Bountiful Baby Automated Sales Strategy Explanation

Hi Everyone!

This is from Nevin Pratt (CEO of Bountiful Baby). I thought you might be curious how our computer algorithm for conducting sales is currently configured, so I thought I would share that information with you (if you are not that curious, then go ahead and skip this email).

The first thing the computer does is a “Days Of Stock” calculation. For each item, that calculation is based on the following: if sales continue at the rate that they have for the last 3 months, how many “days of stock” are remaining for this item? But, why “3 months” of sales data, you might ask? The intent is to discover the best month value that will be the best predictor of future performance— too little data works against that, and too old of data also works against that. It looks like we are zeroing in on “3 months” sales for that calculation, but that number has varied widely in the past as we experimented.

Next, we have tried to discover the best number of “Days Of Stock” that will allow us to reorder in a timely manner, but not too many “Days Of Stock” so that money and warehouse space is unnecessarily tied up. As we have gotten better at predicting that number, we have gradually reduced that number to a lower-and-lower value, above which the items were triggered to be put on sale, and below which things stayed at normal price.

As that number has been gradually reduced, it has resulted in quite a few sales, and quite a few items put on sale— things were being selected for sale that the algorithm would not have selected previously. But we are hoping we are finding a point of stable equilibrium now. If we are correct, the general trend should be towards less-and-less items on sale. But we’ll see.

If you are curious, the current algorithm for determining the kits that are placed on sale and put on our home page is done by the computer as follows:

  1. Get a list of all of the items in category #2411 and subcategories.
  2. Remove any item from the list with “DISCONTINUED” in the title.
  3. Next remove any items from the list that have only 1 (or less) quantity in stock.
  4. Of the remaining items in the list, set their discount percent as follows, based on their respective “Days of Stock” calculation:
    More than 250 days of stock = 60% discount
    201 to 250 days of stock = 50% discount
    171 to 200 days of stock = 40% discount
    150 to 170 days of stock = 30% discount
    Everything else is not on sale.

The computer does this for us, and our home page is set to the kits that the above algorithm decides to put on sale. The calculation is redone approximately twice a day, so kits might move either up or down the sale list, and might even be taken off sale at any time. Also, as our “Inspection Crew” inspects the newly received vinyl, those kits are put into inventory and the computer detects it and might even put the newly inspected kis on sale. Realize also that the kits that are being inspected now were actually ordered months ago, and at that time we had a higher “Days Of Stock” threshold for ordering, which means that for the last little while, and going into the future for a little while longer, newly inspected kits are likely to be put on sale right after they are inspected.

But all of this should start to begin to stabilize, with changes happening less-and-less often. At least, I think that is what will happen. :slight_smile: We shall see.

Don’t know if any of you care, but I thought you might be curious about some of the internal workings of how this is done (non-kit items follow a similar—but not identical—strategy to the kit strategy explained above0.

Hope you are all having a great weekend!

Nevin Pratt
CEO, Bountiful Baby