Business Software

How to Analyze Your Website's Performance

How to Analyze Your Website's Performance
Building a website--whether you’re selling products or creating a blog or another information-based site--has always presented a challenge. You have complete control over your content, but virtually no control over who’s consuming it. Analytics tools have been part of the Web world for years, but many site owners haven’t fully explored them out of fear that they’re too difficult to use.

That was true in the 1990s, when the only way to get good data about your website was to use specially designed software to analyze your Web server’s log files, but that isn't the case anymore. Thanks to a handful of well-honed Web services, anyone with even a rudimentary understanding of the Web can use a Web-analytics tool to draw deep insights about their site’s visitors: Where do they live, when do they come to the site, how long do they spend there, and, critically, how many of them are coming every day?

The Web-analytics market, once vast and fragmented, is now very small. In fact, it has become largely consolidated between two companies: Google and Adobe, who together command about 90 percent of the analytics market (depending on how you count). IBM, Webtrends, and Yahoo all offer analytics tools, but none are major players in the market today. Yahoo recently announced that it would be shutting down its analytics service later this year.

For the purposes of this story, I’ll focus on the market leaders, Google Analytics and Adobe SiteCatalyst.

A Word of Caution

Before I dive into the tools themselves, remember: Analytics are just a snapshot of your website at a point in time (or a range of times). It’s important not to get sucked into the reports, where you can spend hours every day second-guessing yourself, trying to figure out why one page of content performed well and another performed badly.

Oftentimes this is a fool’s errand. There’s just too much randomness on the Web, and the power of social media and community-driven news sites such as Reddit can turn a quick late-night blog post into a completely unexpected hit. You can spend years trying to chase down repeats of those posts to re-create the magic, but that rarely works--and you may find that you’ve exhausted yourself building a lot of the same kinds of content, which no one ends up reading.

When all else fails, throw a funny cat photo onto your site.
Here’s a typical (and fanciful) scenario: The owner of a website/blog about animals finds that a picture of a cat sitting on a toy spaceship accounts for 20 percent of the past month’s traffic, and StumbleUpon is the major referrer, with most of the traffic coming in a two-day span. The site owner then tracks down 100 more pictures of cats on spaceships, dutifully posting them all, but none of them perform well.

Meanwhile, a deeper analysis of the top 100 posts would have shown that the site’s primary content--written advice about pet health, let’s say--makes up 50 percent of the past month’s traffic, and those posts are evergreen in nature, consistently generating traffic month after month, largely from Google. Long after the cat picture has been forgotten, these posts will continue to thrive. What the site owner should have done was continue to produce these pieces, building up more and more “long-tail” content that search engines--and users--find valuable. That doesn’t mean you have to ignore the fluffy “dessert” of the Web--because people always want funny cat pictures--but focusing your site on the meat-and-potatoes items will keep traffic growing and reduce frustration.

The bottom line: Unless you’re a veteran, use analytics as a tool to generate broad insights about your business and to tease out long-term traffic trends. Don’t focus too much on what happens from day to day. Knowing what happened last Monday is less valuable than knowing what tends to happen every Monday. Always use the longest time horizon you can when examining trends. Locate the points where the trend lines undergo a change (did traffic rise until a certain day, then plummet?) and figure out why it happened.

Analytics can tell you only the who, what, and when…and almost never the why.

Next Page: What the Metrics Mean

Subscribe to the Daily Downloads Newsletter

Comments