Spot My Mistake in This TrafficJunky Campaign

I made a mistake this morning.

Here’s the data from a TrafficJunky campaign I set up last night:

Optimisation process

The campaign used CPM bidding.

If you had to pick three ads, which would you keep?

I picked ad1, ad3 and ad4.

One of them was a bad choice.

Here’s why.

Affiliates are trained to look for the best conversion rate.

That’s pretty smart if you’re bidding CPC (paying by the click).

But when bidding CPM, the best conversion rate doesn’t always mean the best ROI.

That’s because it doesn’t take in to account the clickthrough rate of your ads. Less clicks means your conversion rate is taken from a smaller sample, and it therefore has to work a lot harder.

Instead you should be picking the ads that generated the most revenue: ad1, ad3 and ad6.

In this case, ad4 had the second best conversion rate at 3.56% but it had 3 less conversions than ad6 at 3.19%.

While ad4 was more potent for the users who chose to click it, ad6 delivered more revenue through attracting more clicks and maintaining a respectable if lesser conversion rate.

This is something to keep in mind with any CPM campaign.

We are constantly battling for the best combination of a banner that draws conversions, but also enough clicks.

So here’s the next step.

I’ll take the headline and ad copy from ad4 and combine it with the imagery and CTA of ad6.

It’s a process I documented here that has a useful knack of producing banners that increase profit and obliterate the competition.

About the author


A 29 year old high school dropout (slash academic failure) who sold his soul to make money from the Internet. This blog follows the successes, fuck-ups and ball gags of my career in affiliate marketing.


Leave a comment
  • “Instead you should be picking the ads that generated the most revenue: ad1, ad3 and ad6.”

    Should you be taking into account cost as well, so only pick the ads that generated the most profit?

    In the page you linked to in the above post, it mentions:

    “The most important factor isn’t the banner or the LP. It’s the angle itself. Some angles sell, others flop.”

    So, the banner should flow seamlessly to the LP using the same theme/angle?

  • Yep, you should definitely be taking cost in to account. This campaign was freshly launched so each ad was given an equal distribution of impressions. Each ad cost the same.

    As for angles, yes. If your angle doesn’t match up to the copy on the LP, the results are going to be skewed. I use a dynamic LP and pass my angle’s headline/image via the URL string. Saves a lot of time. I will throw up another post about that.

  • If you’re paying for impressions, aren’t you most interested in Revenue/View?

  • Yeah, sorry, to clear up any confusion, the ‘views’ column in the table above refers to ‘clicks’ not impressions. The clicks column refers to landing page clicks. That’s just the way CPVLab titles them for some reason.

  • Hello Finch,

    I’m not really good at math. And I don’t understand why the clickthrough-rate is not calculated with CPM bidding, whilst it is with cpc bidding. Do you mind clearing this up for me and maybe other less math-savvy readers? 🙂 thanks

  • It is calculated, but you only see it at the traffic source end (in TrafficJunky). Which is where my mistake came from. It’s easy to look at the data in CPVLab/iMobi/T202 and pick the best converting ads from a %, but the data has to be cross referenced. Often the best converting ad is not the best performing ad, precisely because CTR is critical to CPM campaigns.

    If this were a CPC campaign, the best 3 ads would have been those I chose originally.

    Because of the CPM nature of the campaign, that wasn’t the case.

  • Finch, smart! A little bit of testing goes a long way. Thanks for sharing the detailed breakdown, and cute add by the way 😉

    Tweeted to my 27K followers!

  • I just wanna know when the next batch of premium posts are coming up? Seriously Finch, shut the heck up and take my money!, you had me at balls lol

  • Even 1 more (or less) conversion for specific ad might change the winners you need to pick,

    Also maybe ad1 was getting most of the impression on specific site, and ad2 got most of the impression on another site, which can change the results, did you looked into it also?

    So you still need to gather more data before deciding where to move am I right?

    For that example how do you decide when enough data is enough?

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>

Copyright © 2009-.