Measuring Online Success in a Rapidly Changing Online Landscape
MW2009 Afternoon Workshop
15 Apr 2009 13:30
At Powerhouse Museum in Sydney, Australia, trying to think about how to measure web visitors.
Probably everything we are using to measure results incorrectly.
1-6 million hits per month are probably not realistic.
We don't know much about who they are or whether they translate into visits.
There is no correct number to the number of people visiting our site though many will try to sell products.
Segment audience.
Powerhouse Museum Children's Website: Play at the Powerhouse Museum
http://play.powerhousemuseum.com
Flash games -- how much time
PDFs -- who downloads them, how many
Log analysis tools often have a hard time distinguishing between spiders and real hits.
Google Analytics Dashboard can show 1 month or 3 month periods. Number of web visits, page views, pages/visit, bounce rate, avg time on site, % new visits.
Can show numbers by day, week, etc. Can limit by locale (eg Australia traffic).
GA has new Advanced Segments feature. See help.
Can break down by city (eg Sydney, Melbourne, Perth, London). Nearer ones show more pages per visit.
True Time on Site.
Bounce rate -- visit one page and leave. 0 time.
A number of Grease Monkey scripts to help with tracking. Works with FF.
makedo page was popular but not locally.
Google Analytics takes a fair amount of configuration to get it tuned well.
Melbourne (MuseumVictoria.com.au)
Top content area has some issues. #2 and #3 are "melbournemuseum" and "MelbourneMuseum". Avoid aggregation with URL rewriting.
Why doesn't GA have a tool to pick two or three entries and aggregate them? It would be easy for them to accomplish.
Don't obsess about accuracy but rather what can be gleaned from it.
Good maps and segmentation opportunities.
Can drill down to specific topics and view pages as well.
Rare to have most traffic at home page. Most other museums have 5-8%.
If you find that people arrive deep in the site, you can change the design to get them to view more of the site.
Filter local views (ie museum IP).
Top landing pages shows first page viewed.
Questions to ask:
What are the key intentions (audience, purpose) for the visitors to your sites?
What constitutes a conversion?
How can you improve the process for visitors?
For example:
• visit museum -> local
• education resources -> global focus on Ontario
• enthusiast knowledge -> global
• games, etc. -> global
Land on hours/location and visit museum may be an indicator of a successful conversion, particularly for local visitors.
Traffic sources -- search is normal to be the highest. Direct traffic relates to advertising where people type in the URL. Keywords searching for your brand.
History.KY.gov
Don't have access to profiles so we'll use Compete.com
Quantcast.com is another good tool.
.gov addresses are hard to work with. Often orgs like to have centralized addresses (eg .com) to let them market and track better.
Google Trends is useful for advertising but we can also use if for other things.
Google Ad Words
AttentionMeter.com
Total Visits is one measurement. However, are those engaged visitors? What defines engaged visitors?
It often helps to have qualitative stories to accompany the statistics.
Safari, when on, hits RSS feed sites every 30 minutes. It does not run the JS so the Google Analytics stats don't rise accordingly.
Often have to state yearly targets of visitors.
Engaged visitors represent about 50% of traffic to server.
In general, we can tell more about the web visitors than we can know about who sees a print ad.
A "hitwise report" from Australian government shows that of government sites visited by Australian IPs, the Powerhouse museum accounts for 1.25% (#14), above all other museums.
What is working? What isn't? How can we make it better?
Analytics determines priority of what should be digitized.
Collections system--what are people looking for (and can/can't find) to determine what to prioritize.
Are the visits keeping or exceeding overall trends for Internet growth.
"What's on" is a common phrase.
Tend to only care about iPhone for mobile traffic.
http://report.reinvigorate.net -- instant snapshot of site. Good for spotting behavior patterns and navigation.
May need to build JS tools to note when page is active.
Social Media
Google Analytics has a visitor loyalty report. Very common for most to visit only on time (eg 79%). However, to make social media work, you need to have high numbers of users who visit 10 or more times.
Can view content, content drill down.
The goal of social media is that people keep coming back often.
GA Advanced Segments and Custom Reporting can allow one to determine clients with multiple visits and use it to determine what topics to use for a social media project.
How large is the community (eg visited more than 100 times, etc.)
Measuring reputation
Whostalkin.com/b>
It's not as important to know how many people are mentioning the site but rather what they say and if it is something that can be used in an anecdotal report.
HowSociable.com
Lots of numbers, many irrelevant. Shows who is linking to you.
Mostly uses site:mysite.com in Google.
Some museums will choose to respond to Twitter messages. For example: "enjoyed my visit to the zoo", "glad you enjoyed visiting the zoo"
URL shorteners (eg tinyurl.com) See http://searchengineland.com) to see what can be tracked.
http://searchengineland.com/analysis-which-url-shortening-service-should-you-use-17204
BackType.com
Shows who is linking to your site in Twitter Tweets. Often looking at http://tinyurl.com or http://tr.im
http://www.powerhousemuseum.com/dmsblog/
Seb's example uses a WordPress plug-in from BackType.com in early beta 0.1. May be something similar for other systems. Really must be moderated.
More Questions
What conversations matter to you?
How do you distinguish them?
What matters about them?
What chatter do you elevate or ignore?
Google has been adding site maps with key links for many sites.
Toolbars:
SeoQuake shows number of links in various tools.
DomainTools.com -- for example, now many links to your site in Wikipedia?
Wikirank.com -- a measure of interest in your organization
Reputation among Wikipedia editors of your institution as an authoritative source.
Flickr stats are pretty basic (geography, etc.). Recent Activity lets you look at the comments. Might want a GreaseMonkey script to drop the short (1-3 word) comments for a given photo.
# of comments and # of favorites is not especially important. The longer comments from regulars seem to be more significant, at least anecdotally, for the annual report.
"Who calls you a Contact" -- subscribe to Flickr
Net Promoter Scores seems to have fallen out of fashion. Asks one question "Would you recommend this to your friends?"
0 not at all likely
10 extremely likely
0-6 are detractors
7-8 are passive
9-10 are promoters
Although it doesn't work well with services and products, it works well for free museums and web sites since there is little financial barrier.
Survey can be placed on site with score.
%promoters - %detractors = NPS (Net Promoter Score)
"How likely is it that you would recommend [the company] to a friend or colleague?"
Rate this article (5 stars).
Content, Images, Style and readability
Not everyone fills it out but those who do provide useful information.
Reports
Google Analytics has some custom report features.
Good book: Web Analytics an Hour a Day. Tips on using the tools to make your site better.
SEO Toolbars
SEOQuake http://ff.seoquake.com
SEO for Firefox
Webmaster toolbar
TinEye.com image search.
GreaseMonkey scripts
GARE (Google Analytics Report Enhancer) -- included Real Time on Site column
Better Google Analytics (has GARE in it plus five other plug ins).
SEOMoz.org ($80/yr for paid tools e.g. Trifecta performs an SEO report). Good blog.
http://kaushik.net/avinash/ -- another good SEO blog with basic info.