November 19, 2005
Under Analysis? II
It's been a week since Google Analytics was made public and the service has been overwhelmed by those of us eager to explore and take advantage of this offering, leading to slow data delivery and prompting Google to temporarily close the service. From the Analytics sign-up page:
Google Analytics has experienced extremely strong demand, and as a result, we have temporarily limited the number of new signups as we increase capacity. In the meantime, please submit your name and email address and we will notify you as soon as we are ready to add new accounts. Thank you for your patience.
While waiting for data to trickle in I've learned a few more things worth mentioning.
First, if you are on a Mac, you'll need to use a Mozilla-based browser like Firefox to view reports, Safari won't do.
You'll need to have a Google Mail account, and anyone for whom you want to create an Analytics account must also have a Gmail address.
During this break-in period as Google adjusts for the unexpected demand, you may find that data will sometimes show up in reports only to disappear later on, or reports will be incomplete. Google states that all data is being captured, so I assume that this missing data will return after things calm down.
The Site Overlay feature has been taken off-line, hopefully temporarily. You can still play with it if you view-source to get the commented-out Site Overlay link.
If you need to track downloads or links coming from Flash or Javascript links, you'll need to add tracking codes to your script calls; see Flash, downloads & More in the Google Analytics Help Center.
There doesn't appear to be a way to see full referrer paths. For instance, you can see that a user came to your site by way of site x, but you cannot see the actual site x url that brought in your visitor.
Last Note: Google Analytics should not be thought of as a total replacement for a traditional log analysis program. Since Analytics can only report on pages containing tracking codes, you won't be able to analyze 404 and server errors, time to serve numbers or bandwidth utilization. However, there are free stats packages that can be relied upon for for this kind of information; right now I'm partial to the open source AWStats, which we've installed for clients here at Threespot Media, used to generate reports for clients who have no stats of their own but do have logs, and finally, for use on this site. Might be a good combination of tools to recommend to clients once Analytics finds its footing.
November 14, 2005
Under Analysis?
Google's purchase of Urchin last March bore fruit today as Google Analytics, a free web analytics program for up to 5m page views monthly unless you have an AdWords account, in which case there is no ceiling.
Urchin is a very powerful program, still available for those who need statistics for intranets or otherwise private networks and appears to have lost little of its power in translation to a free service. Analytics offers 80 predefined reports with 18 predefined segments, the latter of which allow technographic drill-down reports such as Flash plugin version, connection speed, screen resolution and color depth. One of the neatest sounding features is Site Overlay, where your site is rendered in a frame with click/conversion data overlays for each link. You can compare date ranges on any report, do a/b comparisons on ad performance, view drill-down GeoTargeting, Google style in a Flash 7-based report. In fact, the tool is chock-full of DHTML and Flash, which as ArsTechnica coverage points out, may cause some problems if your platform is not Mac or Windows.
Implementation is via a couple lines of Javascript inserted in every page you want tracked, the code itself generated for your use when you sign up for the service. For dynamically generated sites this may simply involve minor changes to one or more templates while static sites could call for a site-wide search and replace operation; either are fairly low-impact.
What does this mean for your clients? Those who have lived with substandard or no reporting at all no longer have a reason to ignore web analytics. Larger clients who have invested in tools may find it worthwhile to migrate to Google Analytics anyway in order to take advantage of cost savings and the lowered hassle of having to maintain their app and server logs.
Clients who have more than 5m page views per month and who do not participate in Google's AdWords program may still want to take a look—are they happy with their web analytics app? Is log management a thorn in their side? Do they advertise already? Have they considered the AdWords program?
One question you should ask your clients is how comfortable they might be with not hosting their own data. The client may have business guidelines or regulations against data sharing, with which a service such as Google Analytics may be so construed. Such a business would also steer clear of similar services such as WebTrends On Demand, which, I might add, is far from free. ;)
I've signed up for the program and am anxiously awaiting my reports -- the service is early-adopter-slammed at the moment with Google reporting a 12-hour processing delay. I hope to have more to say on this after I've had a chance to check out the actual reports instead of the report frameworks I've been exploring while I wait.
This is big news, folks, to be sure, Google's changed the face of web analytics to add value to their advertizing offering and I'm sure we're going to be hearing a lot more about this in the press and from our clients, best to bone up on it now, perhaps a good start would be at Conversion University?
November 10, 2005
Paparazzi
Nearly every platform provides a way to get screen shots. Sometimes you want to get a screengrab of a web page that is too long or large to actually display on your screen: enter Paparazzi and url2bmp. Paparazzi is a free Mac OS X.2+ utility leveraging WebKit to output screengrabs of entire web pages as viewed through Safari, and url2bmp does the same with IE on Windows 95+.
Paparazzi lets you arbitrarily choose window dimensions, though most would select a standard 800x600 or 1024x768 screen res, you can crop the grab so that only the first screen-full is captured. A grab delay may be set to allow plug-in content to load or to offset a slow or congested network connection. A preview pane shows what you end up with before saving.
Supported file formats include JPG, PDF, PNG and TIFF. The PDF version preserves system text. You can choose to save the entire grab, a scaled thumbnail, or both.
Because of the close tie to Safari, Paparazzi can grab whichever url is currently being displayed in Safari. I've created a simple bookmarklet that reverses that -- when on a page you want to grab, selecting the bookmarklet will open Paparazzi and start grabbing. To use, simply drag the Paparazzi-Grab link to your Bookmarks bar, navigate to a site to test and click on the link to try it out.
Designed to meet the same challenge as Paparazzi, url2bmp, which, despite it's name, also outputs JPG, PNG and TIFF formats and relies on a PC IE engine version 4.01 or better. It's slightly less featured, for instance there's no cropping capability, but has the screen res and delay features you need to get the job done.
November 9, 2005
The Shape of Things to Come
Before we get started, let me introduce you to WebKit. WebKit is the open source framework used by Safari and several other OS X applications, based on the Linux KHTML engine from KDE.
Recent Webkit builds implement new CSS3 properties which will someday make designers and html coders very happy indeed. Eventually these properties will allow non-rectangular designs that are trivial to implement, whereas today's browsers require complicated and sometimes tricky table structures to pull off.
For instance, to make a rounded-corner div using CSS3 all one has to do is add border-radius with an appropriate value to your CSS and you're done! If you set a background image for that div, then the image will be clipped by the the value of your border-radius: an easy way to give rounded corners to rectangular images w/o pre-processing. CSS3 also allows defining border images which can be set to stretch along with a div using liquid-layout techniques, all with much less code and effort than the traditional methods.
As of right now, there are no shipping browsers that support these CSS3 properties, the standard itself still being in draft form. In order to experiment with or view the demos I'll point to below, you'll need to download one of the nightly WebKit builds on an OS X box or view the screengrab examples. As the WebKit nightlies are self-contained, a build will co-exist peacefully with your existing version of Safari.
Here's a screengrab of this site [240k, png] viewed through one of the WebKit nightlies. Note the rounded-right edges on the right-hand column and how the 1px border doesn't yet quite follow the rounded lower-right corner on the blog entry divs. The CSS that defines corner radius is ignored by browsers that do not yet understand it.
An excellent example of a clipped rectangular background image [screengrab + example page].
Turns out Mozilla has implemented extensions to CSS that also allow rounded corners, available since NS 6. These extensions were created for use in XUL and in preparation for CSS3, but use a different syntax in implementation from the current spec. Safari currently recognizes Mozilla's shorthand syntax [-moz-border-radius] but not the explicit setting of individual corner radii, where the Mozilla/CSS3 deviation becomes more acute.
Form elements are also affected by the Mozilla extensions. Here's a screengrab of this site's search field and submit button viewed in Firefox.
