Analytical - web analytics for blogmasters

A Multiple Moving Average, or MMA, is a collection of Moving Averages (MAs), plotted together to simultaneously show short- and long-term trends. As with MAs, MMAs can be used to detect trends in any type of real-world data. They’re used in the stock market, but I use them to analyze my website and weblog metrics (traffic, revenue, etc.).

In my opinion, MMAs are more important than MAs. An MA graph just shows you a pattern over a single date range. MMAs show you multiple patterns over several time periods, long and short. If you know how to interpret MMA graphs, they’ll tell you far more about the current health of your blog(s) than just looking at your current site metrics.

I typically use a wide range of “windows” for my MMA forecasting: 7-day, 14d, 28d, then multiples of 28d. Because most human beings live on a 7-day cycle (week), it’s best to use multiples of 7d or 28d. You could use, say, 15d and multiples of 15d, but that would get far too complicated.

It’s important to remember, when plotting MMA multi-graphs, that the patterns showing are an indicator of what has gone by. But by carefully analyzing both short- and long-term patterns, you can estimate what may happen in the short-term and in the long-term.

As with the stock market, there’s no guarantee that your weblog(s) behaviour will continue in a positive direction. However, these MMAs will help you to realize that just because your traffic or revenue has been down for the past 14 days does not mean that your daily average revenue has not been increasing over the long-term.

Long-term patterns put the short-terms in perspective. As with all marketplaces, there are unusual cycles. Your daily traffic is affected by so many factors that it’s impossible to list them all. There may be a holiday in one place, or an important sporting event. Or if your traffic consists mostly of university students, maybe their semeseter just ended and they’ve gone home for summer break.

The two most important indicators of your website(s) health are the 2nd-longest and 3rd-longest windows (time periods) of data that you can produce an MA graph for. For example, on the day that I’m writing this article, I am one day away from the one-year anniversary of putting Google AdSense ads on my hubsite, Chameleon Integration. However, I do my forecasting for my collection of websites and weblogs, not just one site at a time. My longest window of time is 364d (12 x 28d). The second-longest is 336d, followed by 308d.

The MA graphs for each of the latter two windows show what resembles exponential increasing curves, or parts thereof. Somewhere on this page is an example image showing MMA graphs for several windows, all the way up to 252d. If you click on the image, you’ll see a larger version. The three right-most curves show signs of exponential growth.

As you keep adding MA curves for each new window of time (i.e., the next is 280d), you should keep seeing similar exponential curvature. (My example isn’t the best, as the long-term curves look almost linear. You have to look closely and extrapolate into the future.) If you do not see exponential growth for your long-term MAs, you may have a temporary fluctuation. You can either wait until the next window of time to see if a long-term downward trend continues, or you can take preventative action if you’re concerned.

By preventative action, I mean implementing short-term ways to increase your traffic, either by advertising - say Google AdWords - or building backlinks by intelligent, relevant commenting on the sites of colleagues, etc. Visit my BlogSpinner V3 and V2, as well as Darren Rowse’s Problogger for advice in that regard.

If you do continue to see exponential-like growth in long-term MAs, this is good, no matter what has been happening in the short-term. Most phenomena in nature exhibits some sort of exponential growth. It’s not permanent, but such growth is apparent for varying durations of time.

Think of your weblog/ website as a sort of organic entity, which will have a lifecycle in which its traffic/ revenue grows, on average, exponentially, and then eventually hits a point where it will decrease exponentially. The decrease will only be exhibited if you stop updating the site at some point in the future, or your topic of choice ceases to be of interest to a significant number of people around the world. On the other, who knows what the state of the Internet and human interest will be in 10 years.

(c) Copyright 2006-present, Raj Kumar Dash, http://www.mathgurusonline.com/

Technorati Tags: mathblog, math blog, multiple moving averages, rolling averages, forecasting, web metrics, web analytics