Pop Campaigns - How a Simple Test Can Increase ROI Up to 25% (6)
Pop Campaigns - How a Simple Test Can Increase ROI Up to 25%
Hey STMer's!
It's been a minute since I have posted any new material, things have been super busy lately. However, I had shared this tip with one of my coaching students and figured some of you newer affiliates could learn from it as well.
This isn't going to be a long read and this type of test can be setup within a few minutes.
Does Choosing One CDN Over Another Really Matter?
Depending on your campaigns, which GEO's you are running in, the type of offers and overall traffic volume you are getting - choosing the right CDN for your campaign can increase your ROI and profits by up to 25%. Realistically, I commonly see a 10-15% ROI boost when but have seen it greater. The example below is results from yesterday on one of the campaigns I'm using to create a video case study.
Here Are the Results
Statistical Significance - Peak Conversions
Yes, I normally let it run to 90%+ probability but this is sooo close and I may not have the time later to post this thread haha.
What Did I Do?
This part is very simple...
1. I took my best landing page for this campaign, which was hosted on Rackspace CDN, and I uploaded it to AWS Cloudfront.
2. Added the lander I uploaded to Cloudfront into my
3. Let it run for 1 day and spend enough to determine accurate results using statistical significance.
Results were a 11.29% increase in ROI just from that simple split test, for one small campaign, running in one GEO on one source.
The profit difference between the two on this test was only $7.98... however... look at the bigger picture. If I was running in 5 GEOs with 5 campaigns per GEO on 5 traffic sources and a simple test like this yielded 5-25% increase in ROI per campaign, things start to add up quickly.
But, this micro test can definitely help those who are starting out with smaller budgets and want to maximize their first few campaigns to squeeze every penny they can out of it.
When To Use This Test?
From my experience this test gets the best results when you are targeting low payout offers, in large tier 2/3 GEO's. But the same logic will work for all campaigns assuming you can collect enough data.
So, what are you waiting for? Go test 
Cheers,
Andrew
Nice share Andrew!
Actually split-testing each campaign component will make room for further optimizations and add some extra margin
Hm, considering the fact that I'm hosting most of the stuff that requires a CDN on rackspace ... this makes me a bit worried I don't run much pops though, so no panic yet, but the next time I do, I will definitely test cloudfront too.
BTW: did you run a comparable test across several days? Last time I did something like this, the numbers used to differ from one day to the next and in the end, it all balanced out quite evenly.
Very interesting.
Mutuloo, I wouldn't be so worried. From what I understand, this test was done in one GEO only. And CDN providers are known to have different performance per GEO. So in this case, maybe AWS was better, but for another GEO, you could have completely different results. So if you want to optimize based on CDN, it's likely you need to do it individually for each GEO.
Alternatively, you can use a multi-CDN approach, but things will get quickly complex : https://www.maxcdn.com/blog/multi-cdns/
Originally Posted by benjabcn
Originally Posted by matuloo
I don't run much pops though, so no panic yet, but the next time I do, I will definitely test cloudfront too.BTW: did you run a comparable test across several days? Last time I did something like this, the numbers used to differ from one day to the next and in the end, it all balanced out quite evenly.
Great question! I will clarify that I've only done these tests for pop traffic and for campaigns than get under 100k visits per day the difference is usually quite small.
This particular test has only run for 1.25 days but in a GEO I've already thoroughly tested in the past. This mini case study could have been more thorough but I wasn't initially planning to post this. However, it does demonstrate accurate results from past, more thorough tests I've conducted. This test is part of a video case study that I'm building for my coaching students and purposed to be an example.
Some of my previous tests have been very thorough and for periods of time spanning from 3-10 days. There were times that the results balanced out as you mentioned but I can say for certain there were several cases with a very very clear winner. I've also had better performance by not using Rackspace or Cloudfront, but lesser known CDN's in certain geos. If I recall, in an old thread on STM that Caurmen posted (or someone I forget now) they shared this but I also use it for looking into new CDNs to test when doing high volume traffic in a geo: https://www.cedexis.com/get-the-data/country-report/
To give a little more insight on how I've benefited from this...
1. Focus on the 80/20 as you should always be doing. In this case, if you decided to test this do so on the campaigns doing the biggest part of your volume.
2. To really see the benefit of this your campaign should be getting atleast 250k visits per day and run it for a minimum of 3 days. I have previously tested this on campaigns doing 1-2 million visits per day and ran for 7 days, it certainly can result in a nice boost. Some of the most significant results for me have been in large geos with poor to fair internet connectivity.
3. Something else to keep in mind... even when a nice boost happens, if the 'winning' CDN has a more expensive pricing structure it may not be as significant of a boost as you think. (Example: 10% increase in ROI but the CDN fees increase and take 3-5% of that due to overall traffic volume and depending on how profitable your campaign is already).
4. This is not usually a test I recommend at the beginning of your optimization, this is a later stage optimization hack. Starting off with Rackspace or Cloudfront is more than enough to begin with, so get profitable first then consider testing this if the time/upside seems worth it.
5. There are certain GEOs that I have found specific CDN's to use that for one reason or another have consistently proven to have better performance than Rackspace or Cloudfront - I'm not overly techy so I can't explain why but it does. So I've noted which GEOs and which CDN's work best for them, usually referring to large population GEOs that I can buy hundreds of thousand or millions of visitors easily.
6. I originally got into testing this type of thing about a year ago when I read Emanuels case study on his utility app banner campaign, he switched over to Cloudfront and had big improvements. And I just like doing small fun tests from time to time.
Cheers,
Andrew
Home > Paid Traffic Sources > Mobile