Over the years I’ve been involved in a number of web-based companies. All had great ideas for their business model. One had one of them had a great idea for classified ads. It had the latest in taxonomic matching and advanced search capabilities. If you were looking for a Mustang, it could tell direct you to ads for cars or horses depending on context and other factors. Its search capabilities were ahead of the time. It had pretty much every bell and whistle the newspapers asked for and that the design folks could think of.
Then Craigslist came alone. Craigslist was free (at least compared to newspaper classified ad sites where the newspapers typically charged.) It had no taxonomic matching. Its search capabilities were and still are bare-bones. In fact, it very much relies on the user to narrow down and define searches.
But it succeeded where the other product failed for what I believe one very simple reason. It was simply blazingly fast. It didn’t matter if it returned bad results the first time. It was so fast the user didn’t mind typing in new search parameters and narrowing down their search. It was faster than any of the “advanced” newspaper classified engines I saw. Sure, they might try to do a better job of returning results, but the honest truth was, in most cases people would end up doing multiple searches anyway trying to narrow down their search. And in the time it took to do 2-3 searches with a typical website, Craigslist allowed the user to do 10-15 searches. Time was money and people wanted to do things quickly.
Over the years with numerous sites I’ve seen the design get in the way of the end-user. The truth is, 80% of the time, people will use 20% of the features, but they want those 20% to be as fast as possible.
So, keep it simple and keep it very fast.
One of these days though I’ll relate the story of the 3,000 mile Steinway search.