ONA session: Understanding metrics
Oct. 2, 2009
By Clyde Bentley, Ph.D., 2009-2010 Donald W. Reynolds Fellow, Associate Professor, Missouri School of Journalism
Great stuff. Note how bad unique visitors are as measure, which corresponds with the Belden research I did with Greg Harmon several years ago.
(Rough notes, not verbatim.)
Dana Chin, USC
Because you can get all that data from the metrics programs, it doesn’t mean you have to use it.
- Basically there are three basic numbers, two ratios and one proportion
- Visits per unique visitor
- Page visits per visit
- The bounce rate
Unique visitors – not a good measure:
This is computers, not people. It is a very poor measure that can be easily over and under counted. If you have three computers (work, home and at a hotel) you are counted as three unique visitors. Conversely, if you have a computer used by multiple people (as in a school computer) it is counted as one visitor.
Advertisers are not buying on uniques and it is not a good measure of use.
Whenever you see “unique visitors” it should be accompanied by a time period. You can never add up unique visitors (July + August =). You must use the system to get the uniques for whole period.
News organizations should use the figure for THE WEEK, not the month. It will give you a lower number because it constrains the repeats. In this world of too much data, use the week. Go over them in your budget meeting.
Math of visits:
A visit is a period of activity separated by at least 30 minutes of inactivity. If a person goes to your site and leaves the computer then comes back later in the day, they are counted as two visits. But if leaves for 29 minutes, it is one visit.
Here are the good number.
- Visits per week: (number visits / weekly unique visitors)
Say that on average a visitor is coming 2.5 visits per week. If you are a primary information divider on a 24/7-day cycle, that is low. The goal is 7 visits per week. This is a great number to calculate for sections and microsites.
Also look at the daily number. A sports page should see a Saturday bump, etc. This is a good measure of “did your content work?” This is a very compelling number and a very easy number to get. Get the number from Omniture or Analytics and throw them into an Excel spreadsheet. - Page views per visit:
When your visitors come to your site, are they engaging in the content. - Bounce rate:
When one visit goes to the site and immediately leaves. Expressed as a proportion.
The issue with bounce rate is what is on the page. CNN reported in January that its bounce rate was above 50%. Very unhappy about it. If the bounce rate on the page is lots of headlines and you know people can get the information and leave quickly, you can count on a high bounce rate.
A bounce rate is not really bad if you want to just point them to information and let them go away as a service.
If your bounce rate is high, ask questions. What is it about a home page that is driving people away? Is it too crowded, are the search criteria giving people the wrong information? No information to bring them inside?
Number of bounces plus number of visis that started with home page and had 2-plus page views should equal 100% of visits.
When you see a bounce rate you don’t like, immediately dig deeper. This is like taking your temperature, an indication of something else that is going wrong.
Always remember that bounce rate is for a particular page, so look at the page where the most people enter. Usually the home page. Look at the content of the articles on the page. Compare their impact. However, Google reports an overall bounce rate they define as “Bounce rate is the percentage of single-page visits or visits in which the person left your site from the entrance (landing) page.”
Video metrics:
Very unclear measures right now that we are working on. For now, stick to basics.
Basic video, slide show metrics:
- Number of people who viewed it, by week, plus sharing number
- Videos viewed by visitor
- Percent of video that was viewed (a VERY important measure for advertisers)
- Engagement: comments, ratings. The percent who posted or rated it along with positive and negative influences.
Newsroom numbers vs. advertising numbers:
Newsroom is census data (100%). Advertising uses panel data, like Nielsen. Measures a group of people. Most census data is confidential, while the Nielsen and comScore information is public.
Social Media Rules:
Look for engagement over numbers. Audience, loyalty, influence and action. Metrics can help map the goals. Set your goals FIRST and then use the numbers to check them.
Social media is a constant call to action, so look for that action.
Twitter: Perfect tweet has a call to action, a link, a hashtag or keyword and a comment. This means that the perfect tweet is actually less than 120 characters. Especially because you must leave room for a reTweet.
Newsnumbers.com has a set of tools out there that work for measuring Twitter.
Topsy: is a search engine that tells you if you are part of the conversation.
MyTweeple and Tweetreach: look for influencers. Review your reach. For a high school football tweet, do you have the coach? Review the reach and following/follower ration.
Twittersheep: analyzes your follower profiles to assess their likelihood of engagement.
You can do popup surveys but remember you are not getting people who are not coming you your site.
Frankly, it’s your audience. Define your success by what your audience is doing and the members are thinking.
chinn@usc.edu
http://www.newsnumbers.com