Home > Mobile > Follow-along Campaigns

---

International Mobile ONLY (17)

---

Getting my feet wet with mobile and cluttering the forums with too many threads, I'll keep my posts organized here!

Resources I've followed to get here over the past week:

• Overview (what to Target): http://stmforum.com/forum/showthread...bile-Marketing
• Making Ads and Banners: http://stmforum.com/forum/showthread...-Mobile-Banner (doing this tutorial is what made Mobile call out to me, I have a lot of fun making animated banners!)
• Making Landers: http://stmforum.com/forum/showthread...Actually-Works

Here is my tracking setup:

• Tracking: CPVLab on Linode (hosted in Dallas)
• CDN: MaxCDN

For my offer details:

• Countries: Targeting Spain only
• Offer: The first offer I've picked is Moshield Antivirus ES on F5 and C2M, it's a pin submit with $9.60 payout

Campaign targeting is Android and Country only, and I've split this up into two campaigns:

• C7_MOS-ES_CAR_APP: Carrier only, In-App traffic
• C6_MOS-ES_CAR_WEB: Carrier only, Web traffic

Overview

C6_MOS-ES_CAR_WEB: Carrier only, Web traffic
* Impressions: 9,449
* Clicks: 51
* CTR: 0.54%

C7_MOS-ES_CAR_APP: Carrier only, In-App traffic
* Impressions: 74,879
* Clicks: 431
* CTR: 0.58%

Banner Breakdown: (I'm cutting banners per caurmen's guide)


C7_MOS-ES_CAR_APP: Carrier only, In-App traffic

C6_MOS-ES_CAR_WEB: Carrier only, Web traffic

I cut ads 3 and 5 because they didnt get any clicks after 1200 impressions



Landing Pages
Getting tired making the tables! I'll just list the overall stats:
LP: 273 clicks, CTR 18.68%, no conversions
DL: 253 clicks, no conversions

Conclusions and Next Steps

• Ads: Looks like I have a pretty good ad (Ad4) so I should make more variations of that.
• Landing pages: Still working through understanding CPVLab so not sure what LP to cut. Caurmen's guide says to not cut anything until 5 conversions existing between them, so I'll hold off on that.
• Sites/Publishers: There is one site that is getting about a third of my placements. Not sure what to do about this, whether it's good or bad, since I have no conversions

Next steps:

• Keep collecting more data for the pin submit? Not sure how many clicks I should be expected before seeing a conversion. Network stats is about 0.5% CVR...I figured I should have seen one by now?
• Launch a CPI campaign

As a side note...I believe I set up tracking in CPVLab backwards: I currently use the banner name as the "Ad" and the publisher as the "Keyword." I believe it should be the other way around because CPVLab default flow is Ad->Keyword->LP->Offer. So if that's correct, I'll add fixing this as a next action too

---

If you are targeting many placements, you cannot cut ads by caurmen's guide. You even commented in the thread where he mentioned it

http://stmforum.com/forum/showthread...l-Significance

---

Couldn't I still cut ads based on the split test calculator? Perhaps I should use the statistical conversion tool on publishers instead of ads?

---

There are a range of approaches to placement cutting. IMO, I'd recommend cutting them on ROI as per Section 10 of the guide (before you cut any ads unless the ads are doing stunningly badly).

Multiple-placement campaigns are more statistically complicated in general, because you've got two unknown components - ad quality and placement quality. Generally I'd currently recommend narrowing down placement quality first.

I know there's another guide needed for this - will be on it soon!

---

Got it, I have ~400 clicks to my LP so far and see that these two placements are getting the 47% of the traffic. I will pause these placements to send traffic to the other 17 sites:

---

Nice detailed stats. I approach mobile a little different to most here however if i was testing your offer on a network like incrowdads/decisive i'd target the most popular 3 (or so) sites and top 3 (or so) handsets only, to eliminate the expenses needed to get statistical significance on promoting an offer with such a high cpi. Just my 2c

---

Interesting, I may try out your strategy. My thinking was that since the top 2 placements already received 100+ clicks and no conversions, that I should let the other pubs "catch up" so-to-speak

---

Just re-read your post and noted the payout of your offer - ouch. That's pretty high for mobile: as mykeyfocus says, you're going to need a fairly large budget!

What does 54 clicks translate to in terms of how much you've spent to get them?

---

Update on the PIN Submit
Here's my latest stats, no conversions yet. Pretty sure it's my ads and landing page.es

I see from the payout I'll need a larger budget for this. But what's a good estimate of what kind of budget I'll need?

3x offer payout per ad times 10 ads times number of placements? So $27*10*20 = $5400....is that correct?

Launched a CPI Campaign

On the bright side, I launched a CPI campaign and have two conversions! It's a 0.30 payout on install and I'm direct linking so far. That one has 210 clicks, so a CR of 1.9%. I have it only set up to target Carrier and in-App banners (not testing wifi and web traffic yet). I plan to "clone" this campaign after making more ads. All my ads are around 0.5% CTR for this campaign

---

If you are getting your feet wet in mobile, my recommendation is.... PICK ANOTHER COUNTRY!

Spain is an expensive country to target and at that CPA, you're gonna require a larger budget to test. If I were you, I would pick a lower payout (say $3 and below) offer and test it first. At least you are able to obtain data faster and optimize faster. That's the key to a winning campaign. Once you start to see green, you can build up your cash flow to test new offers, and build up your confidence level too when testing new things.

Imagine if you lose spent few hundred and make 1-2 conversions on this campaign, would you say "Argh mobile is not for me, it's so tough!".

Picking the right campaign when you are starting out is just like someone going for a singing audition. Pick the wrong song and you are gone.

Keep us posted on your result!

---

Yep what kok said! Start low and easy. CPI is the way to go man. Spain is one of the toughest for sure, especially with the high payout you are working with. You should also be aiming for a 1% + CTR for your ads if you are running on 320x50 banner placements.

---

Alright, I hear ya! I had the gambler's mentality and threw up a single lander with a couple of ads hoping for easy conversions. I'm not testing enough Landing pages on the PIN submit to say I'm giving it a real shot.

So...here are stats from the CPI offer I launched in a less competitive country. I'm running 6 ads and have this split tested as follows:

• Carrier and In-App 320x50
• Carrier and Web 320x50
• Wifi and In-App 320x50
• Wifi and Web 320x50

I'm not really getting impressions for the last three, and I think it's because I'm bidding the same amount for them all (I'm bidding 35 cents, no real reason, I just picked a number at random. I am thinking of increasing my bids for the bottom 3 campaigns to get more data and impressions)

Here are the top 10 pubs for Carrier and In-App, since I only have conversions for this due to low traffic for the other 3

And there's another 20 or so pubs with only 1 or 2 clicks.

Here are how my ads are doing, none of them are above 1%. I'm tempted to cut ads when I reach 8000 impressions for all, but I will keep them all running if it's above 0.1% until I cut more pubs. (actually, I cut Ad #2 because i made it really late at night so the ad quality wasn't up to my personal standards....that...and there were no clicks after 1200 impressions. data speaks for itself)


EDIT: Looking back through my stats for pub #248 that I cut: 60% of the traffic was through Ad#1. I am now wondering if I gave it a fair shot with the other ads. I'm going to keep my "cut" decision and let this be a lesson for me to launch campaigns faster with more ads at once

---

What's the payout on your new offer? That'll be critical to the statistical significance levels you need.

(Or at least, to calculating expected ROI, from which you can then obtain statistical significance.)

---

The payout on the new offer is 0.31, and I upped my bid to 0.37.

My biggest question is whether I'm doing this calculation correctly...I'll walk through my thought process for those following along

ROI Calculator for CPC
I'm changing the ROI calculator from caurmen's Tutorial #10 to do a CPC calculation instead of impressions.

My ads are currently getting 0.5% CTR, and I think it's enlightening to see how the Minimum Viable Conversion Rate drops proportionally to CTR changes -- I should make better converting ads!

Pubs Performance

Now, I'll run these 10 pubs through http://statpages.org/confint.html to get the confidence interval to find my High-end confidence level. Oh, by the way, the total cost has changed because I updated the average CPC value in CPVLab to be more accurate.


Analysis

If my ads are at the current CTR of 0.5%, I can cut right now to one pub. However, if I can get my ad CTR up to 1%, I will need more data to make a decision.

According to the calculator, 9 visitors with no conversions is a cut at 0.5% CTR. 21 visitors with no conversions is a cut with 1% CTR. And this begins to fall into the trap of "testing to significance" instead of choosing a number to test at. I'll keep my number at 30 visitors to decide my cut, since it's pretty cheap ($3 per pub) to test.

Conclusion

So guys, if this makes sense I'll be cutting the top 5 pubs and let the rest run to 30 visitors

---

I finished a bidding test by pushing my bid to $5 and seeing what kind of impressions and the results are interesting: not much difference between my $0.37 bid and my $5.00 bid

54000 Impressions/day x 0.5% CTR x 30% CVR x $0.31 Payout makes this a $25/day campaign, optimistically

Hmm....makes me wonder if I should switch offers in different countries now (I'm targeting Asian countries). I should launch a CPI in Spain

---

All looks very solid - well done!

It's this kind of hard work - whether carefully checking your stats or grinding out new ads - that lets you succeed in AM.

---

This is a great thread. About time someone went through the grind properly and talked actual numbers. Inspiring!

---

Home > Mobile > Follow-along Campaigns