Testing 50,000 images/ads? (17)
Hello
If i test 50,000 images of mature women and give each image 100 impressions, will i find a high ctr ad?
How can I find high ctr images for adult dating?
Thanks in advance
Is this considered trolling?
I am serious, I am not trolling. I've been testing many ads x2 the payout for the last 2 months. But can't get my CTR for USA higher than 0.15%.
How do super affiliates get higher CTRs?
I have also tested my landing page a lot for weeks calculating statical significance, but the conversion rate is around the same 2.1% as before.
I'm no expert yet, but from what I've seen so far you might wanna test new banners in batches of 10 and cut according to Caurmens guide.. also, you might wanna test 1 thing at a time, like offers first.. then move to split testing landing pages or banners next.. but too many variables at once will be hard to interpret all the stats..
As for the 50,000 images, that would take tens of thousands of dollars to get any useful data.. I think if you're using banners you see around the web and you can't break .15% after testing 50-100 then it's not going to happen.
Have you tried using the STM Adult spy? Very useful for banner ideas
I have tested and found a decent landing page and a good offer. I have tested plenty of ads from the competition. It's wierd how the #1 ads on pornhub USA only get like 0.2% CTR. (I've tested them)
I see the only way to get the campaign profitable would be to increase ad ctr, or to increase the conversion rate.
But I am not sure what element to test now?
Test more ads at x2 the payout?
Is it possible to get ads with a conversion rate of 3% or 4% in adult dating?
I'm not sure bud I'm kinda in the same boat haha.. hopefully someone else will chime in soon.
damn 50,000 images.
I would test way more then 100 imps man, INSANE ads will slip through the cracks for sure, you're going to need more data to know and 50,000 ads tested is gonna be costly lol
You want high CTR?
Test weird shit, one of my most successful images was a girl I found bending over doing laundry lol, weird obscure images = crazy ctr, but crazy ctr does not mean profit.
Testing 50K images will cost you approx. $150K average on FB.
Do you have potential for such type of testing ( Money and human resource ). I am talking about broad testing. Recursive testing will cost 30x more.
For dating campaigns, you don't need to these many images. Start with 10-20 images. Learn, what kind of image will work. With experience, you will soon realize 5 out of your 10 images win.
Start with <20 images. Test, rinse and repeat.
That's insane.
You can test 100,000, the problem might be in your targeting not in your banners. There are so many variables controlled by the traffic source that it doesn't worth it. I'm in the banner optimization business for about 3 years but I haven't heard any to test so many ads.
I think you should focus more on the landers, you have much more space to play along, there are much more influential elements that can boost conversions and with multi-variate testing you'll find some good landers, especially if you have the budget.
Firstly - testing an image for 100 impressions will give you almost no useful data. You'll need a few thousand impressions at least depending on overall CTR to start narrowing down the best images.
And testing 50k images, quite apart from being prohibitively expensive, isn't a good way to iterate to what works. As davidwikes says, start with 10 or 20 very DIFFERENT images, learn what works, and experiment in batches from there.
How many impression does a banner need for gathering useful data? x2 the payout?
a few thousand will give you a good idea. At 5-10k it settles and is more confident your getting results that have had time to properly converge.
100imps is pretty useless. I've had 5% ctr and 0% ctr with imps that low, anything can happen. shoot for atleast 1-3k before making any choices imo/
if its really awful ctr after 1k i usually just axe it ...
Originally Posted by igv2014
Yep, as Zeno says, the requirement for data depends on two things: your expected CTR and your payout.
You can't cut on CTR until you've reached statistical significance to the point where you know the ad won't hit your expected CTR. Use the statistical significance calculators in Part 10 of the Getting Started Guide to figure that out - for an average target CTR of around 0.15%, you're looking for about 5,000 impressions to have enough data to cut ads performing at less than 0.1%, and more impressions for ads performing higher. For an ad that's looking for a CTR of above 1%, you can start cutting very low-performing ads around 1,000 impressions (cut things with 0.4% CTR or lower), and by 3,000 impressions you can cut things that are performing below 0.6% CTR.
You only cut on CTR to remove obvious no-hopers, though. After that, you want to cut on CVR, using the process I lay out in Part 10 of the guide. You can only really start on that at 2x payout (3x payout on mobile, minimum) in spend on each ad.
100 impression is no where near enough.
That being said try this...
-Categorize them into groups. (or hire someone to do so for you)
-Organize into 5-10 groups
-Pick 2 images from each
Optimize accordingly.
The person who clicks on a banner does so because of something that happens inside their brains. Everything he sees on the page is interpreted in his mind, and all kinds of crazy shit goes on up there! That means that it can be any combination of image, color, headline, background, etc. And this may also be different between target sites and placements as well. For instance, the background color behind the text on one of my banners did raise CTR from below 1% to over 2% on Pornhub. Why? No idea.
Home > Paid Traffic Sources > Adult Traffic (NSFW)