By Tim Caynes on Jul 17, 2009
Hot, well, warm on the heels of our storage finder, we have recently deployed the new server finder, which replaced a previous incantation of finding functionality which was held together by hello world string and was somewhat creaking at the edges. The new server finder is based on the architecture we developed for the storage finder - they are, in fact, 2 renditions of the same finding platform - and so leverages the features that make it so worth the investment of effort.
As with the storage deployment, the key to making the server finder successful was, well, a number of things, but the main thrust of our efforts was defining the data architecture that make the relationships between product groups, products and product attributes a meaningful one. This can only happen with some Herculean efforts being undertaken by our publishing teams in conjunction with the product marketing teams, who really understand what is important and relevant about the products they market. Really, the finder itself is just a layer of abstraction on top of the data set underneath, and in theory (as we are at pains to try and progress), can be applied to any well-structured data set. What matters, is whether the data that a customer, user, or casual visitor is presented with, and the methods they can use to interrogate that data, enables them to reach an appropriate destination. In other words, they might know where they want to go, they might have a vague idea, or they may have no idea at all, but if we've done our job as well as we should be doing it, the directed searches and filters that the finding platform utilizes should provide a the equivalent of a product sat-nav, but avoid the 18-wheelers that get grounded on hump-back bridges in the middle of Hertfordshire on the way to the new Tesco Express.
Probably an analogy too far there, but it is by way of illustrating that the key to the finding platform is the data that it manipulates. I mean, we did a number of detailed usability trails, with various rapid and high-fidelity prototypes, struggled over the tiniest nuances of labels and gradients, fought compromise on page region refreshes and a followed number of other noteworthy user experience best practices, but in the end, if we built our application infrastructure on top of a taxonomy akin to a river bed full of shopping trolleys, we'd only be providing half a solution, which, in fact, is no solution at all.
We've still got a number of things to work on that didn't make it into the first release, such as enabling product comparisons across products and, more difficult, across product in different product families, but take a look for yourself and let us know what you think. Comments are more than welcome, especially ones that are nice.