Home > Paid Traffic Sources > Mobile

---

CTR for landers on CDN all lower than VPS? (9)

---

I understood how speed is crucial, especially for mobile so I moved over to AWS CDN from my previously VPS hosted version.

Surprisingly, I've got 6 landers, and all 5 of them reported a lower CTR than my VPS hosted version.

Granted the clicks aren't A LOT for statistical significance, but there seems to be quite an obvious pattern here.


Any one has thoughts about this? Should I just revert back to my old VPS, or attempt another CDN?
(New Zealand geo)

---

You sent 5x as many clicks to the CDN as you did to the VPS, I'm no expert on statistics but that has to not be good. I know on my own camps, stats always get a bit worse when you get into higher volume.

Is the VPS located in in New Zealand?

---

It's based in Chicago

Yeah, I was doing a small test on my original landers, then I got my CDN up and transferred all the traffic there. I know from statistical significance these volume is way too little, but when all 5 results aligns together, I figured it holds a little more weight.

---

To make absolutely sure you are getting representative data, you need to run both CDN and VPS at the same time, so like a proper split test. If you collect the data during different hours of the day, it will be skewed by the simple fact that people behave differently as the day passes by. Same goes for different days of the week.

On a side note, its very well possible that you will see better results with your VPS over the CDN. CDN isnt a cure for all problems, there are situations where VPS will do a better job. But first run a proper split test before making the final decision.

---

Ive just did more testing, and realised that across all my landers, VPS seems to generate (100% identical results across over 5 pages on AWS CDN vs VPS)

My VPS is located in Chicago. My Geo is in Aus.

Will getting a new CDN solve this issue, or does this VPS just kick ass?

---

Hmm. If you're targeting AUS, try getting a VPS there and see what your results are from that.

Also, what's your tracker? That may be relevant.

---

Hey caurmen, it was AMAZING seeing you on stage last few days! You were awesome!

Well I'm not at the stage of purchasing new VPS/hosting for every geo I'm in, boot strapping a lot atm. But this really got my attention see how the results were similar across all my identical landers split tested, and the VPS wasn't even close to my geo.

Any setting in AWS I might've missed that could've caused this?

I'm using Voluum

---

Thank you! It was great to be there.

Hm. It might be that Cloudfront is giving very slow results for AUS for some reason.

I'd recommend adding some real user tracking with Pingdom to your landers. That'll give you an idea on speed and also help narrow down what isn't working specifically.

Also, as a first step - fpt.Pingdom.com has an option to test from Melbourne. Run your CDN and VPS landers through that and see what results you get. That may shed some light on the situation.

---

Based on my own experience, you will do well with the VPS, Im seeing similar results with my own tests, I will most likely do a case studdy on this soon. But I would NOT stop using the CDN if I was you.

I use both a VPS and a CDN - VPS for advanced stuff, CDN for the simple static LPs. On top of that, CDN is very cost effective, you dont have to worry about system resources and its simple to use.

I also have a few dedicated servers still , where I run some old sites, so those come to use at times and I dont really care much about where they are located as my own tests didnt prove it would be necesary Just make sure that whatever VPS you are using, it must be well optimized and responding as fast as possible.

---

Home > Paid Traffic Sources > Mobile