Live event blog: Crash course in web analytics

By Cole Kennedy

Ready to follow along with Brad Best's talk on web analytics? There's still time to swing by RJI 100A (the Palmer Room) and sit in!

We're off! Brad is telling us a little bit about himself. He started his career in advertising and made his way over to strategic communication. He credits his interest in analytics to his use of statistics in fantasy football and baseball. He realized that the same sort of statistical principles applied to websites.

First question to ask yourself, no matter what the website is: "why does your website exist?"
  • Step 1: Get "qualified" people to your website. Do you have a target audience?
  • Step 2: Get them to take the desired action. What do you want visitors to do?
  • Step 3: Get them to come back. Did you meet or exceed expectations?
The first news analyst: the paper boy on the corner of the street. Where did he go to sell the news, and why? He went to busy street corners, because that's where the foot traffic was. He was analyzing information to determine how to maximize his sales.

  • Impressions: the number of views on an ad.
  • CPM: cost per thousand (the M is a Roman numeral) impressions.
  • Unique impressions: measured over a given period of time; 'uniques over six months.' News organizations might be more interested in people who are return visitors.
  • Visitor: doesn't account for how many pages a person views; they simply come to the website.
  • Visit: refreshes after 30 minutes, so a person can visit multiple times per day.
  • Bounce Rate: rate of visitors who come to one page, and immediately leave. A high bounce rate will lead to a low visit duration, sometimes artificially.
  • CPL: cost per lead; an alternative way for websites to charge advertisers for more comprehensive engagement, like sign-up forms.
  • CPA: cost per action; a generic term describing whatever action a website specifies.
  • CPW: cost per whatever; what Best likes to use because companies tend to have their own terms for measurements.

Websites need to measure engagement because you can't assume that people will just come to your website and buy stuff or read stories. Measurement and analytics can help websites learn if they are meeting their audience's expectations, and learn which types of stories work and keep visitors coming back. It can also help news organizations understand how people consume the news, whether it's mobile, or web, or app. This kind of understanding can increase ad revenue and subscriptions, and help foster a more informed citizenry.

User experience is king! Engagement metrics can help reveal how the UX is working.

Though web analytics might not always be 100% accurate due to sharing computers and other situations. Compared to Nielsen metrics or newspaper circulation, it can be very accurate. Scale of the audience helps to make sure insights are valid.

Web analytics offer the ability to test. A/B testing allows websites to direct different proportions of their traffic to different versions of the website to understand what is working, and what is not. Multivariate testing allows web designers to test multiple elements on a single webpage at a single time, but it's a very complex process.

After figuring out business objectives, the next step is determining key performance indicators (KPIs). Within the organization, stakeholders need to be found, so that certain people can be kept aware of website performance. Find a champion for your cause; discover the person within the organization to help make changes to improve website performance.

Most importantly: web analytics are a constant learning process! Data should inform your journalism, not totally control it. Don't chase page views, chase user experience!

Some web analytics resources:
Video of the event: 

No comments:

Post a Comment