"Count what is countable. Measure what is measureable. What is not measureable, make measureable." -- Galileo

Friday, May 29, 2009

Blank Spots in Visitor Loyalty

While things are hopping over at Plone Symposium East, I thought I'd finish up my series on Plone.org visitor statistics from Google Analytics. Tonight I'd like to discuss several under-appreciated metrics: loyalty, recency, length of visit, and depth of visit. I'm inspired by a blog posting by Avinash Kaushik that promulgated these measures for non-e-commerce websites.

Loyalty. Plone.org visitor loyalty has four modes, a mode being a local maxima in the frequency graph. A huge number of visitors (51%) come to the site only once, but then there is an up-tick after 8 visits. Of those, there is another up-tick after 25 and another after 200. From the graph, one would suspect that visitors returning more than 8 times represent a distinct population.

The long and the short of it is that Plone.org updates its content frequently with new documentation and products coming online daily. Returning visitors who come back over 200 times are clearly deeply interested in Plone. Even those who revisit just 9 times probably are users of a Plone site or owners of a simple Plone portal who need more than just a single answer to a question or the one-time download of an online reference book.

Recency. Fully 82% of all Plone.org visitors came within the past 24 hours. There is also a noticable peak in the "8-14 days ago" visitor population. These are routine visitors who make a habit of returning to Plone.org 2-4 times per month. As an action item, should the Plone community “sell” harder the value of repeat visits to our audience?

Length of Visit. About 55% of all visitors stay for less than 10 seconds. Obviously they were looking for something else and instantly recognized that their search had taken them too far afield. Of those who stayed more than a few seconds, 22% stayed for 1-3 minutes, 23% for 3-10 minutes, and 21% for 10-30 minutes.

This seems to indicate that about equal numbers of site visitors find their answer or locate their download within 180 seconds, within 300 seconds, or within a half hour. Considering the number of brief how-to's as well as the length of some tutorials, this probably is to be expected. A laudable goal for Plone.org might be to increase the under one minute population, which would indicate that site users found their answers quicker.

Kaushik states that "If you are a support website then should you be embarrassed if 20% of your audience was on the site for more than ten minutes!" Reducing the number of 10-30 minute visitors would be another excellent goal for Plone.org. That would mean that solutions to problems were becoming simpler and no longer needed lengthy, detailed tutorials. That said, I think there will always be a need for comprehensive tutorials that explain a process from start to finish in complete detail.

Of course, these metrics fail to capture those Plone users who are able to answer their own questions just due to the high usability of the Plone sites without needing to look up any online references.

Depth of Visit. As might be expected, 50% of all visitors stop after a single page. These no doubt correspond to the 51% of single-time visitors, the same ones who stay less than 10 seconds.

You'll note, however, that this is bimodal with almost 5% of all site users viewing 20 or more pages. I'm guessing these are developers and power users.

Its a testimony to Plone's internal search engine that visitors who stay beyond a single page view find what they're looking for (or if you're a pessimist, they give up) within a handful of pages. Turns out the average is 2.6 pages. That's about right for the following use-case: go to Plone.org, enter a search term, view the results page, select a relevant result and view it.

This concludes my "Blank Spots in..." tour of Plone.org statistics from Google Analytics. I'll try to get back every six months or so and see if things are moving around. With all the recent changes to the home page, I would expect to see some metrics on the move and in a positive way.

Tuesday, May 19, 2009

Blank Spots in the Week

Last week I took a brief look at some Plone demographics along with some pageview observations. Tonight I'm taking a gander at the periodicity of visits to Plone.org. Here's the graph of visits over time.





I've identified the maximum (A) and two minima (B). The maximum is easily explained--Tuesday August 21, 2007 was the rollout date for Plone 3.0. The minima are also easily explained--the annual December holidays, Christmas through New Year's.

What of the sawtooth pattern? Its the weekly periodicity of activity. Week in and week out, Plone visitors peak early (Monday or Tuesday) and taper off significantly on Saturday and Sunday.

The hourly data (below) show an up-tick beginning at midnight with another rise around 0700. (All times are Google Standard Time, that is, -0700 GMT.) That first rise coincides with Europe going to work and the second bump begins as the eastern seaboard of the Americas gets going. Things taper off as the west coast closes up shop.





From all this we can conclude that the general Plone population of visitors are not overwhelmingly hobbyists. Both annual and weekly patterns support the view that working professionals are using Plone.org with down time for weekends and annual holidays. The hourly data reinforce this with the pattern matching that of the working hours for Plone's largest populations, Europe and the Americas.

My last observation is the remarkable consistency in visits. There's no need for regression on that first figure--its obvious to the eye. Plone.org visits have been incredibly steady, except for the periodicities noted above, ever since data has been tracked by Google Analytics.

For those who look at Google trends and claim that Plone is declining and becoming a niche player in the CMS market, this is the strongest data I've seen that refutes that position. Plone.org visits over time show a strong and consistant population of visitors, the majority of which are professional.

Furthermore, one would expect most visits for support (see last week's post) to occur during the development of new sites. As sites enter stable production mode and support visits drop off, other new sites must be cropping up to replace them. The horizontal trend line of Plone.org visitors is an indicator that new sites are being created at a fairly steady rate over a several year period.

For those who would argue that the horizontal trend is simply the same band of loyalists hitting the site day after day, stay tuned. More to come on visitor loyalty in my next post, as I continue to probe Plone.org's statistics.

Friday, May 1, 2009

World Plone Day

Happy May Day. This morning's post is a collection of miscellany while my main article on Plone.org statistics cooks on a back burner.

For anyone who hosted a WPD event last week, kindly take a moment to fill out the simple survey form that is collecting data on the many different events out there. The current results only show Istanbul, Albuquerque, Pisa, Taipei, Paris, and Cologne. Thanks.

I found an interesting article in the Sydney Morning Herald about CMS, especially FOSS vs subscription services, by Valerie Khoo. That in turn has lead to a lively discussion at CMS Report that continues up to this week.

CMS Wire is staying on top of things and I see they have a new end-of-the-month feature looking at what's up and coming in the month ahead. They made mention of both Plone symposia and it's always good to get press like that.

Web Hosting Geeks has an article that in its own way reiterates my own position: let requirements drive your CMS decision. The article discusses WordPress vs Joomla, but frankly one could just as easily strike out WordPress and replace it with Drupal or SharePoint and switch Joomla for Plone.

In an unrelated item, Packt just rolled out their author website, which may inspire some to pick up their virtual pen and write that book. Of interest to me is Packt's announcement of their 2009 Author Awards. Voting is now open but the rules limit contenders to authors of new or revised titles from 2008. With several new Plone titles coming out this year, plan on this being a hot item next year for the 2010 Author Awards.

In closing I'd like to thank Nate Aune, Jon Stahl, Chris Johnson, David Brenneman, Ross Patterson, and Alexander Limi for their hard work at the NTEN Conference this past week. From the looks of the online material at NTEN it was a great conference. I wish I'd kept better notes on the Twittersphere this week--there was a noticible up-tick in positive traffic that I attribute to the Plone booth at NTEN. I also note that there's now a Plone twibe at http://twibes.com/plone.