Hillstrom's Merchandise Forensics

We focus on the presentation side of the story, and that's important, no doubt. We don't focus on the merchandise. Merchandise is more important!

I would prefer to have two customers buying $20 items than to have one customer purchasing a $40 item, as long as the gross margins are identical.

look at repurchase rates for customers buying new items, comparing them to repurchase rates for customers buying existing items.

Run A Factor Analysis. For me, this is fun! I analyze the weighted percentage variables (channels, categories, months, price points, new buyers, loyal buyers, promotions). I only extract two factors, using varimax rotation. I save each of the two calculated factors as new variables.

Identify Weights. I run a regression model to learn how important historical transactions are, compared to current transactions, using twelve month future demand as the dependent variable, and twelve-month historical variables (0-12, 13-24, 25-36, 37-48, 49-60, 61+) as independent variables. I only run the model on 0-12 month buyers.

Create Weighted Variables. I create weighted historical variables for each advertising/marketing channel (catalog, email, online, search, that kind of thing), each merchandising category, each month (including a separate variable for Cyber Monday), each price point category (0-10, 10-20, 20-30, 30-50, 50+), for transactions tied to a marketing prom

When certain new items emerge as potential winners, the marketing team must respond by giving these items featured real estate in email campaigns and landing pages, even the home page for that matter.

timeframe that is just too small to matter.

A Merchandise Forensics analysis allows us to figure out why a business is struggling. Our current suite of analytical tools are ideally suited for figuring out why conversion/response rates struggle.