Expert: Tim Flight, GPSReview.net
#6 of 30
About the Expert
Tim Flight is the owner of GPSReview.net, a blog devoted to reviewing GPS devices and providing the latest news about them. Tim is a website developer, blogger, mac guru, and an aviation guy from Carrabassett Valley, Maine. Tim is a passionate, devoted and influential authority on GPS devices. One of the cornerstone’s of this blogger’s success has been his ability to crunch numbers and do analysis as to what works and what doesn’t on his blog. In his guest blog post, Tim, in his usual fashion, enlights and informs readers on doing some number crunching for success on their own blogs.
There are two types of people in this world, those who enjoy crunching numbers, and those who don’t. Regardless of the camp you reside in, crunching numbers is an important part of success. To effectively use crunch data you need to understand exactly what the numbers represent, and then plan for what you can do with the data.. Try to only track data that is actionable. It might be fun to track what day of the week you have the highest CTR, but is there much you can do about it? Would you really make a change to your site on certain days of the week to change the CTR on those days? Tracking only actionable data can apply both to the type of data collected and the frequency which you track that data. Is it worth it to track CPC on a daily basis? Probably not. CPC fluctuations from many of the most popular advertising systems are common these days. Therefore looking at CTR on a daily basis is probably too frequent, whereas looking at it from month to month will have much more meaning. Do you track page views on a daily basis? Is it actionable? If your page views drop one day by 15% over the previous day, is there really much you can do the very next day to increase page views? Maybe, but those changes wouldn’t likely be something that would remain long-term. Therefore it might not make sense to track page views on a monthly basis. Are there changes you could make that might increase page views on a monthly basis, looking more long term? This is far more likely. For me, beyond total revenue the most important item to track is revenue per unique visitor. This gives me the most consistent figure that weeds out “exceptions”. Some days you might get ads that result in a high CTR but low CPC, but the revenue per unique visitor would remain relatively the same.
Let’s say you have a graph of total revenue for each day. One day you get a news scoop and have a sudden surge in traffic. If you just graph revenue, you will have a spike in your graph for that day. When looking back at that graph a week later you will want to explain that spike, and you will need to look at other data to figure out why. However a graph of revenue per unique visitor will probably not show a spike for that day. Therefore this figure becomes a better figure to gauge the daily “health” of your site. Another example is tracking eCPM. You decide one day to double the amount of advertisements on each page. Even if you slightly increase your revenue, your eCPM will go down! Therefore comparing eCPM before and after you change the number of advertisements on a page is almost meaningless. However if the change was effective you would notice a change in revenue per unique visitor. When looking at the different type of data you track, make sure you understand the differences between data that can receive growth over time as well as data that should have a more level trend. For example we all strive to constantly build traffic. Therefore we hope the number of daily unique visitors continues to grow. When graphed, we expect that line to trend up. On the other side of the spectrum there are figures like CPC and CTR. It is nice when these figures grow, but that is not the expectation. Doubling your traffic will not double your CTR. Therefore when looking at graphs of data, try to keep figures that are expected to change over time together, and keep figures that are expected to remain level over time together. As practice exercises, frequently take a look at your data, spot some idiosyncrasies, and try to figure out why the data changed at that point. You do this by understanding what makes up that data point. Did your eCPM drop? Then you either added to the number of ads on the page, your CTR dropped, or your CPC dropped. If you added to the number of ads on the page, then there is the explanation. If the CTR dropped, check to see if you recently altered any ad placements or styles. If your CPC dropped… well, not much you can do about that except experiment with a different type of ad.
Another important tip to remember is that bigger isn’t always better. Do you always want to be increasing your CTR (click through rate)? Probably not. I know of steps I could take today which would double my CTR tomorrow, but they would probably not be healthy to the long-term success of my site. Often, I’ve found myself wanting to decrease a figure like CTR. Most people generally want people coming back to their sites. If you over-optimize a site then people might never see what a great site you have. You might make some short-term money, your long-term prognosis for growth probably isn’t that good. Finally, you have to apply what you know about your site when interpreting the data. For example let’s say you have one site that averages 1.5 page views per unique visitor and another site that averages 5 page views per unique visitor? Which site is doing a better job? You can’t really tell. It could be that the people on the first site are getting deep linked into the site, finding exactly what they want, and then moving on. Or it could be that they were turned off by the site and moved on. With the second site it could be that your internal navigation is a mess and people are frequently clicking around trying to find what they came there fore. So analysis of data isn’t always a left-brained activity. You will need to use your intuition to understand the data. Tracking data can be lots of fun, and it can yield lots of critical information about your site. However in order not to become too obsessed with stats and to allow more time for writing content, try to only track data that is actionable, and produces useful information.