Over 10 years ago, a grocery chain in the Northeast, Stop & Shop, launched the customer loyalty or savings card. You know that card you need to swipe to get the advertised savings. After a couple of years they pulled the card for three reasons: one, the customers weren't happy with it; two, state law required the store to give everyone the savings anyway (if they asked for it); and three, they weren't getting a return on the data.
We didn't have Big data back then. It think back then we called it BI.
Today, just about every store (notably not Publix in Florida) has a savings/loyalty card. What do they do with that data? Well, not much really. They don't use it to make it easier on customers. Only Walgreens (that I know of) gives you points for usage, which makes it more of a loyalty card than others.
One use is to put as many aisles in your path for milk and bread as possible, so you will buy more. The Consumption Economy in full swing.
If you have children, the grocery store or the pharmacy now have landmines of sugar and toys at every turn - stocked at a child's eye level to get mom to buy (spend more, raise ARPU)
It is difficult for a store to get as personalized as Amazon, whose website knows you, your shopping habits, purchases, wish list, history, etc. But seriously they don't try.
Another issue for stores is that for the most part stores are just distribution for brands like Coca-Cola, Pepsi and Procter & Gamble. Like Tech Data or Ingram, retail stores live on a thin margin. Grocery stores see about a 1-3% margin on grocery items. Co-marketing dollars are how they survive and advertise.
Obviously, the store has high margin on its own branded stuff and non-grocery items (like magazines, batteries and impulse buys).
Everyone talks about Big Data, but what is the store supposed to do with it? Mail you coupons to push you to higher margin products? They kind of already do that in bulk. Plus do you think Pepsi wants Publix pushing Coke? Who do you think pays for the coupons and sales? The manufacturers.
Not much different from Tech Data really.