I don't see why anything fancy would need to be done.
Please verify for your study that the concept choice (as recorded in the .CSV, .CHO, or .CHS export of your CBC data) would tell you exactly what position on the shelf a SKU appeared in (e.g. top shelf, bottom shelf, middle shelf). For example, concepts 1 through 6 are always top-shelf products, 7 through 12 are middle shelf products, etc.
Then, with some back-end post-processing of your .CSV or .CHO file (whichever you prefer), you could add some new independent variables (attributes) to the experimental design, such as a new attribute describing if the concept is on the top, middle, or bottom shelf. Assuming you have an experiment where the SKUs changed positions on the shelves (say, across people), you could estimate a shelf position effect in addition to the typical brand and price effects. This would need to be done by modifying your .CSV of .CHO file (to add the new attribute to the design) and submitting the new .CSV or .CHO file to CBC/HB standalone system or Latent Class standalone system.