Redesigned Site

CTR plummeted. change it back?

[smoger]

Hey all,..

I recently(2 days ago) launched a much better looking layout for my site. all content is the same, just looks a lot nicer, and my CTR the past 2 days has been horrible.

should i just change it back and face the reality that my site has to look like crap in order to make a few bucks... or is it worth it to wait and see what happens? surely someone else must have gone through this at some point..

[WolfLover]

Hello there. I did not do a site redesign, however, my site would look so much better if I could use the 728x90 at the top OR the AdLinks 728x15 that looks like navigation or both.

Every time I've tried that, and it looks beautiful on my site, the eCPM and CTR goes straight down. However, when I have been using the wide skyscraper, which looks ok, I've had a better CTR and eCPM.

I tried this experiment last night at midnight G time. I added a 336x280 large rectangle at the top (which moved my navigation down on both sides, and I am making about double eCPM today and nearly double CTR.

I have NO idea why this is. I would love for my site to have a cleaner look, but it does not make nearly as much from AdSense when it does. If it looks "like crap", my visitors click, makes no sense to me.

[gamiziuk]

I suggest you give it at least a week (7 days) before tinkering with it again. I find my traffic varies from day-to-day of the week. Fri-Sat-Sun are usually my worst days of the week.

[leadegroot]

Yes, try to get a good solid sample to make decisions on - I try to get at least a 1000 page views before I decide.
(Mind you, I always split test any changes anyway - this minimises the issues of 'has my traffic just coincidentally changed for some non-related reason?')

[inbound]

1000 page views is not a 'good solid sample' in my opinion. It would only be so if you have a very high CTR to start with.

A week makes sense as it allows for the daily fluctuations.

However, I agree that split testing is a good idea. This allows you to compare performance directly but it does present technical issues if you have not done it before (separating the channels, tracking the user sessions for consistency by individual user etc).

[UserFriendly]

By split-testing, do you mean use the ugly stylesheet and the smart stylesheet at the same time, serving one or the other randomly, and then checking the data to see which ended up most successful after 1,000 page impressions?

[maximillianos]

Another point to consider. Did your redesign change any of the URL names or parameters being passed to the pages? If so, G will consider them new pages and it may take a while before they start showing relevent ads again.