hayley 365


experiments in ad layouts

16 Jan 2014

2014-01-16 13:51:49 -0600

So today, I'm starting to see the effects of the layout changes to one of my sites, and I guess it pretty much confirms that there were accidental clicks occurring.

I'm pretty sure that you're not allowed to discuss specific CTRs, so I'll just say that the CTR is down enough that it's probably not just coincidence.

If you read my post yesterday, one of the things I was uncertain about was just how much whitespace to add. Given my current layout, the more whitespace I added, the smaller the actual content got, so I wanted to find a good balance.

Well, I can't know at this point if the changes I made are enough to appease the mighty google, but they definitely seem to have had a positive effect on accidental clicks without negatively affecting the main content too much.

But all of this led to the thought, "geesh, how many of the clicks today were accidental?" As someone who rarely ever clicks ads, I've often been shocked at how many ads get clicked. I mean, it's still a tiny percentage of pageviews where ads are clicked anyway, but apparently that number was a little inflated due to the layout being too cramped.

I had the realization that I might be able to get some data that would help in making a guess on how many ads were still accidental.

Basically, I have 2 equally sized ads right next to each other. Given my layout, if it's accidental, one of them is more likely to be clicked than the other. What I'll never be able to know is whether there was a more tempting ad in one spot versus the other, but I figure that over time, the two ad spots would probably be pretty equal in terms of temptability.

So what I realized that I could do was to create two separate ad units instead of reusing the same ad block twice. Then the plan is to see if there's a significant CTR difference between the two ad units. I'm not sure what would qualify as significant, but a large difference could definitely be indicative that there's an accidental click problem going on.

Apparently, I could've also accomplished this ad separation with custom channels, but I've never looked into them. I probably will in the future though because running experiments like this could prove very useful. I'm not sure if custom channels would make it any easier to run the experiments though.

Oh, another problem with trying to determine the accidental clicks is that, in my layout's case, the ad that would be more likely to be accidentally clicked is also the one that is more prominent and more likely to be noticed by a user.

I wonder if heatmap-type analytics packages are compatible with adsense. That would be the most obvious answer to determining whether there were accidental clicks occurring. I mean, if a heatmap showed that most of your ad clicks were occurring on the very edges of an ad, well there you go. Accidental clicks.

The preliminary results from these changes are pretty foreboding for the project in general. And yet another reason why I need to hurry up and diversify my projects away from adsense.

Specifically, I really want to get some direct-pay projects going. But I keep stalling for various reasons. Ooh, let's list them all:

I think that about sums it up for now.

Frankly, it's the technical stuff that worries me more than anything.

See, I'm expecting that my first direct-pay business idea will fail or fizzle. I'm expecting that it's going to take several ideas before I find the one that actually works.

And the technical stuff that scares me right now, will actually be roughly the same for any of these imaginary future projects. Basically, I figure a lot of this stuff out once and then I'll be set. Then, it will be all about the critical business part of finding out what people are willing to pay for and figuring out how to get in touch with those wonderful people.

The business thing is really the make or break part, but the technical stuff is what's got me the most nervous.

I'm also uncomfortable with the idea of spending money on something like adwords to actually test these ideas before I launch into them (especially if I'm going to have a lot less income to play with due to the adsense changes).

The plan is that I'm supposed to build one product that implements the entire ecosystem of what I need to figure out technically. If I'm blessed, it will also work as a business, but it's mostly supposed to serve as a learning tool for all of the technical stuff that's got me tripped up.

Hell, maybe the fix is to just solve the technical stuff in isolation. Like, just have some piddly website, a freaking "hello world" page even, but use it to figure out things like authentication and a payment system.

I should do that.

But will I?