April 16th, 2008

In very early January I decided to run an experiment where I’d put Adsense on this site for a few months, not do anything differently than I normally would, and see how much a personal site could earn.  Although I only planned on having ads on this site for January and February, they were so unobtrusive that I’ve still got them now in April.  I suppose that portion of the experiment worked — the ads didn’t bother me nor anyone else enough for anyone to complain about them.

My experiment was to only show ads to people that came from search engines, and those that have not commented on my site in a while, or ever (shame on you).  In fact, if you came from a search engine, the site would show you two sets of ads; one on the top of the post and one on the bottom.  This method of displaying ads very selectively, I think, has improved the overall statistics regarding my ads.  Which I’ll share with you now – in perhaps too great of detail for some of you.

My click through rate is at 0.33% and my average eCPM is $0.94 (though today it is inexplicably at nearly $3.50).  If I’m able to keep the click through rate fairly high, and continue to increase the traffic on my site, I’d stand to do pretty well for a personal blog within about a year.

Above is a graph showing my sites overall page views over a 1-year period, not counting my own hits to this site.  Obviously I wasn’t interested in these numbers until I began this experiment on January 3rd of this year, which I’ve marked with a 1. Since then the trend has noticeably angled in the desirable direction, which I’ve marked with a 2.  The number 3 on this chart is the current month, which is not fully calculated until the month is over, which is why it appears to have such a steep drop.

The only thing I’ve done differently on my blog since January is attempt to be fairly regular with writing, and as you may know I’ve been bringing all of my content to my site rather than spreading it all over, and I’ve tried to maintain a certain level of quality with everything I’ve published here.  Combining those three things is what I believe made my site’s traffic increase as it has.  My actual numbers are nothing to brag about, to be sure, but so long as the trend continues in the right direction you could say that the experiment was a resounding success.

But, was I able to accomplish my goals?  Yes.  Here is what I said my goal was in January:

“I don’t have any hard number goals that I’d like to achieve, but I wouldn’t mind being able to buy myself one lunch per month off of this site’s ads. Seems like a low goal? Remember, this is a personal Web site with a very small number of subscribers, less then five-hundred posts, and I only manage to write on here when I find the time. So how much does a good lunch cost nowadays?”

In the very first month of my running this experiment I made enough to buy a lunch.  In fact, based on the amount I paid today for lunch at Panera Bread, I’d be able to afford two lunches for the month of January’s earnings.  Success!

But wait, in February I would have only been able to buy a little more than one lunch.  Still a success, but odd that the amount I made in February wasn’t more than it was in January, since my overall page views were higher.  Here’s why; the number of clicks on ads that I had in February was only half that of January.  The main reason, I believe, this occurred was because of search referrals.  The main way I make money from ads on this site would be through search referrals, especially since I show two sets of ads on those page views (which probably brings down my CTR in retrospect).  If the traffic I get from search engines decreases over the course of the month so will my earnings.

Thrilling I know, but there’s more.

What about March?  March was, statistically, the best month this site has had in at least a year.  March was much better than February but still not quite as good as January.  While traffic from search engines did indeed increase, it didn’t increase enough to make up for the losses incurred in February, or even to match that of what was made in January.

Now we’re onto April.  You’ve probably painted a camel-like graph for yourself in your head with January and March being the humps and February being the valley in the middle.  So, where will April end up? With the month already half over, this site has made more than all of February, and almost as much as all of March.  I believe the month of April’s earnings will end up being slightly more than that of January.

What I’ve actually learned

Ads don’t have to be obtrusive to make money. The most important thing when serving ads is that you serve them to people that are most likely to click on them.  If you show an ad to 1,000 people that won’t click on it, you’ve just decreased your CPM average.  Context is also very important, but Google takes care of that for me.  However, if you’re reading this and in the middle of building your own ad system, know that properly serving an ad contextually is just as important as serving an ad to the right person.

I know what you’re thinking; this is advertising 101.  Not exactly.  I’m speaking of actually turning advertisements off if you think the person viewing the web site is not going to click the ads.  Do not even bother showing an ad to a person that will not click it.

And I’m not talking about demographics here.  I’m talking about scope.  If the person you are serving an ad to is there to see a very specific “thing”, and the ad is not well suited for that “thing”, you won’t get a click.  Plain and simple.  However, if the person you are serving an ad to is looking for something related to the ad, or is highly vested in something related to the ad, or even better yet is incredibly interested in something related to the ad – that’s when you’re going to get a click.

Here is the best example I can come up with on the fly.  If I’m reading an article on CNN.com, I don’t even notice the advertisements because I’m there to read the article.  However, if I’m looking for information, products, services, etc. that the article just happens to be about, and the ads are relevant, then the ads become part of the content.  The ads become valuable.

The second thing I learned is that having a lot of traffic doesn’t equate to more click-thrus.  February’s earnings dip is a perfect example.  Having a friend, or just about any other web site, link to you will not help the number of clicks you get on your ads.  If I was earning money based purely on CPM, or cost per thousand impressions, then it would, but not if I’m earning money solely on the CPC, or cost per click, model. (There are probably arguments to this, and I’d love to hear them.  The first thing that comes to mind is that the Long Tail will improve with Page Rank.)

Advertising will always be the most obtrusive when they are based only on number of impressions.  If I was earning money on CPM, I’d be showing an ad at every opportunity.  In fact, I’d probably put them right in the middle of my post.  This is why advertisements based on only on CPM should be replaced by poignant sponsorship campaigns that make sense.  More on this at another time.

New goals and some changes

I think I’m going to make some changes to the way I handle search engine referral page views.  As of right now I show an ad both at the top and the bottom of the post.  I think I might adjust it so that I only show an ad on the top of the post and not at the bottom.  This way I’m not showing two ads, which “fluffs” the impressions statistic, and lowers my CTR average.  I am not sure if this will help or hurt me, but I’m open to suggestions here if any of you have experience.

My new goal is that my earnings per month increases at a rate of at least 10% per month.  So if I make $1.00 this month, I want to make $1.10 next month.  I don’t even know if this is possible or plausible.  But it is a goal nonetheless, and a measure at which to weigh this experiment when I revisit it in the future.

If you have any comments, suggestions, tips, or tricks – please share them in the comments.  Hey, by commenting you don’t have to see the ads at all, so why not?