Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Amazon’s Battle to Break into the Grocery Market

bloomberg.com

Post reply on HN
Print view View original

31–40 of 112 posts

Re: Amazon’s Battle to Break into the Grocery Market

#31

Post by antognini »

dangero wrote: post #23
Gargoyle wrote: post #5

Earlier quoted context omitted.

I'll be a believer when it can handle a Thomas Crown scenario - multiple people in the same outfit crisscrossing around each other. While a bowler hat flashmob is unlikely, a group of similar dark jackets etc isn't.

This scenario is the same as any other machine learning environment: The output is prediction with a % confidence score. This means the threshold can be set such that worst case certain people need to be checked out by hand. Fallback to standard checkout means they still cut down quite a bit on the needed human interaction and they can set KPIs around improving those stats.

While it's sometimes okay to interpret the output of a model as a confidence, you do so at your own peril because that's not really what the model is optimizing. There are often cases where the model outputs a very high "confidence", but the model uncertainty is (or should be) very high. This is generally the case anytime you give the model data that is very far (in some sense) from the data that it's been trained on.

Re: Amazon’s Battle to Break into the Grocery Market

#32

Post by Johnie »

Spooky23 wrote: post #18

It's cool that they are trying to innovate here, but it seems like a weird area to focus on. The high growth areas in grocery are things like prepared foods and produce. Owning the middle of the store to try to out-Walmart Walmart seems like a low-value proposition. I've been using Amazon Fresh for a few months. The delivery aspect is awesome, but the rest of the business is pretty questionable. They cannot keep anyt…

>I've been using Amazon Fresh for a few months.

This is an artifact of living in Silicon Valley bubble. The rest of the country don't live this way.

Re: Amazon’s Battle to Break into the Grocery Market

#33

Post by alexmingoia »

jenamety wrote: post #22
Spooky23 wrote: post #18

It's cool that they are trying to innovate here, but it seems like a weird area to focus on. The high growth areas in grocery are things like prepared foods and produce. Owning the middle of the store to try to out-Walmart Walmart seems like a low-value proposition. I've been using Amazon Fresh for a few months. The delivery aspect is awesome, but the rest of the business is pretty questionable. They cannot keep anyt…

>it seems like a weird area to focus on i see it as you the consumer are the focus, not the products you buy. Knowing you as a consumer is the goal to be able to serve your every purchase. I'm really surprised Amazon doesn't have a branded credit card that offers unbeatable rewards, thereby seeing all user transactions.

They do have a branded credit card. You get 5% back on all Amazon purchases and other Amazon benefits. It's a visa I think provided through chase.

Re: Amazon’s Battle to Break into the Grocery Market

#35

Post by jhpankow »

ghaff wrote: post #28
jenamety wrote: post #19

Earlier quoted context omitted.

So if i knew you order toothpaste on average every 48 days, i could put a new one in cheapo snail mail on day 43. who is in the best position to predict this and take it completely off your mind? current grocers or amazon? your scenario presumes you need to invoke the order yourself. Amazon is going to be predicting your needs (poorly at first but improving over time). I'd wager this big picture includes getting audi…

Amazon already has subscriptions. Personally, I don't buy anything with a combination of frequency and predictability that makes a subscription make sense but obviously some people do. (Diapers I assume are one thing.) And that becomes harder with with perishables. Customizable but pre-populated (based on prior purchases) would seem to make a lot more sense.

My SO has signed is up for subscriptions for certain recurring purchases. Our usage is too erratic to put on a schedule. I have a new job at home: finding a place for all the things we get automagically delivered from Amazon that we aren't ready to consume. I finally spoke up when we had three containers of dishwasher soap and nowhere to put them.

Re: Amazon’s Battle to Break into the Grocery Market

#36

Post by trome »

bogomipz wrote: post #29
trome wrote: post #24

Earlier quoted context omitted.

Sure, but who'd actually trust Amazon with their food? I know I wouldn't, they've screwed up 3 out of 5 orders in the past year, leading me to have little faith that they won't give me E Coli covered veggies. They're opening a large, car centric Amazon Grocery pickup location just over in Ballard as a test, but the particular location they picked is a parking lot for 3 to 4 hours a day, and is over 3 blocks from tran…

Can you say what happens when they screw up the order? Is it your responsibility to wait around for someone to come collect the wrong parts of the order?

Depends, twice I've had to ship it back, and once Amazon told me there were no returns and refunded me (Samsung Evo+ 64GB that only did 6MB/s, rated at minimum 10MB/s). That being said, I need to throw that MicroSD card away, whenever I use it in a camera or single board computer instead of a tested Evo, it causes poor performance.

Re: Amazon’s Battle to Break into the Grocery Market

#37

Post by j-c-hewitt »

afarrell wrote: post #30

Amazon needs to get a lot better at selling people complimentary goods. Once I buy something, I frequently see multiple advertisements for that exact product. Why can't they develop a database of products that compliment each other and so when someone buys a kitchen tool, they start advertising cookbooks. Example: "On Prime Day, Members purchased over 215,000 Instant Pot 7-in-1 Multi-Functional Pressure Cookers" http…

They do remarket to people automatically if many people have purchased both product_x and product_y. If you sell stuff to people on Amazon, you could also manually set up a Facebook, Google, etc. remarketing campaign to do just that using your customer emails.

They actually do advertise products to you in your favorite category based on your recent purchases.

In general you see the same product remarketed to you over and over because it's an additional layer of difficulty to add people who have purchased a product from you to an exclusion list. It's definitely more effective to do that (and Amazon does try to do that to you if you either haven't purchased it after a certain period or have purchased it on that particular device), but with any really complicated website with millions of products some things are going to get lost.

>If you have a database of 100s of thousands of people who all are suddenly looking for recipes and raw materials for a pressure cooker, why not advertise to them?

If Instant Pot wanted to break Amazon's TOS and make some extra money selling their Amazon customer lists to publishers, they could probably do that. Otherwise Instant Pot would have to get into the publishing business themselves to get access to those lists.

Re: Amazon’s Battle to Break into the Grocery Market

#38

Post by gwern »

> Workers at Amazon Fresh, the company’s grocery-delivery business, threw away about a third of the bananas it purchased because the service only sold the fruit in bunches of five, the student concluded. Employees trimmed each bunch down to size and chucked the excess.

If anyone was wondering about the bananas, the paper (https://dspace.mit.edu/bitstream/handle/1721.1/99025/9213069...) says (pg33) that after it was identified, they switched to selling 2lb bags of bananas, which the banana growers were already able to provide, so they scrapped the whole exactly-5-bananas-and-tear-to-make-fit.

> In addition to the obvious financial impact of this, many Associates found the process frustrating since it was highly repetitive, and they did not feel they were creating value for the customer

Well, that's one way to phrase it.

Re: Amazon’s Battle to Break into the Grocery Market

#39

Post by bearcobra »

trome wrote: post #24
thedarkginger wrote: post #4

I know this discusses a few strategic plays Amazon is making but shortfalls of the self-checkout project (AmazonGo) are interesting: "In its video touting Amazon Go, the company said it was aiming to open the site to the public in “early 2017,” and it hasn’t provided an update to that timing. But the technology has been crashing in tests when the store gets too crowded and requires human quality control, people watch…

Sure, but who'd actually trust Amazon with their food? I know I wouldn't, they've screwed up 3 out of 5 orders in the past year, leading me to have little faith that they won't give me E Coli covered veggies. They're opening a large, car centric Amazon Grocery pickup location just over in Ballard as a test, but the particular location they picked is a parking lot for 3 to 4 hours a day, and is over 3 blocks from tran…

I've ordered groceries with Prime Now a few times and never once was worried about trust. I'd probably trust their produce more as the people who have also touched it aren't random shoppers.

Re: Amazon’s Battle to Break into the Grocery Market

#40

Post by thedarkginger »

trome wrote: post #24
thedarkginger wrote: post #4

I know this discusses a few strategic plays Amazon is making but shortfalls of the self-checkout project (AmazonGo) are interesting: "In its video touting Amazon Go, the company said it was aiming to open the site to the public in “early 2017,” and it hasn’t provided an update to that timing. But the technology has been crashing in tests when the store gets too crowded and requires human quality control, people watch…

Sure, but who'd actually trust Amazon with their food? I know I wouldn't, they've screwed up 3 out of 5 orders in the past year, leading me to have little faith that they won't give me E Coli covered veggies. They're opening a large, car centric Amazon Grocery pickup location just over in Ballard as a test, but the particular location they picked is a parking lot for 3 to 4 hours a day, and is over 3 blocks from tran…

Since the article stresses repeatedly that Amazon is all-in on produce... I think this is a big question for their ML approach.

One of the annoying parts of not-really-self-checkout that is available in big grocery stores today is that you have to scroll through a screen, choose the correct brand (i.e. Dole Banana), and then weigh it before it prices your order.

Wonder what their strategy is to price variable weight products (i.e. produce) without introducing new friction into the user experience

Post reply on HN