Background

In the early Spring of 2003, I answered a job posting at my University for a temporary position unpacking boxes of antique porcelain for possible photographing and posting on a website. At this point I was a recent graduate with a degree in History, and had sent my resume out to dozens of companies with little or no response. I was hired immediately, and began learning of my employer’s ambitions to create an online web business based upon the vast inventory he inherited from his mother. I began working with the company he had already hired to build an e-commerce website for him, and my temporary position quickly developed into the full-time one of general manager of an emerging online company. The website company he hired provided all the infrastructure I needed. Everything from an interactive database administration interface to the front-end website template were provided for me. It was up to me to decide upon what sort of category structure made the most sense given the products in the warehouse, and to write relevant content for the site accompanying these categories. It was then that I first learned of the concept of “search engine optimization,” and slowly the company began educating me in these techniques as I handled the day-to-day task of building the site according to their instructions. It was with this project that the foundation for my career as an information architect were laid.

Category Management

The tsnumai of information I encountered upon first being showed the warehouse where the antiques were stored provided me a vertiable meditation in the art of category management. There were literally hundreds of boxes filled with about the same number of distinct patterns of antique china, lined in rows along the concrete floor of this approximately 2000 square foot warehouse. The dinnerware pieces were completely mixed up, meaning that any one box could contain several different patterns. Our first task, then, was to consolidate the hundreds of patterns into distinct boxes. This had to be done manually, as the owners and I wandered about the warehouse with plates and saucers in hand, playing a massive game of ‘Memory’ as we matched patterns and moved endless stacks of expensive, fragile dinnerware into their proper places. And yet, the dinnerware patterns themselves represented only one level of organization. Other attributes included the types of dinnerware (dinner plate, lunch plate, soup tureen, etc.,) as well as manufacturer (Haviland, Meissen, Dresden, etc.) However, it was not just dinnerware my employers had inherited, but figural scenes as well. These could also be categorized by manufacturer (which sometimes corresponded with dinnerware manufacturers, sometimes not) as well as whether they were a singe “figure” or “figurine,” or a “figural scene.” Besides all this, there was also a good amount of cut glass and silverware, which we basically ignored. My technical understanding at this time was rather limited, although I have always had a natural ability with computers. (I wrote my first computer program in BASIC at age 7.) What I ultimately decided was to have one category which represented the majority of the total inventory: Haviland dinnerware. And beneath this were a couple of hundred categories named partly according to Haviland’s own category scheme, but mostly according to the Schleiger system which I learned was the industry standard for this line of China. I then created separate top level categories, based upon manufacturer and product type. Thus there was a top category for Meissen Dinnerware, as well as categories for Meissen Figurines and Meissen Figural Scenes. The same was true for Dresden and “Dresden-style” manufacturers. Admittedly, the scheme I ultimately decided upon was not perfect, but it was the best I could do given my level of technological understanding at that time. Combined with the search engine optimization of the site, however, the products were ultimately very easy to find and I do not recall very many calls where a user could not find what he/she was looking for, if in fact it existed on the site.

Search Engine Optimization

From the very beginning, the search term we were targeting was “antique china,”which was sought by search engine users hundreds of times each day. I spent most of my time, however, researching and writing copy on each of the manufacturers represented in our inventory. Each day I brought home books that had also been passed down to my employers from the original proprietor, and wrote extensive histories of the different manufacturers as well as some general history of the discovery of porcelain in Europe. After writing histories of all the manufacturers, I set about writing product descriptions for all of the products we had uploaded to the site. (Of course, I also spent a good deal of time photographing these products and determining pricing, which are other topics altogether.) Soon we were getting calls about our china, and we were gaining the attention of other websites which were happy to link to us. Looking at our webstats, I watched the traffic grow from a handful to a hundred a day, then to hundreds a day. Finally, after only a few months, we hit one thousand unique visitors in a single day. It was around this time that I came to work one day to find my boss in a fury of excitement. “We’re #1 in Yahoo for ‘antique china!’” he told me. And sure enough we were. He had a huge blowup of the results page printed and mounted on his wall. (We soon gained a #1 ranking with Google as well, and I have only recently fully understood why we attained this ranking with Yahoo first.) ConclusionNeedless to say, this was an exciting and largely successful attempt at information architecture, though at the time I did not even know such a term existed. What I lacked, however, was a strong technical understanding of the technology we were borrowing from the web design company; specifically ASP and SQL. Afraid to touch the code, much less the database calls and procedures, I was left with the standard interface they provided all of their clients, and the few minor adjustments I was able to convince them to make. Were I to do it over again, I would do a good number of things differently. The most significant difference, however, would be the manner in which I categorized the products. It is my belief that static category structures, if they must exist at all, should be based upon a single, irreducible attribute. In this case, I would have probably chosen “type.” “Tea cup,” “soup bowl,” and even “figurine” or “candleabra” would be product types. Every other attribute such as “manufacturer” or “pattern name” would be entered at the product level. Had we chosen to fully incorporate the glass and silver merchandise, an additional attribute of “material” could have been used. The value of such a scheme is that it would have given maximum flexibility to the user based upon what he/she actually knew about his/her own wants or needs. This is important for all types of products, but in the case of antique china it was particularly cogent. While some people knew they wanted a dinner plate by Haviland in the Schleiger 322 pattern, there were others that just knew they wanted a pretty cup and saucer for their curio cabinet. A fluid scheme that made use of dynamic “virtual categories” would have been extremely helpful for this website.