This analysis requires an assumption that may / may not be valid.
Say a customer puts three items in his/her shopping cart. Is the first item what "started" the order? We don't truly know, but if you assume the first item is the item the customer wanted and then added on other items, well, you have the basis for an order starter analysis.
You code each item based on the order the item was placed in the shopping cart. Then you analyze all items, averaging the "order" out ... eventually you will see that some items appear 1st or 2nd in a shopping cart and other items appear 2nd or 3rd in a shopping cart. The former item starts orders, the latter item complements orders.
Order starters eventually become the items that the marketer features ... which creates a problem because the order starters eventually appear to be order starters because of marketing strategy and not because they actually start orders.
But for the most part ... the analysis is enlightening. It's one you should perform. At some companies, you'll see womens apparel early in the order starting process and mens apparel later ... it tells you that the woman is buying for the man, for instance. You'll see other categories (stuff like Home, for instance) added on. You'll see those "Featured on TV" items at the top of the order starter process ... this tells you that marketing is playing a key role in the order starter process.
Perform the analysis - it's enlightening!