Home > Paid Traffic Sources > Adult Traffic (NSFW)

Culling Banners - What Would You Do? (9)

03-08-2014 08:03 AM #1 beastmode (Member)
Culling Banners - What Would You Do?

I recently ran a campaign:

1. 17 Different banner angle styles, 3 variations of each style.

2. I ran each of the banners to 5000 impressions or more. So if you add up each variation of the impressions of the same style, it'd be around 28,000 impressions. I think this should be statistically significant.

3. I was left with 3 banner angle styles that clearly beat the rest. Here were the 3 that won:

A - 28,000 impressions with a CTR of around 0.28% for the whole group (the average of each of the 3 banners from that angle)

B - 28,000 impressions with a CTR of around 0.29% for the whole group (the average of each of the 3 banners from that angle)

C - 28,000 impressions with a CTR of around 0.44% for the whole group (the average of each of the 3 banners from that angle)

My question would be:

1. Would you bother optimizing the A and B banners to make the CTR higher or would you just run with C angle and try and tweak the variations of that angle to get a higher CTR and scrap the rest?

2. Let's pretend that C is the best angle, I would then need to tweak my landing and offer page to match that angle as closely as possible right?

03-08-2014 08:16 AM #2 cmdeal (Veteran Member)

Quote Originally Posted by beastmode View Post
I recently ran a campaign:

1. 17 Different banner angle styles, 3 variations of each style.

2. I ran each of the banners to 5000 impressions or more. So if you add up each variation of the impressions of the same style, it'd be around 28,000 impressions. I think this should be statistically significant.

3. I was left with 3 banner angle styles that clearly beat the rest. Here were the 3 that won:

A - 28,000 impressions with a CTR of around 0.28% for the whole group (the average of each of the 3 banners from that angle)

B - 28,000 impressions with a CTR of around 0.29% for the whole group (the average of each of the 3 banners from that angle)

C - 28,000 impressions with a CTR of around 0.44% for the whole group (the average of each of the 3 banners from that angle)

My question would be:

1. Would you bother optimizing the A and B banners to make the CTR higher or would you just run with C angle and try and tweak the variations of that angle to get a higher CTR and scrap the rest?

2. Let's pretend that C is the best angle, I would then need to tweak my landing and offer page to match that angle as closely as possible right?

As for your first question ... in performance marketing, CTR is an intermediate metric, not the goal in itself (except in the rare instance that you are getting paid on a CPC basis). Of A, B, C you need to determine the profit that you have generated for each variation. You need to use a profit based metric to make the correct business decisions.

As for your second question ... yes. A big yes.

03-08-2014 10:57 AM #3 beastmode (Member)

Thanks CMDEAL

But I am testing a CPS offer, I am not sending enough overall clicks to the sales page to determine whether or not it's making me money.

I sent 500 clicks, but had 25% CTR on the landing page, so I sent like 125 clicks to the CPS offer. So it's not enough data to figure out what is profitable.

I just wanted to know WHICH banners would get me clicks. Once I can get clicks, then I'll focus on optimizing the sales page to fit that angle (I own the offer, so I have the flexibility to change the offer to make it suit whatever angle people like).

But thanks for answering, I just made 7 variations based off winning angle C to see which variation of that angle is the best, then I'll optimize my campaign further.

03-08-2014 01:06 PM #4 caurmen (Administrator)

Honestly, for a CPS deal I'd recommend trying to get a stronger sense of purchase intent than just a clickthrough. You're risking ending up with the classic high CTR, low CVR banner.

Can you run a simple email collection test? Set up a lander that captures emails of people interested in the product you'd be pushing them to, then fire traffic from each of the three banners to that lander. Track and see which banner style gives you the most email addresses - that's a much stronger signal of purchase intent.

This will give you a decent number of "conversions" for a much lower cost than running a full test on the offer, allowing you to see which banner is more likely to drive the right type of traffic.

Plus, you've then built an email list as a side-effect

03-08-2014 04:41 PM #5 beastmode (Member)

Quote Originally Posted by caurmen View Post
Honestly, for a CPS deal I'd recommend trying to get a stronger sense of purchase intent than just a clickthrough. You're risking ending up with the classic high CTR, low CVR banner.

Can you run a simple email collection test? Set up a lander that captures emails of people interested in the product you'd be pushing them to, then fire traffic from each of the three banners to that lander. Track and see which banner style gives you the most email addresses - that's a much stronger signal of purchase intent.

This will give you a decent number of "conversions" for a much lower cost than running a full test on the offer, allowing you to see which banner is more likely to drive the right type of traffic.

Plus, you've then built an email list as a side-effect
That's a smart idea, I'll do that!

03-08-2014 08:21 PM #6 zeno (Administrator)

For CPS offers it is great to have multiple pixels, e.g. like with Clickbank how you have order form impressions. Not possible with all offers/networks but it's useful to get data from the funnel not just the very end result.

Caurmen's idea is great too!

03-23-2014 11:29 AM #7 marcovandaar (Member)

Quote Originally Posted by zeno View Post
For CPS offers it is great to have multiple pixels, e.g. like with Clickbank how you have order form impressions. Not possible with all offers/networks but it's useful to get data from the funnel not just the very end result.

That's something i've been looking for but haven't found the perferct solution to do that... right now i'm using google analytics funnels

03-24-2014 02:51 PM #8 Finch (Moderator)

Ideally you would want to test each angle with the landing page already customised and then base your decision off ROI.

A quick hack to do this is to design a dynamic LP and pass two variables:

Headline ID and image ID.

You'd pass header1, header2, header3 and etc in the URL string. Then serve a different header based on the angle used in the banner. Header1 = "BBWs in xxx want to meet", Header2 = "Photos of someone you know.." etc etc.

Same for images.

If you want to instantly turbocharge your landing page CTR, pass an image ID in your URL and then use it to serve the same image that you used in your banner.

Consistency doubles down on a good angle.

03-25-2014 11:16 AM #9 dirtbag (Member)

Quote Originally Posted by Finch View Post
Ideally you would want to test each angle with the landing page already customised and then base your decision off ROI.

A quick hack to do this is to design a dynamic LP and pass two variables:

Headline ID and image ID.

You'd pass header1, header2, header3 and etc in the URL string. Then serve a different header based on the angle used in the banner. Header1 = "BBWs in xxx want to meet", Header2 = "Photos of someone you know.." etc etc.

Same for images.

If you want to instantly turbocharge your landing page CTR, pass an image ID in your URL and then use it to serve the same image that you used in your banner.

Consistency doubles down on a good angle.
Something I'll definitely have to give a try now that I have a good idea how to achieve it.

Home > Paid Traffic Sources > Adult Traffic (NSFW)