The open concept worked at first, because I perceived an issue I had when shopping: some shops have far too many rooms, hallways and areas in which to look for items. It makes shopping nearly impossible for those in a rush, although some may find difficult shopping experiences fun. I wanted to avoid that situation, especially where customers don’t even see your products because they were in a room the customer never encountered. So I simply laid out all my items in easy to view locations. My first store had no ceiling and very few walls.
However, as the store grew and more products were added, the requirement for more textures (mainly on product boxes) continued to grow. It came to a point where a visitor would be faced with loading (albeit automatically) over 150 textures in order to see the store. Tracking software indicated some visitors would appear and then quickly leave. The obvious conclusion was that they were overwhelmed by the gray view and simply went somewhere else without shopping. Not good!
After weeks of running in this mode, I now conclude the texture load theory was wrong. While the store did come into view much faster when teleporting in, sales dropped off significantly. At first I took this to indicate a continuation of the degraded economy in general, but several friends (Amber, Haley and Marlee) suggested the visibly blocked store sections conveyed a different impression to shoppers - hidden products must be creepy! Since shopping is a very psychological matter, especially in SL, I thought there’s probably something to this theory.
Ten days ago I took down the obscuring walls to once again reveal the entire store to view. As expected, long texture load is now quite evident when teleporting in, but what would happen to sales? After ten days of “open” operation, I can safely say that sales are up significantly!
The conclusion? While texture rezzing performance is always important, the psychological factors that develop from the store’s visual appearance trump performance when it comes to shopping.