Web Applications

Time and Audience Advertising Models for the Web Thursday 5/18/2006 11:27:39 AM (CST) - Michael Wells    The first commercial websites I built was a webzine.  A city site focused on promoting events, restaurants, and so on.  But because it was 1996, there just weren't any advertising models for the web, so I settled on a tv-style advertising model.  We used the weblogs to calculate the number of viewers, when they viewed, and the amount of time they viewed, rather than the number of pages they hit.  We did this largely because hits didn't mean much to the people we were selling adspace to, but they did understand tv and radio advertising models, and understood concepts like primetime.  Why didn't this take off?  I'm not sure... too may people saw the web as a magazine (due to it's print-and-image-based presentation at that time) rather than as a tv broadcasting network (due to it's real-time ability to serve content based on time of day, geography, and so on). I've been doing sites now for just over 10 years, and I still like the tv advertising model better.  Doesn't it make more sense? I do think we're likely to see a push to switch the model soon, by sites that are very media-heavy.  For example, if I'm YouTube, I know that I can keep users on a single page for minutes at a time.  That's a lot different from other websites, and I want to use that in my advertising costs.  If I stick your ad on a page where people are looking at your logo for an average of 3 minutes, I sure want that to cost you more than a different page with an average of 1 minute.  [ Comment on this Entry ] [ Email this Entry ]

Hacking Web Forms (and making them Usable) Wednesday 5/10/2006 10:56:56 AM (CST) - Michael Wells    If you're like me, the web has undoubtedly become the largest neighborhood in your personal infocity, and there are a handful of restaurants there that get essentially all of your business. You return to these same restaurants again and again, often several times a day -- for the convenience, nice ambiance, but primarily because the food is good and it's something you need.  Even with all those perks, if you're visiting 3 times a day, and after a month the Maitre'd still doesn't even recognize you, are you going to keep going back forever?  Most likely not. What website operators don't always clue in to is that their customers expect nothing less.  Modern consumers are willing to trade some degree of personal touch for better, faster, cheaper... but convenience of use is simply a non-negotiable factor. [ Comment on this Entry ] [ Email this Entry ]

Css url() and Relative Paths Monday 5/1/2006 5:11:49 PM (CST) - Michael Wells    I must have forgotten this little behavior of Css -- when you use a relative path in the url() construct, e.g. background-image: url(def/abc.gif); the path is evaluated relative to the locaiton of the Css file, and not relative to the location of the page referencing the Css file. That seems pretty far off-skew with other browser behaviors, but it turns out to be very handy in a number of cases.  For example, if you're developing a .NET app that has to run in both virtual directory mode and in a root path configuration... this approach can prevent your Css files from breaking.  And simplify many of your image references, all at the same time. [ Comment on this Entry ] [ Email this Entry ]

How do you design a remixable web application? Monday 7/4/2005 7:21:18 PM (CST) - Michael Wells    Jon Udell queries, "How do you design a remixable web application"?  An excellent, if extremely brief entry on a topic that's easily worthy of a few good books. Jon points out that many of the interesting applications we see today are built on top of other applications.  Saavy developers take a touch of Google, a splash of eBay, a dollop of CraigsList, a sprinkling of deli.cio.us, and create something entirely new.  And why not?  [ Comment on this Entry ] [ Email this Entry ]

Dynamic CSS Generator Saturday 8/2/2003 11:51:32 PM (CST) - Michael Wells    I haven't considered this thoroughly, but it seems that for web applications that must be customizable, that the most effective way to customize the look and feel is to implement dynamically-generated Cascading Stylesheets (CSS). [ Comment on this Entry ] [ Email this Entry ]

Unique Web Interfaces Saturday 8/2/2003 12:35:35 PM (CST) - Michael Wells    About a year ago, I came across the WebFX Dynamic Webboard, which caught my attention for its unique web-based UI design.  In addition to being quite cool... It does a great job of emulating unique GUI components, such as the scrollable folder list at left, the toolbar at the very top, and the hierarchical listview at top. It is very detailed in its interaction.  Mouseovers on list items show a hover effect that is pleasing and useful. The toolbar at top particularly caught my eye.  The developers managed to make the toolbar match Internet Explorer's toolbar design and appearance almost perfectly, which easily fools the user into thinking that the browser itself has changed.  It has not; this is part of the HTML webpage.  To convince yourself, right-click on it, and you will see IE's pop-up browser content menu. [ Comment on this Entry ] [ Email this Entry ]

Real-Time WebServer Traffic Analysis Saturday 5/31/2003 5:31:06 PM (CST) - Michael Wells    It might already exist... (if it does, let me know) but I'm getting weary with the limited capability that a webserver log provides.  The information itself is only semi-adequate; but more importantly, the fact that the information is stored for future analysis (rather than real-time display) is frustrating. It seems fairly obvious that real-time analysis engines should exist.  A simple graphical display, showing a website in a hierarchical fashion, color-coded to indicate traffic-heavy areas.  For companies like CNN.com, which take a more real-time, broadcast-oriented approach to their websites, this type of capability seems absolutely essential... and quite simple to build. [ Comment on this Entry ] [ Email this Entry ]