Comment by basseq | Hacker News Reader

Comment by basseq | Hacker News Reader

Parent Full thread

Even if quality is there, how ripe do you want? Are you going to use those avocados tonight, or later in the week? Do you like green bananas or ones that are turning brown?

Let alone the "what looks good today?" or "what's on sale today?" method of shopping!

Even packaged goods are tough. When I'm in a store, I know quickly which product I want. I may not remember the brand name, but I can scan the shelf and find it.

Online, I have to sort through a myriad of options, not to mention sizes and variants.

BBQ sauce, for instance, returns 58 results on Peapod. Sweet Baby Ray's, which I like, is 22 of those. There are 13 different variants (original, hickory & brown sugar, honey, buffalo sauce, sweet'n spicy, honey mustard, honey teriyaki, sweet teriyaki, sweet vidalia onion, maple, sweet chili glaze, honey chipotle, no sugar added) and 7 different sizes (14oz, 16oz, 18oz, 18.5oz, 28oz, 40oz, 80oz).

Holy crap!

I know what the label looks like, and I know the bottle is about yea-big. The images online are 200x200... I can barely SEE the label.

Then, what if what I want isn't in stock? Is it going to appear on the page? Am I just not going to get it? Would I get a substitute, as I might if I was in the store?

By the time I do that mental exercise for 50 items, I might as well drive to the store.

These are hard challenges.

One thing that might make a difference is integration with club cards. Giant knows what I buy down to frequency and UPC. Why haven't they done better integration to link my in-person buying with encouraging me to buy the same items online?

For me, the wish is for scripting. Most of my buying is algorithmic, but the filters don't match my algorithms.

It is difficult to express, "Please give me the cereal that I like the most that is under $X/lb. If there is an exceptional deal, buy more." without an API or scripting.

Which, if you break it down further:

1. Everyone's "algorithm" is different. Even if you could bake every possible filter into your eGrocery product, can you imagine the UX nightmare?

2. The "that I like the most" is a tough thing to quantify. Who is "I"? Me? My wife? My kids? My mom who's coming to visit this weekend? What do I like? Specific brands? Things I've bought before? From that store, or a different one? Things I might like because they share characteristics?

That's why user-scripting makes sense!

The downside, for the vendor, is that I'd also wind up scripting price-searching at multiple local grocery stores for equivalent items. If I can just curbside pick up what I need at both stores, they're literally next-door to each other.

The irony is that if those two stores were to cooperate and work together on a unified scripting system, both would gain sales on average; yeah, if an ear of corn is 50¢ at Safeway and 75¢ at Raley's, Raley's would lose that sale, but if Raley's happens to also sell mayonnaise for $1 v. Safeway's $1.50, and I can order both at the same time to fruitlessly attempt to satiate my insatiable hunger for elotes, then both would get a sale (and same deal if one or the other is cheaper on the chili powder and/or cotija).

We used to have such an API in the UK with Tesco, but that looks to have withered: https://www.tescolabs.com/category/api/

This is like writing a cost function for a global optimization algorithm for a problem with tons of parameters. You will very soon find out that the optimizer is much more adept at finding highly rated, but extremely useless outliers than you are at writing a competent cost function.

Herein lies the path to coming home to a cart filled with very-cheap-cause-it-expires-tomorrow produce and no tooth paste because it didn't make the cut under some top dollars limit you introduced to fix another optimization loophole.

I mean, this sounds like every other day at the store for me.

I’ve noticed Wegman’s has recently been separating fruit by “ripe today” and “ripe in a few days” and it’s really great. Especially for things like peaches and avocados that seem to have a pretty limited window between hard as a rock and rotten.

Wow, that does sound great! Are they doing that in store only (like separate bins?), online, or both?

They’re in separate bins at my local store and each bin has a sign. Since I know which fruits they separate out like that I just specify when I order online. They even put a little sticker on the “ripe today” produce if you order a mix so you know which to eat first.

"Let alone the "what looks good today?" or "what's on sale today?" method of shopping!"

That one is really important with fresh food. Maybe the lettuce looks good or the cucumbers, who knows?

This could solved with a little more thought from those doing the pick and pack.

If someone orders 3kg of bananas or a dozen avos - mix it up with some ripe now and some on the green/firm side.

The biggest issue we had was out of stock items. If you're in the store and a key ingredient is out you decide on a replacement meal which just isn't possible online, as there was no guarantee that what you ordered would still be available when they got through to your delivery.

Because they instead decided to invest their tech money into autonomous floor cleaning robots, if the Giant stores near Harrisburg are any indication.

Reply on news.ycombinator.com