Tuesday, August 30, 2011

Stores should have apps to locate products on their shelves

Here's an idea I have wishing was implemented for many years:

How many times have you visited a grocery store or hardware store and had a hard time finding needed items? It may be because you are unfamiliar with the store's layout, or because the items you want are obscure, or can fit into multiple categories. I find the problem particularly pernicious in huge hardware stores, when you are looking for items you may only ever buy once, and have no idea how the store is arranged.

Therefore, in every grocery store or hardware store, you should be able to search to determine if the store stocks the product and exactly where it is located.

This would be an ideal mobile app. Your Android or iOS device could quickly lead you to where a needed product is located. Furthermore, you should be able to plan your shopping on your home computer even before leaving to go shopping.

The challenge for store owners would be indexing their shelves, assuming they don't already do this for inventory-management purposes.. However consider the following analysis. Imagine a typical grocery store: Shelves have to be stocked constantly, and staff have to be knowledgable about where items are located. A certain amount of staff time is dedicated to helping customers find items. Also, customers would be attracted to a store with a good index, since they would be able to more rapidly locate items. All this argues in favour of it being a net-postive investment to have staff index shelves.

How much time would staff need to index shelves? Imagine there was a device available that could scan bar codes and using enhanced GPS, automatically enter the location into the store's index. Imagine then that someone could scan one product type every 10 seconds, and there are 6 product types per meter of shelf space, five shelves high, and 250m of shelves in the store overall. That would result in the 7500 products being indexable in about 25 hours of work - only a little more than half a work week. If indexing were redone only every 3 months, then the overhead would be 1 20th of a full-time-equivalent person. The indexing overhead could be reduced even further if it was tied to the restocking process (whenever someone is restocking the shelves, they scan the item).

Interestingly Tesco in the UK is making it possible for people to plug their iPad's into grocery carts! Why to add search to this.

The possibilities for enhancing this service are endless:

  • People could create a customized list of the products they buy (perhaps by scanning barcodes), and then check off the items from this list that they have run out of. The system could then give customers an optimized route through the grocery or hardware store that would lead them to pick up each item in sequence, alerting them to when they have arrived at the location where needed items can be found on the shelf. This would make shopping significantly faster. I suppose it would also reduce impulse buying, which may be one reason such a system has not yet been deployed.
  • If this mechanism were standardized across the industry, and a database of current prices was also maintained, people could do zero-thought comparison shopping, create multiple customized lists, allowing them to visit the shelves of several stores to achieve the overall lowest combination of prices. Of course, this would increase competition, and result in downward pressure on prices, which may be another reason why such a system has not yet been widely deployed my merchants.
  • Even if a store itself doesn't maintain such a database. perhaps an external enterprise could do it for the stores in a given city. The cost would be low enough that a low subscription fee ought to pay for the service. I would be willing to pay, for example $20 per year for access to indexes of the 10 grocery and hardware stores near me. It would require 750 subscribers to break even if someone is employed at $30,000 per year and full indexing is every three months. However if indexing was done in full only yearly, with incremental indexing for seasonal items less frequently, then costs could be substantially reduced. Of course, stores may raise legal challenges, asking people to leave who are not buying. However, all it would take would be fore one or two stores to allow indexing before the others would have to permit it, lest they be left out.
  • And what about crowd-sourcing? If enough people sign up to index a few items using their smartphones, every time they go to a store, the indexing problem may be solvable at a very low cost.
The user interface for search in such a system would have to be well designed. Many products have near synonyms, or are hard to spell. For example a search for 'spaghetti sauce' should find pasta sauce as well, and a a search for 'spagety source' should also find it, since not everyone can spell.

Other challenges relate to creating usable shelf maps with precise-enough positioning information. However I strongly believe that these are technology challenges for which good software engineering can really provide the solutions, and for which the hardware embedded in today's mobile devices would be sufficient.

There are some people already working on this kind of application: For example this article talks about ongoing work at the University of Washington, outlining some of the challenges. This article talks about planograms, which are the shelf maps that are already in use in the industry to assist with product placement.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.