Cyde Weys Musings

Recently I’ve been playing around with Google AdSense ad formats and colors on my more popular blog Supreme Commander Talk. The results have actually been a bit counter to established wisdom, which says that while image and animated image ads are flashy and annoying, one should leave them enabled, because they have a higher click-through ratio and thus bring in more money. I found the opposite to be the case.

Two weeks ago, I modified Supreme Commander Talk’s ad-serving script to use two different ad formats, one that was all text and one that enabled image ads. It served up one of the two ad formats randomly on a page-by-page basis, so half of the blog’s visitors saw pure text ads and the other half saw the ad format that allowed image ads. The results were startling. The non-images format was earning nearly double on a per-page basis. I’ve since removed the image ads and have gone to text-only ads now. I think I may have a few explanations for these results.

My site was getting heavily inundated with low-value image ads. The leaderboard ad format that I’m running can display up to four text ads side by side, but only one image ad. So when an image was displayed it only gave one potentially relevant choice rather than four. All of the image ads were very low paid, so even if they did get clicked more often (which I doubt, though Google won’t provide data on this), the total revenue wasn’t good. Supreme Commander Talk is a hardcore gaming blog for a hardcore gaming audience, and most of the image ads ended up being ads for various massively multiplayer online role-playing games (MMORPGs), e.g. “Join Furcadia now!” or “Join Adventure Quest now!”, etc. These ads just aren’t worth much.

My audience is also a bad group to target ads to. Hardcore gamers are very adept at using the Internet, and they pretty much just mentally filter out ads and would almost never think of clicking on them. Combine this with all of the bad, low-paying, oftentimes irrelevant ads that gaming blogs seem to attract, and you see the problem I’m having over there. Text ads ended up being better because the image ads paid little, were annoying and subliminally filtered out by experienced web users, and were advertising online games much inferior to the game that is the subject of the blog, Supreme Commander. Now if my gaming blog was about various online RPGs, the ads would’ve fared better.

What’s very telling is that earnings from this blog and Supreme Commander Talk are roughly equal, even though SupComTalk has readership numbers at least five times as high. The relevance and profit-per-click of ads on the two blogs isn’t even remotely comparable. My next experiment on SupComTalk that I will report back on in another week or two is whether it’s better to use a clashing color scheme that stands out or one that blends into the page. Also, it’s important not to over-generalize my findings. Image ads may still be better than text ads overall, it’s just that for one particular subject matter they didn’t do well at all.

I’m still playing around with the AdSense on this site. It was taking up too much of the sidebar on the right so I’ve moved it up into the black empty space to the right of the header. I almost fear that it’s too visible there, though. Well, I’ll run it this way for awhile and see how the hits compare. I’m up to a whopping $6.99 in over a month, which doesn’t even pay the server’s electric costs. This is not a money-making endeavor for me, but rather, simply playing around with the AdSense. I have gotten a much better feel of how it works.

Oh, and I love how you need to accumulate $100 in your account before Google ever starts paying out. At my present rate, that’ll take over a year! I have electric bills to pay in the mean time.

See more of my posts on Google AdSense.

I would really like to know where in the world they get these numbers from. To be perfectly honest, when I typed my URL into the script, I was expecting it to tell me this blog was worth about $20. But $5,645.40?! Don’t be ridiculous.

So far I’ve made a grand total of $6.99 using Google AdSense in the better part of a month. That’s the only monetization I’m seeing out of this site (and it doesn’t even cover the server’s electric bill). That’s a far, far cry from the 4-figure value this blog is supposedly worth. How do I get my hands on that kind of money? Are there really people out there looking to buy up smallish blogs for thousands of dollars? The very notion is ludicrous.

If someone really does want to buy this blog for $5,645.40, though, I can’t really see myself refusing. By the way, Pharyngula is supposedly worth $1,212,631.92 by these “metrics”.

You may have noticed by now that the Google AdSense test I’m running in the navigation bar (to the right side of the screen) goes under the heading of adcruft. I figure I owe an explanation.

In the techy world, the suffix -cruft is used to denote content that is generally worthless, lacking in quality, selectively biased, or esoteric; basically, junk. Fancruft, for instance, would describe Wikipedia articles that are created by fanatical fans of some obscure television series. It’s stuff that nobody else really cares about. Wikipedia has lots of fancruft; in fact, it has over 100,000 articles on individual episodes of various television shows. So keep that in mind the next time someone is bragging about Wikipedia’s 1.6 million articles. The thing about Wikipedia’s cruft is that it’s so pervasive it’s just not worth trying to get rid of. If you polled the majority of the population they’d want it deleted, but start actually trying to delete it and the fancrufters show up in disproportionate numbers and bitterly contest it. So basically, it’s just not worth the trouble to delete this kind of stuff.

But I digress. Another kind of cruft would be adcruft, which I would argue, applies to all advertising, and thus the words ad and adcruft are redundant. Anyway, the label is just there on the navigation bar to delineate it from the content that I consider worthwhile (you know, the actual content of the blog). It also serves the purpose of letting people running ad-blocking software know that there’s something there that’s not being displayed. I’ve been using the AdBlock extension for Firefox for a good while now, and there’s lots of sites out there that I don’t even realize are running ads because they are removed so completely. Even if the ads aren’t being displayed, I generally like to know which sites are running them and which aren’t, hence why I’ve made it obvious on my blog.

By the way, the term cruft most likely originates from Cruft Laboratory at Harvard. The laboratory has been used for heavy research in the past, and so it was always throwing away obselete equipment that had become worthless. As Wikipedia says, “By the whimsical humor of the student body, if the place filled with useless machinery is called Cruft Hall, the machinery itself must be cruft.” I should point out that I actually saw Cruft Laboratory at Harvard during Wikimania 2006. It was interesting to see the physical roots of a word. Anyway, one of us took this picture.

My little experiment with Google AdSense is entering the close of its second week. So far it’s been fun and educational, but I’ve made very very little money. Just $0.02 off of site-targeted ad impressions, to be exact.

See, the way Google AdSense works is that the majority of ads (or the majority of ads a small site will see) are “contextual”, meaning that they are determined automatically based on the content on the page. They only pay on a per-click basis, so a million people could see the ad, but if no one clicks on it, you don’t make any money. That’s the problem I’ve been facing.

Site-targetted ads are advertisments that are chosen by advertisers to run on a list of sites. Generally, smaller sites will end up being included in huge robotically-determined lists of hundreds of other sites, and advertisers will choose portions of these lists to run site-targetted ads on. Large sites that run AdSense end up getting targetted individually rather than in a group.

Most of ads displayed here on Cyde Weys Musings end up being contextual ads, because this site is small enough that it doesn’t get included in too many site-targeted advertising campaigns. Which is too bad. I’d like to see some more of the targeted ones. Not only do they actually pay me (a few pennies), but they’re more interesting to read, as they’re generally more related to the site’s material.

See more of my posts on Google AdSense.

Update: Okay, I lied. It’s up to $0.04 now. Apparently Google is actually keeping track of those fractions of a cent and not rounding them down, so even though only two days in my AdSense report are listed at 1 cent, all of the other 0 cent days add up to another 2 cents between them. At this rate, I’ll get rich twice as quickly!

I must say, I’m impressed with Google’s application of their company’s maxim “Do no evil” to the rough-and-tumble world of online advertising. Google requires valid tax information before they will send money earned through Google’s ad systems, and they are also prepared to withhold as much as necessary if the user has previously had a problem with the tax man. Google seems committed to preventing online tax fraud, which tends to run rampant in the virtual realm.

Let’s compare this with Second Life which, as far as I know, doesn’t require any tax information, even though people are supposedly quitting their real jobs and making a living solely in Second Life. What are the chances that all of these people are honestly reporting all of their (very real) virtual income? I’m willing to bet that the majority of money withdrawn from the Second Life game economy in the form of real currency ends up going unreported. And since these online companies can’t seemingly be trusted to regulate themselves (like Google does), it would seem like government regulation isn’t too far away. Nothing gets Uncle Sam more pissed off than people not paying their taxes. I suspect that, fairly soon, any online game that sanctions real-money trading (RMT) is going to be required to deal with tax issues, just like any real-life business that pays out money to workers.

By the way, I’ve been experimenting a little bit with the Google link box advertisement on this site. It looks better than the Google AdSense as it isn’t doesn’t appear so “commercial”, but then again, it doesn’t pay per impression. It only pays when someone clicks through to one of the links and then clicks on an ad from there. But still, it’s interesting. For instance, I can put it in the admin interface and see what Google thinks are the five most relevant phrases for my content. These make excellent primers for choosing tags to put on the post.

I could see link boxes being more readily accepted at Wikipedia than Google AdSense. People already often go to Wikipedia to find out information about a subject, so adding a few links to commercial resources would just provide added utility, and make WMF some money to help keep the servers running.

As for the earnings from Google AdSense impressions, so far I’ve made a big fat $0.00. I have made a tiny bit from two people that clicked on links. It’s starting to appear that the majority of the money people earn through AdSense is from ads being clicked, not from ads being viewed.

See more of my posts on Google AdSense.

Update: I think it’s best that I stop discussing AdSense for a little while, because every time I do, AdSense gets all meta on me and starts only displaying ads on the front page about AdSense and AdSense-related stuff. This is most definitely not A Good Thing. I think it’s some kind of Google self-defense mechanism. If Google AdSense is mentioned in any of the content, much unequal weight is given to that, probably as some sort of a defense against people trying to get attention to their ads by mentioning them.

So Google AdSense has been up and running on this site for about a day now (I very recently got it working, after having some trouble with the Adblock Firefox extension). Here are my impressions.

Google doesn’t really seem to be able to get the best ads for this site. Yes, they are largely relevant, but the artificial intelligence trying to correlate page content with ad content isn’t really up to snuff. Often it won’t be able to pick a suitable ad at all, leaving a black space. This shouldn’t be happening. Also, Google AdSense is way too narrow. It chooses an ad based on the content of that page only, regardless of what is on the rest of the site. So the single page views of some of my smaller blog entries just don’t have anything relevant on them (here’s an example). It should have some memory of content across pages, and tailor the ads on each pages to the content of the site as a whole. As it is, it just takes little snapshots, treating each page as if it were a completely independent site, and it misses the mark a lot of the time.

So far the earnings have been abysmal, but I’m okay with that, because I didn’t start this experiment seriously trying to make money. Also, the number of impressions Google is reporting isn’t squaring up at all with the number of visitors my Apache logs are telling me I’m getting. Are that many of my visitors really running ad-blocking software? Then again, I am only running one ad box, and the smallest one available at that. I’ll try switching this stuff up a bit soon, and seeing how the number of impressions AdSense reports changes by. AdSense also has a nifty thing called a link box, which doesn’t display ads directly, but rather, takes you to a page on Google listing a bunch of ads relevant to a particular category of content it saw on your page. That’ll be fun to play around with.

I’ve been playing around with AdSense at the suggestion of some of my friends from Wikipedia. We were throwing around the idea of running AdSense ads on the Wikimedia Foundation’s projects and we realized none of us had any first-hand experience, so I volunteered to use my blog as a guinea pig to report back some preliminary findings. As far as we can tell, and we think we did the calculations accurately, the Wikimedia Foundation is passing up somewhere between several hundred thousand and one million dollars per day by not running advertising. This is a colossal amount of money, and it could be used to do so much good in the free content community. Up until now funding has been mostly worked out, but with the failure of the recent winter fundraiser, which came in several hundred thousand short of the $1.5 million goal, things are looking slightly iffy. If worse comes to worst and hosting and salary bills become unpayable, you can be sure that the Foundation will put up advertising for a few days to cover the cost rather than risk going under.

I still find it sad that we could make as much money in two or three days with advertising as we did over the course of a month-long fund raiser.

I previously wrote that I’d be trying out AdSense soon. Well, that time has come. It’s not as easy as one would hope, and not for technical reasons. I have to enter tax and payment information before I can start receiving payments. Yuck. Although this probably won’t ever get in my way, as Google only sends a check once the account balance reaches $100, and I really don’t foresee that happening. I don’t have the patience to put up with, what?, a year of ads before I receive any monetary benefit out of it. So I’m approaching this strictly as an experiment on using Google AdSense, and I don’t have high hopes of ever seeing any money out of it.

Google has a walk through that details how to set up and start using an AdSense account. Unfortunately, it’s a little bit annoying to listen to. The woman speaking it has a slight lisp, which is most unfortunate because she very frequently has to say words like “AdSense” and “ads”. Also, it sounds like she’s speaking a bit too close to the microphone; there’s the occasional breathy puff of air.

The actual setup of making ads is easy. Google has a wizard that makes the appropriate HTML code given a step-by-step set of directions. For instance, do you want to make a link box or an ad box? Do you want ads with text and images, just images, or just text? Etc. Everyone should be able to figure this stuff out. For the record, you’re limited to three ad boxes and one link box per page. But I can’t imagine putting that many ads up. Google also has a handy page listing all of the various available ad formats. I’ve chosen the basic (and small) 120×240 vertical banner of text ads, which has two ads in it. I don’t really want to clutter up my site too much, but I’m limiting those payments I’m never going to see anyway. Later on I’ll experiment with different sizes of banner ads and report back on the payment differentials between them.

So I’ve added the ad code into this site, but Google AdSense is saying it could take up to 48 hours to actually start displaying. Grrr. I guess the server cluster that serves up the ads needs to be synchronized with the server cluster that handles account management, and this doesn’t happen instantly. But hopefully they’ll start displaying soon and I can start seeing what kind of hits Google thinks are relevant to the content of this site.

The next post in the series.

Update: The ads are working. I just wasn’t seeing them because I had AdBlock turned on (d’oh).