Home > Paid Traffic Sources > Mobile

CTRs, Conversions, and Statistical Significance (15)

02-18-2014 06:44 PM #1 rfcapital (Member)
CTRs, Conversions, and Statistical Significance

I've read the statistical significance post in the how to get started guide, so I know some basics, but I'm not sure about anything because I'm a noob.

Here's my question:

I have an ad on a mobile app install offer (Android US) running on Decisive:

Direct linking (I know, I know, landers, I'll get to it eventually, I was eager to get started)
18,800 impressions
0.3% CTR
0 Conversions
$0.13 CPC
$0.80 Payout

It seems to me that it would take a miracle for this campaign to turn around, but with the amount of impressions I received in such a short period, I'm wondering if the same concepts of statistical significance apply.

So, when do you cut your ads running on mobile? Should I expect a higher CTR and conversion rate than a typical PPC offer through desktop traffic?

02-19-2014 04:36 PM #2 caurmen (Administrator)

I'd cut this based on ROI - see Section 10 of the Getting Started Guide

In this case, given your stats I'd assume you've seen 56 clicks. Running that through the stats calculator linked in the guide, that gives you a maximum possible conversion rate of 6%, where you'd need a conversion rate of 16% to break even given those CPCs.

At that point, the ad ain't worth it - kill it.

02-19-2014 05:10 PM #3 kamaleon (Member)

@caurmen isn't this more complex than passing it by the stats calculator? The ad can be seen in many placements, and it can be profitable in one placement but not in another...

02-19-2014 06:16 PM #4 rfcapital (Member)

Quote Originally Posted by caurmen View Post
In this case, given your stats I'd assume you've seen 56 clicks. Running that through the stats calculator linked in the guide, that gives you a maximum possible conversion rate of 6%, where you'd need a conversion rate of 16% to break even given those CPCs.

At that point, the ad ain't worth it - kill it.
You were right on, I had 56-60 clicks depending on where I'm looking (Voluum or Decisive). I was missing the maximum possible conversion rate - I knew my conversion rate was stupid high for the campaign to actually work. It's a shame too, my CTR was pretty decent (compared to some of my desktop CTRs)

02-19-2014 06:55 PM #5 kamaleon (Member)

Quote Originally Posted by caurmen View Post
I'd cut this based on ROI - see Section 10 of the Getting Started Guide

In this case, given your stats I'd assume you've seen 56 clicks. Running that through the stats calculator linked in the guide, that gives you a maximum possible conversion rate of 6%, where you'd need a conversion rate of 16% to break even given those CPCs.

At that point, the ad ain't worth it - kill it.
@caurmen why are you taking the clicks into the stats calculator? You shouldn't take the impressions of the ad?

02-19-2014 08:08 PM #6 rfcapital (Member)

Quote Originally Posted by kamaleon View Post
@caurmen why are you taking the clicks into the stats calculator? You shouldn't take the impressions of the ad?
He's calculating two different numbers. Impressions give you the numbers you need to make money, clicks project the numbers you will get if you continue, given your current data.

If your impressions tell you that you need 16% to make money, and your clicks tell you that you'll be lucky to get 6%, time to cut the ad.

02-19-2014 08:29 PM #7 kamaleon (Member)

These are my results

02-20-2014 05:06 PM #8 caurmen (Administrator)

kamaleon - there are two different ways to look at the same data, either by cost of impressions or cost of clicks. Given rfcapital had listed cost of clicks, that was easier to calculate, so I went that way

You are absolutely right about placements, however - if those clicks were spread across a lot of different placements, that's a different story. rfcapital - were you running one placement there or a range of placements? Also, how many handsets were you seeing in the data?

02-20-2014 05:13 PM #9 rfcapital (Member)

Quote Originally Posted by caurmen View Post
kamaleon - there are two different ways to look at the same data, either by cost of impressions or cost of clicks. Given rfcapital had listed cost of clicks, that was easier to calculate, so I went that way

You are absolutely right about placements, however - if those clicks were spread across a lot of different placements, that's a different story. rfcapital - were you running one placement there or a range of placements? Also, how many handsets were you seeing in the data?
Running a range of placements, only real targeting was Android OS, I wanted to get some data before I tried to target.

02-22-2014 12:22 PM #10 caurmen (Administrator)

@rfcapital - Ah, right. In that case, I'd recommend giving it a bit more traffic and seeing if you start getting any conversions at all. Too many variables to rule the offer out on a $7 spend if you're still narrowing down handset, placement, etc.

02-24-2014 02:10 PM #11 kamaleon (Member)

@caurmen so mobile is too complex and too many variables to cut anything with the rules applied to adult or POF, correct? It is more of an instinct rule for cutting placements?

02-24-2014 04:30 PM #12 gosu22 (Member)

I'm running the same thing, with no targeting besides Android. No conversions also, but unsure of when to cut. 3x offer payout per ad applies? Otherwise I don't have info to cut without any conversions.

Treat the difference carrier and handsets as ads? Cut based on LP CTR?

02-25-2014 03:26 PM #13 caurmen (Administrator)

OK, I'm going to do a full tutorial on this in a little while. For now, I would recommend cutting placements first based on ROI, then start cutting ads.

Other mobile folks - I know this is still a rapidly developing area, so how do you tend to cut multi-placement campaigns?

02-26-2014 08:37 AM #14 Mr Green (Administrator)

I do cut out of instinct.

Every campaign I have run on a specific network, no matter what niche, I use that data in some form for my next campaign. So I'm only start blind on my first campaign I create for that traffic source. Which is usually merged over from another source.

Cutting at 3x the payout is a good starting bar.

@gosu how much have you spent, and what's the payout for this campaign? Is the offer proven?

03-09-2014 10:56 AM #15 kamaleon (Member)

@gosu22 @rfcapital This thread is dead guys? You get replies from admins and you disappear?

Home > Paid Traffic Sources > Mobile