Take what works and bottle it.

One of the key reasons that web analytics are so important is because they allow us to take a truly objective look at our websites and our Internet marketing campaigns. Still, however, I find a lot of companies repeating the same mistakes because they simply will not listen to the rational voice provided by their web analytics reports. As good of a marketer as you may be, your search marketing campaigns, online advertising, and your web content all stand to benefit from thinking quantitatively.


If you're looking to improve any of the elements listed above, my first suggestion is simple: Take your data and work backwards. Start with conversions. Figure out the key goal or key goals of your site, whether that's a purchase, a contact form or a newsletter signup, see who completed these processes and retrace their steps back as far as you can looking for patterns or trends. While you're at it, make sure you have any information on offline tactics for your timeframe handy to reference against. Working back upstream can allow you to identify the most common pages that converting visitors have seen which gives you a good indication of your most effective content.


If 54 percent of your visitors who made a purchase saw your About Us page, it would seem that a sizeable audience wants some assurance about who they're purchasing from before they pass along their credit card information. It might be wise to add more of this information throughout the site and perhaps add a call to action within the site that more proactively guides people to the About Us page. Going further, you can track conversions back to specific search engines, referrers, and even keywords. Notice that the majority of your ad spend goes to your Yahoo campaign, but accounts for none of your conversions? Are certain ad groups seeing extremely high click throughs, time on site and conversion rates? Take this information and run with it! Use related variations, expand successful campaigns onto new engines, but most importantly - take action. But as analytics guru Avinash Kaushik often says, correlation does not imply causality, so even when you make such changes, keep tracking! The data may be objective, but our interpretations of it are still subject to error.