Last fall, I wrote a long piece for Newsweek about big-box retailers, in which I concluded that Amazon.com had basically made their business model unsustainable; the only question was whether they could switch to a smaller, more curated business format before Amazon put them out of business. Now it’s time to ask whether grocers should be watching their backs.

Yesterday’s Heard on the Street column at the Wall Street Journal discusses Amazon’s push into groceries -- currently operating only in Seattle, but expected to expand into other major cities soon. How worried should Safeway and Harris Teeter be?

To answer that question, let’s start by assessing Amazon’s immense strengths as a retailer. To my mind, those strengths are twofold: scale and relentless distribution excellence. Yes, its Web technology is excellent, it has great strategic intelligence, and it has some pretty good purchasing teams. But the things that other companies can’t replicate are its scale and its warehouse operations; you don’t have to be that much of a merchandising genius if your store carries, um, everything.

Big-box retailers got big by putting huge stores on cheap land, offering customers one-stop shopping for an entire category. Amazon got bigger by offering its customers one-stop shopping for almost everything, from even huger warehouses on cheap land in the middle of nowhere. Previously, only a small number of catalog sellers and home shopping networks such as QVC could match that kind of scale, but these strategies had various limitations (the effect of Amazon’s emergence on QVC’s strategy and operations is itself a fascinating subject for a different post). Amazon almost matched the convenience of a local big-box retailer, at lower prices and without you having to leave your house.

But not in groceries -- at least not yet. Amazon’s distribution expertise is in warehouse management, not shipping -- shipping is handled by UPS, FedEx, and LaserShip (with increasing input from Amazon, to be sure). UPS and FedEx don’t do grocery deliveries, so that means Amazon will have to either do it themselves or partner with someone who’s already expert at those kinds of logistics. Opening a grocery delivery service involves a whole new set of skills and trade-offs that Amazon hasn’t faced before. For one thing, while the change in Internet sales tax laws means that Amazon will no longer take a big financial hit by putting warehouses in high-tax states, it still has a cost problem: You can’t just pick a nice piece of inexpensive land four hours from anywhere anyone has heard of, because then you’ll waste a full workday on the round trip. For another, Amazon doesn’t sell a lot of products that have to be kept carefully temperature controlled and then tossed out after a few weeks if you don’t sell them.

Grocery delivery also poses a price challenge that Amazon hasn’t faced before. It’s pretty easy for Amazon to undercut Best Buy, not only because it can forgo the expensive leases in shopping malls, but also because its labor model is so much more efficient. Best Buy clerks get paid to spend a lot of time standing around the store, scanning the floor for stray tendrils of dust and the shelves for stray items that have gone out of stock or were put back in the wrong place. On a good day they go into the back and fetch you a pricey television or appliance maybe five times; their productivity is constrained by the number of customers who come in. Amazon’s warehouse workers, on the other hand, have amazing -- some would say inhuman -- productivity. Every minute they are working, they are making money for the company.

Amazon can’t produce this kind of cost differential in groceries, because grocery stores rely on free labor -- yours -- to do all the laborious bits of fulfilling orders. You wander around the store with your cart, pile it all in, compare the prices of various brands of frozen chicken tenders, and these days, you may even be your own checkout girl. Then you drive it home to your house, unload the car and put the groceries in the kitchen. In order to deliver your groceries, Amazon has to pay someone at least $10 an hour to do all that work for you.

Of course the grocery store, like Best Buy, has to pay folks to keep the shelves stocked and the store cleaned. But fewer people than you’d think: The big commercial food companies do a lot of product shelving these days, especially for those fancy displays you see at the end of aisles. And a grocery shopping trip is vastly more labor intensive than almost any other kind of retail; when I interviewed Peg Merzbacher of the Internet grocer Peapod for that story, she told me that after long refinement, it takes 20 minutes to “pick” a $150 order. And that’s from their stripped-down “warerooms,” which carry about 11,000 of their fastest-moving items (compared with perhaps twice that in a typical store). When they tried to pick the orders from the full store selection, that same order would take an hour.

After you’ve picked the order, of course, you have to pay a driver to take it over to someone’s house and deliver it. So Amazon starts off with a built in cost-disadvantage compared with standard retailers -- the opposite of the situation they had with big-box stores. There are ways that you can make up those disadvantages, of course. For one thing, its warehouses may not be far out in the hinterlands, but they can be placed in suburban locations that are cheaper than an urban grocery stores. And Merzbacher told me that delivery services suffered vastly less “shrinkage” and “spoilage.” Shrinkage is when someone steals your stuff; spoilage is when food goes to waste -- either because you didn’t sell it or because an employee or customer somehow ruined it.

With a Peapod order, they know exactly how much deli meat they need, how much potato salad to make, and so forth. And customers can’t pull a pack of chicken breasts out of the refrigerator, carry it around the store for 40 minutes, then decide they don’t want it and leave it in the dairy section, forcing you to throw it away. That’s why ordering your groceries from Peapod is a bit more expensive than going to the store yourself, but not outrageously so.

Still, it’s not cheaper, is it? And unlike most of the products that Amazon sells, people really do often need stuff right this minute, not two days or a week from now. That’s why Merzbacher is skeptical that online groceries pose an existential threat to bricks-and-mortar grocers; she views them mostly as complements. You might have your weekly delivery that comes at a convenient time and contains all your staples and heavy items, and then drop in at your local store for a few things that you need to finish dinner.

That’s presumably one of the reasons that Merzbacher didn’t seem that worried about Amazon. There’s a sort of paradox at the heart of online grocery: The grocery business requires big scale, because the profit margins are razor thin, so you need a lot of volume to really make money. However, because the margins are razor thin, you have to be really, really careful about spoilage. If you order too much food and don’t sell it, you’ll hemorrhage cash. That’s what killed early Web grocers like Webvan.

Peapod was well positioned to handle that challenge because Royal Ahold, which owns it, also operates grocery stores, like Stop-and-Shop and Giant. In the case of Stop-and-Shop, the “warerooms” are actually in the stores, minimizing the risk that they’ll end up overstocked with a bunch of food that goes bad. But they still have to pay very careful attention.

“It’s a tricky business,” Merzbacher told me, “not to be entered into lightly.”

Of course, I’m never ready to count Amazon out: For one thing, it doesn't seem all that fixated on actually, y’know, making a profit; and for another, Jeff Bezos is a strategic genius with a relentless focus on flawless implementation. The pilot program in Seattle, Amazon Fresh, seems to be doing OK, and, according to the Wall Street Journal, it’s spurring add-on sales of other items.

But I do think that moving into the grocery business will be the biggest challenge that Amazon has taken on in a long while. I will be watching avidly to see how it manages it.