2 campaigns, and lowered the bids.

All the best,
Tudor

---

Hey Tudor,

Dealing with the huge inventory of PropellerAds is kinda tricky without a lander since you are direct-linking your offers you are somehow flying blind without any insight on the traffic quality.
What I would suggest is adding a landing page in your funnel to filter out low CTR placements first, then continue on split-testing offers to focus on the best LP-Offer combination.

Getting back to your questions here are a few suggestions:
- You can cut placements aggressively by pausing placements that have spent 2x the payout without conversions.
- To cut placements based on impressions you have to consider their cost if there are no conversions and compare that placements to the converting ones. Maybe is not worth to bid on a placement with extremely cheap traffic that doesn't generate conversions and no one else is bidding to.
- It is hard to say that the problem is the Carrier in your case. Basically if most of the traffic is coming from these 2 carriers and very little from the other 2, then you should try increasing your bid to have a more balanced traffic spread among 4 carriers.

---

For pop stats, these would be the methods to use for split-testing landers and offers, respectively:

https://stmforum.com/forum/showthrea...Banners-Part-1

https://stmforum.com/forum/showthrea...211#post289211

In your specific case though, because you're direct-linking to individual offers, none of the above applies (I provided the links anyways for future reference - when you start running landers.)

Basically all you're doing is test offers until you find one that converts at least once every 5x payout spent on traffic, then keep an eye on your placement stats and run until you can tell whether there's a chance you can reach green by cutting placements. If so, keep running, if not, pause and test another offer.

As for how to cut placements, I do have a tool for that, but frankly I think it's overkill for pop:

https://stmforum.com/forum/showthrea...Banners-Part-2

It's accurate, but it needs a lot of data before it would make a verdict. I wouldn't recommend using this tool for pop, because pop camps don't stay profitable for long, so efficiency is the way to go.

Instead, I would suggest to cut placements that are in loss by 2x payout or more, regardless of whether it's had conversions or not.

Keep in mind though, that just because a placement is in loss at the current bid, doesn't mean it can't become profitable at a lower bid. This is why testing bids is important. I would suggest to set up multiple camps with different bids, run them for a bit to collect data, pause the camps that look hopeless (i.e. unlikely to be profitable after cutting placements), then cut placements for the remaining camps.

You can't really base decisions on the number of impressions, because this metric isn't directly related to ROI. Other metrics such as offer payout and traffic cost will play more major roles.

For example: 10k impressions without a conversion, should I pause the camp or not?

Case 1: Traffic costs $0.50 CPM, payout = $5. 10k impressions would only be the equivalent of 1x payout in spend. Of course it would be too early to make a decision.

Case 2: Traffic costs $5 CPM, payout = $0.50. 10k impressions would cost $50 (!!) which would be 100x payout in spend. In that case you should have paused the camp WAAAAY before it reached that kind of spend!

Hope that's clear! Please feel free to ask for further clarification.



Amy

---

Hi Amy,

Thank you for your answer. I uploaded below a screenshot from my Voluum acount. The data is from yesterday and today.
The first 4 campaigns are all paused. The first 4 campaigns are for a Video offer, that is targeting ZA, Android users, and
4 different carriers.

The payout for the offer is $0.57, and i was paying around 2$ CPM. The first 2 carriers made conversions, and the last 2 didn't so i paused them after 2000 impressions.

When i first started them, i was targeting all the operating systems, but then i realized i need more control, so i created an
particular campaign for each of the operating systems.



Maybe i should add a screenshot for each day, because that will reflect the changes i made in terms of splitting the campaigns based on the OS.

Here is the one in 9/6/2017



Here is the one for 9/7/2017



So, we can clearly see that the first campaign had an -80% ROI yesterday and today it was at -55%.
I guess it helped to split the campaign by OS. Overall for that campaign i have spent $18.03 and made only $5.15. Besides the OS split,
i also lowered my bids.

Lower bids affect the quality of traffic? I mean if i pay less for my CPM, will i get crappy visits, or the traffic is the same no matter the bid?
You said above that i should find an offer that converts at least 1/5x payout. In my case that would mean:

- 5*0.57=$2.85

and i was paying a little less than that...

- $18.03 spend / 9 conversions = 2$

Yesterday before the split and losing some bad zone ids i was paying: $11.65$ / 4 conversions = $2.91
Today after the split on OS and losing some bad zoneids i was paying: $6.38 / 5 conversions = $1.27

So, i managed to cut the CPA by more than 60%, but even so, i don't know what to do next to cut it even more, to get below $0.57.
The traffic seems so expensive, and on propeller i cannot bid less than $0.50.

What should i do with the second one that yesterday made 7 conversions, but today didn't made any?

Thank you very much for your time,
Tudor

---

Thank you for your answer. I really appreciate it.

Tudor

---

ZA is a tough geo to crack - even the 2 big carriers would normally take a lot of optimization to get profitable. So you're doing extremely well!

Yes - setting up granular camps like you did to target each carrier+OS combination is a super-good thing when you're running in a big competitive geo like ZA. However, all the camps can be difficult to manage. An alternate method would be to set up one camp per carrier first, each targeting all OSs, then observe OS stats to see if there are OSs that are performing considerably better/worse. You can blacklist the bad ones and/or break out the good ones in their own camp. That would minimize the number of camps you need to manage.

It would help if you could show some placement stats. If you do, please show 2 versions: One sorted by decreasing impressions, and the other sorted by decreasing conversions.

How much traffic do you still have? As long as you still have a good amount, just keep cutting placements.

Also - don't just aim to decrease traffic costs. When you bid low, you can't get traffic from some of the best placements - because your competition will bid higher for them after seeing how well they convert, so the traffic source won't give you any volume from them (i.e. multiple people offering different prices for the same goods, so goods are being sold to the highest bidders). ZA is one of those geos where you'll need to invest money upfront into cutting stuff. But once you do, you can benefit from the leftover placements for a long time, running offer after offer on them.

So what I would suggest is to clone your camp a couple of times and assign higher bids to them, then cut placements aggressively for each camp. Anything that's in loss by 1x payout, cut it. Be prepared to lose money though if you decide to go this way - remember, this would be an investment for all future camps. If you're not wanting to invest heavily, then stay away from ZA for now. There are lots of way cheaper geos you can go conquer to build up cashflow.





Amy

---

That's what i did.

At the moment i stopped all traffic, for the past 3 days. I am waiting for an response from Propeller Ads to cut some browser traffic, that is killing me.

That is the answer i was looking for regarding the bids. Thank you very much!!!

At this point i've cut most of my campaigns, and kept only 2. The ones that made conversions. I didn't want to start with ZA, but other offers that were for Malaysia or any other country that you suggested in your guide, weren't approved. So, i started out with what was approved. Right now i am waiting for Propeller to cut some browser traffic that i don't want into my campaigns, and then i will make them live again. I hope i will see better results.

Do you have a guide for Mobile and Landing pages? The one you made is for direct linking, and i am curious if there is something here that explains mobile and landing pages like you did in your guide.

All the best,
Tudor





Amy[/QUOTE]

---

Sounds good!

Writing that at the moment - it's very difficult trying to come up with a one-size-fits-all approach. It's much much easier to look at a specific set of stats and provide suggestions.

In the meantime, feel free to ask questions about landers - will do my best to provide guidance. Here are some links that may help:

https://stmforum.com/forum/showthrea...l=1#post324392



Amy

---

Home > Paid Traffic Sources > Mobile