It’s Up to You to Fix What’s Broken

Many companies measure their Web site availability from a “service level” perspective. In general, this means the percentage of time that the site is available from the Internet and that the systems are up and working properly. The gold standard for Web site availability is “five nines” uptime, or 99.999%.

Achieving high service levels is no longer expensive or difficult. For example, an external Web site monitoring service can check availability every two minutes for $50 a month. In addition, the use of redundant bandwidth providers reduces connectivity downtime to near zero, and this redundancy is usually available as a matter of course in hosted data centers.

Excellent system monitoring software, available for a few thousand dollars, tracks the status of your infrastructure software and systems, and warns you when they are not running smoothly.

Finally, a load balancer with a small amount of redundant equipment can ensure that any internal hardware errors do not disrupt the site’s performance or availability even at peak loads.

But end-users — your customers — do not measure Web site availability from a service-level perspective. To be sure, if traditionally measured service levels are low, then users will be frustrated. But this is not the only aspect of availability, as is commonly assumed. Rather, from the user’s perspective the only measure that matters is whether his or her interaction with the site was successful.

From a user’s perspective, the measures are “Was my question answered? Did I find a hotel room? Did I pay my bills successfully? Did I find the item I wanted to buy? Did I feel confident that my transaction occurred as expected?”

What follows are specific examples of problems your Web site might exhibit, and that cause users to view your site as “unavailable” from their perspective:

  • Users are accessing your site via a link from Google, from one of your affiliates, or from someone’s weblog entry, and that link has an error or is out-of-date.
  • You may think you are not responsible for the errors other sites make, but the end users don’t care who is responsible — and you are the one losing customers.

  • The instructions and navigation on your site are unclear, and a certain segment of users are not providing requested information as intended, so their transactions fail.
  • User error? Not from their perspective — they just clicked to your competitor’s site.

  • You just updated the software for your Web application for the third time this quarter. It has several new features that customers appreciate, but it has a bug that just one in one thousand users encounter.
  • From the perspective of your users, your service level just went from five-nines to three-nines.

  • Like most Web sites, you’ve designed and tested your application using Internet Explorer. When it is accessed with Firefox some of the key application buttons are not visible.
  • Your sales to this increasingly large audience (estimated at five-to-20 percent of users and growing rapidly) drop like a rock.

    The good news is these problems can be mitigated to some extent with proper quality assurance and design practices. A Web site is, at its core, a piece of application software, and an excellent site will use best practices for software development.

    But this will not eliminate the problems, and ultimately the only way to stay on top of them is to monitor the application in production.

    A recent E-Commerce Performance study showed that 72% of users do not report Web site problems they just go elsewhere. To keep these fickle customers, you must fix what is broken.

    David J. Jilk is CEO of Xaffire, an innovator in live Web session capture and playback technology. He has an extensive business and technical background in both the software industry and the Internet. In addition, Jilk independently advises startup firms and provides technology due-diligence services for a variety of investors and acquirers. He can be reached at [email protected].