ComradeWolf's PPC Journal

Thanks a lot guys, this thread and picture are full of gold!

My new test has almost finished (7 clicks left). I pulled 1 sale. This time I was testing a long (around 500 words) review LP (not on WP, just a page on a domain) in which I tried coming off very enthusiasistic and sincere. Basically the twist this time was that it involved a story of a user, not just a review (the inspiration from this is from Bencivenga's seminar, where he was talking about how a story-ad managed to beat a control ad).

Now I am going to test a new twist to LP: add value to the LP (I think it might beat the story-LP, if it doesn't I'll run a second test for the story-LP, the way I usually do, if there is at least 1 sale). I am not sure yet how exactly I will do this: either with an e-book or in some other way (because I think a lot of people will just d/l the e-book and start reading it, and this might ruin momentum).

I am also starting to test ads using the concepts from the picture in post #20 for my other group of KWs.
 


Comrade,

I'm going to give you some advice that I wish someone had given me when I was just starting out.

You need to ask yourself the question: how do I know when I'm done testing? That is, when can I be relatively certain that I know what the cost per conversion is for my current campaign? Here's how you find the answer using some basic statistics (for the sake of brevity, I'll just be giving you the formulas, not explaining where they come from).

Ultimately, what you're going to want to calculate is a 95% confidence interval around your conversion rate: this is a range around the conversion rate within which you can be 95% certain the true conversion rate lies. If the lower bound of the cost per conversion is above your offer payout, you can stop testing and go back to the drawing board. If the upper bound is below your offer payout, you're profitable and you can increase your ad spend.

Here's how you calculate the confidence interval. First of course, there's your conversion rate(CR), this is just:

CR=#Conversions/#Visitors

Then there's a quantity called the standard error (SE) (don't worry about where this comes from):

SE=Sqrt(CR(1-CR)/#Visitors)

where "Sqrt()" just means take the square root of everything in the parentheses.

Now that you have the standard error, you can use it to calculate the lower bound (LB) and upper bound (UP) of the 95% confidence interval.

LB = CR - 1.65*SE

Of course, the LB can't be less than zero, so if you get something negative you can just set it to zero.

UB = CR + 1.65*SE

And similarly, the UB can never be greater than 100% so if you get something more than 100%, just set the UB to 100%.

Here's an example:

Let's say you spend $1,135.00 to send 1243 visitors to a landing page, and 13 of them convert.

By these formulas, you have a conversion rate of 1.05%, and you can be 95% certain that the true rate is between 0.57% and 1.52%. This also gives you a cost per lead of $87.31 and you can be 95% sure that your cost per lead is going to end up somewhere between $60.00 and $160.26.

And there you have it. This gives you a disciplined way to figure out when you can stop testing. And if you're super fucking lazy, you can use this spreadsheet I made for you: https://docs.google.com/spreadsheet/ccc?key=0Av-uhxrFJuPtdE94QmRubzZMQ0p2NTFFRzQ4MWRzd0E&usp=sharing

You'll need to make a copy before you can edit it. Notice that the spreadsheet also has a field for the cost, and gives you a lower and upper bound on the cost per conversion. I won't bother explaining how I did that... an exercise for the reader as they say.

Anyway, good luck!
 
  • Like
Reactions: Staccs
Minderwinter, thanks a lot for the template and for spending time to write everything so thoroughly!

I am starting to use that right now, really helpful.
 
^I'm glad I took stats in high school. I used to ace the tests on confidence intervals.

Finally get a chance to use something I learned in school.
 
It takes ages to get those clicks. Probably people are going on a holiday and as a result they surf the Internet less often. Anyway, I'll keep waiting for 100 visitors.

How I implemented the concept of value I've talked about before: I am basically giving away free info relevant to the niche and product on my LP and then at the end I say that there are more secrets and tricks and if visitors want to learn them they must click on the link.

As a customer I'd view this advice as valuable because I don't just say "Work out more!". Instead I give thorough step-by-step instructions for around 2-3 weeks backed up by scientific studies and experience of other users. I literally explain what the user must do daily to get closer to the set goal.

But the test hasn't finished yet, so I can't really say whether visitors find this information valuable too. :)
 
What are you using for your landing pages? I'm looking into using wordpress and just curious if you or anyone recommends a good theme? I know that wordpress has a landing page plugin. Does anyone have any experience with that?

What would be ideal for me is to have wordpress site with 10 or so different pages that are all variations of the ad for split-testing.
 
Jaguar7 said:
What are you using for your landing pages? I'm looking into using wordpress and just curious if you or anyone recommends a good theme? I know that wordpress has a landing page plugin. Does anyone have any experience with that?

What would be ideal for me is to have wordpress site with 10 or so different pages that are all variations of the ad for split-testing.
Click to expand...

For my WP tests I was using the default theme because I wanted the blog to look like that of an average user, who is not that savvy. So I can't really give you proper advice.

But on one of the forums I read a case study by a guy who was using flexibilit 3 theme and was happy about its performance.
 
The test has finished. CTR to the salespage was pretty high - 34% (the largest I've got was around 50% on an LP of a different style, around a month ago). But there were no sales this time. By the way, I have noticed that if I get sales in a test it is always the sale made by some of the first 30-40 visitors. I am not sure if it's a coincidence or not.

Using Minderwinter's formulae and guidelines (thanks a lot!), I calculated that in my previous tests my cost per conversion is around $30 and statistically it will end up between $10 and $374. To improve this situation I can:

a) decrease ad expenses by increasing QS or bidding on cheaper words or bidding less on my current KWs so that my ads will end up having lower positions

b) increase the number of conversions to 4 (seems more feasible, even though the max I've had so far were 2 sales)

Also, the ads I was using for this test were targetting spontaneous/humanistic visitors (big thanks to CCarter and thehobbster!) in the surfing mode. The CTR of the ads was pretty high - 3,48%.

Next I am going to run one more test on the previous LP (since it turned out to be more effective that this one - by 1 sale, lol) and tweak it.

Though I will probably have to do it in around a week or two because I am going on a business trip to support my colleagues on a project overseas and I expect the work on it to be very time-consuming. But I will do my best to keep updating the thread and share my test results.
 
The project got suspended so I won't have to go anywhere.

So I signed up for a month of PPCC to get a different perspective on PPC (and partly as a source of inspiration).

May be I am not experienced enough yet to fully grasp its greatness, but right now it seems a bit weird. The guides (at least in the first month) focus only on email submits/downloads and a bit on list building. The worst part about them is that the are way too simplistic in my opinion: you pick an offer, then KWs, and then you get cash. I think the author skips some very important parts to make it look easier (or may be it is really easier when you go for email submits and related stuff). I guess this is done deliberately because some of those guides are later sold on WaFo.

So it's not that great (in my opinion) and I won't be subscribing for the next month, but at least this got me thinking that I should probably try out some other offers (like email submits and similar stuff) and niches when I am done with testing angles/tweaking my current LP.
 
ComradeWolf said:
For my WP tests I was using the default theme because I wanted the blog to look like that of an average user, who is not that savvy. So I can't really give you proper advice.

But on one of the forums I read a case study by a guy who was using flexibilit 3 theme and was happy about its performance.
Click to expand...

Oh I see, going for the blog feel. I'm wondering if that would convert better than just a sales page with some testimonials. I could see a blog being more trustworthy, but I guess its all about testing. Good luck.
 
Another day, another test. I've signed up with MB and Peerfly and now I am also testing a weight loss offer in addition to my clickbank offer. The weight loss offer looked very promising, until all of sudden I stopped getting clicks and impressions because of some problem with editorial guidelines on Bing.

So now it will take Bing guys around 2-3 days to settle the problem. Of course I will be calling them daily, because usually they are very slow when it comes to various requests, but I hope that the problem will be solved soon.


Jaguar7 said:
Oh I see, going for the blog feel. I'm wondering if that would convert better than just a sales page with some testimonials. I could see a blog being more trustworthy, but I guess its all about testing. Good luck.
Click to expand...

Yeah, try it, probably it will work for you.
Thanks!
 
All of my Bing campaigns have been suspended due to poor user LP experience: I am not getting any impressions or clicks, QS for campaigns dropped to 1/10. As a result, my CPC will surge (to approx. $1-1,5). I guess my account sparked huge interest because I started promoting a weight loss supplement (even though a lot of people are promoting it as well, but many of them use long LPs that look like fake review blogs or fake news blogs).

Now I must improve my LP user experience to get the score back up. Suprisingly enough, prior to that check by the Bing compliance guys, there haven't been any problems with user experience on any of my LPs (including the ones that were downgraded today) for around 6 months. This looks very suspicious, but I guess this would happen earlier or later so I am quite ok with that and not that pissed.

I will also try email/zip submits as soon as I am done with tweaking my LPs.
 
Out of curiosity I optimized only 2 LPs out of 4. Funnily enough, LPs which were not tweaked got to 6/10-8/10 quite soon, while optimized LPs only got to 3/10-4/10. I am not sure how Bing's score calculation algorithm works, but these results are fairly controversial.

I am split-testing new LPs and headlines. No results yet. I am not sticking no Bencivenga's stuff anymore. Right now I can't get it to work. Probably it works for some other people, for me it doesn't really do anything.

I will just stick to Cashvertising and will test Tested Advertising Methods which I am reading.
 
Another day, another test. 1 weight loss supplement sale and 1 fitness product sale. Starting a new split-test for headlines, will move on to CTAs/images afterwards, because new headlines don't really lead to significant improvements in results.

I think that I am quite close to the ideal LP which will work for these offers (or maybe I am just being delusional and trying to inspire myself). So I stay fairly persistent.
 
Looking to jump back into the paid game myself, who are you using for a host if you don't mind me asking?
 
The new test for the weight loss supplement didn't get any sales. I am now testing a more emotional LP that is a bit longer.

But I got 2 sales from the fitness Clickbank product. Each sale came from LPs with different images which I was split-testing. The test is still running, because I want to make sure I have enough data. Looks like this is my winning LP and now I just must spend some more time on testing a lot of variaitons for its elements to get more sales or break even.

Out of curiosity I have also launched a test for an e-mail submit on 7search. Let's see how it turns out.
 
You must log in or register to reply here.
Top Bottom
Back