They may have changed it, but a few years back I did product feeds to Amazon and had to read the documentation. Basically Amazon will only search the title of the product page, nothing else.

That's why the title of the products are stupid long with everything crammed in. But yeah, you can find anything on Amazon, the site is basically broken unless you know the exact thing you want, brand, name and all.

It's not even that good at searching books anymore.

If I search for Pixel 3 case on Amazon, it will also return Pixel 3 XL cases.

If I search for 1440p IPS monitor, it will mix in some monitors that aren't 1440p, and some that aren't IPS.

I'm not sure if it's because they don't want to solve the problem, because they don't care, or because somehow having customers need to spend more time on Amazon to research their purchases perversely leads to people spending more time on Amazon seeing other items they might also later buy.

Another possibility is search is hard. If all they returned is things with "Pixel 3" well, that includes "Pixel 3 XL". Ok so they look for exact string "Pixel 3". Now nothing matches like "Red Pixel 3 Case"

note, I have never implemented search but suspect it's hard problem.

Still Amazon's search is pretty awful. I've found adding something in the category I want to my cart gets me better results than being on the item itself. In other words I might search for "wrench" and it shows me 50 items of which 2 are remotely close the the type of wrench I'm looking for. I click one, get new page for that product and below it shows other wrenches but they aren't remotely like the one the page is for. So I add it to my cart and suddenly it suggests a bunch of stuff that is actually close to what I wanted.

I also don't understand the pitching me things I already bought. I get pitching me disposables I already bought. I don't get pitching me a monitor or soldering iron or another mouse or keyboard. I'd think either via statistical analysis or via manual tagging they could figure out what items are more likely to be purchased again and not recommend another PS4 just after I bought a PS4? But who knows. It could be this is harder than it sounds or it could also be they suck at search or haven't gotten around to it yet or ...

Pretty sure that Amazon search is bad on purpose... For a while they didnt even allow you to sort search results unless you selected a category (some products can be in multiple or wrong categories so you ended up hiding good results if you selected a category).

There is a 3rd possibility: sellers are completing the product page with incorrect metadata. This might even be down to simple user error eg someone non-technical given the data entry job.

I'm not saying Amazon isn't full of dark patterns and other problems but if there is one thing I've noticed on nearly every online store since the dawn of online shopping, it's that there is always the odd product that has been incorrectly classified. This problem only becomes magnified when you allow 3rd parties to complete product pages.

For me though, the most annoying anti-pattern goes to eBay and how it will bulk out it's results with items from other countries even when you specifically state a local region.

Presumably their data shows them that if they mix in some monitors that aren't 1440p or aren't IPS that customers still sometimes buy those monitors anyway. From Amazon's perspective the search functionality is more of a product discovery functionality. If they only showed what the customer was searching for they would miss out on sales to people who didn't start out with a clear idea of what they wanted.

fine, then come to a middle-ground:

search bar (on the top) for window-shoppers

selection boxes (on the left) for focus-shoppers

If I select 1440p under Resolution, I do not want non-1440p monitors in search results.

The biggest problem with the selection boxes (when they do appear) is the garbage data. In this scenario, you'd get something like "1440p" "1440 pixels" and "1.4KP", sorted nowhere near each other.

Two days later, I'm looking at listings for Pet Supplies > Fish Food, and one of the filter categories is "Number of players".

Audible does this thing where when I search for an author, it constructs a bag of words based on the titles of books that show up for that author, then retrieve based on that bag of words, then rank by like popularity of the returned books. So if a different, more popular author has written books with similar titles (which is very easy given a trilogy with storm-themed titles) you will occasionally see a book by the wrong author as the number one result.

Every time it happens I want to quit my job and force them to hire me to fix it.

Yea i've seen this many times with price comparison engines, i even used to work for one and was head of search back then. We had to actually "tune/down-weight" phone accessories vs phones. Ppl typing in "iphones" wants "iphones". Ppl typing in iphone cases wants "iphone cases". The rudimentary problem is both of the text tokens are in the titles and are popular products. "precision vs recall vs popularity/sorting". We had a similar but lesser problem with books (the titles are so vast and ambiguous). Lol you get a book called 'iphone'. The leading ecomm platform in my country (s.a) has this book problem in search results and mm-tokens still. Looking at you T..A..L ;).

End of the day, search is hard and personal.

> If I search for Pixel 3 case on Amazon, it will also return Pixel 3 XL cases.

In reverse, someone looking for a Pixel 3 XL cases but only wrote Pixel 3 cases, would have been much more pissed than you were if it was has specific has you want.

I'd like them to add country of origin to the facets, but that would allow me to eliminate 90% of Amazon so they never will...

I've noticed a trend lately where items will "ship from the US" with an estimated 2-3 day delivery date only to be delayed at the sender, eventually "ship" a month later, and arrive with international postage a few days afterward. I'd love to filter that kind of crap out as well, but I'm not sure how Amazon could go about actually implementing it.

By tracking actually delivery data and displaying some aggregated statistics on the product page?

That's happening on Aliexpress and other places too - I've wondered about this. An aliexpress seller said ships from US, arrives within a 5 days or free, then 1.5 months later it finally arrives and they get dispute overturned because it was delivered. So pathetic. The stupid postage gives away their "shipped from US" game.

Read the fine print, and launch the dispute right away. Oh, and look at the seller's reputation/history.

But maybe I just wasn't affected because I don't even look for locally shipped items, as I'd be better off going to B2B-focused niche retail locations in the rough vicinity.

I think the EU requires Amazon to list the seller's country of residence.

There's a difference between listing it and being able to filter it in the search results. I'd like to remove Alibaba that gets blended in, essentially.

Country of residence is a vague and unenforceable concept.

Larger issue is getting the data honestly.

In reality, it will eliminate 10% of say two categories and maybe 100% of 5 others.

Amazon's search engine is pure garbage and for a company this size, and with their status, it's quite embarrassing.

Just a quick example: I saw some nice furniture on Amazon by a brand named Rivet. Turns out this is one of their in-house brands. I wanted to know more about their other products so in one of their listing, I clicked on the brand's hyperlink, expecting to be redirected to the brand's page...

But nope, I landed on a generic search page with, among other things, metal rivet items! Truly mind boggling. So in the end, I had to use Google search to find the actual page.

From having spoken to people who worked on object clustering for Froogle in the past, I think the base issue is that it's surprisingly hard to extract all the metadata accurately from descriptions. Even if you provide an file format and allow merchants to upload all their item descriptions with structured data, you'll still just end up getting bad data from a lot of merchants, and then you have the problem of whether you want to accept the inaccurate data provided by the merchant who has an incentive to misstate things, or the inaccurate data provided by your flawed automatic text extraction systems.

I'm not sure if there's a good workaround for this besides buying from places specializing in the specific types of items you're looking for, which comes with its own set of tradeoffs in price, convenience, etc.

> you'll still just end up getting bad data from a lot of merchants,

The last few times I bought anything computer related on amazon was just that. Hundreds of entries with bad and probably copy pasted technical data. They would have to vet the data at least randomly and ban problematic sellers to make their search even remotely usable. This is one area where specialized online stores outshine Amazon and it will probably remain that way.

> I can't find an A19 LED light bulb with 2700K color temperature and CRI 90 or better.

https://store.yujiintl.com/collections/residential-lighting-...

I am a happy customer of this vendor. I bought all the misdescribed hard to find stuff on amazon first... what worked at all failed quickly. The yuji stuff seems a little pricey, but factoring in defect rate it isn't.

Mostly I've purchased 2700k high density strip (now probably several grand worth), but I've also bought a few cases of a19 bulbs and am also very happy with them.

I can confirm. The one time I had some of the LEDs die is when I tried to hack a fan-less 150 W LED desk lamp together form 9W COBs thermal-epoxied to a trade fare showpiece from a Chinese streetlight/factory-downlight company, that they used for 200 W. But I had it upside-down and without a heat spreader layer, which I blame the LED's death on.

Other than that, I'm happy with a 100 W, 5 m long, 1800 K, 95 CRI strip with a 15~16 kHz PWM dimmer, and my collection of the high-power remote phosphor bulbs and my share of a box of pseudo-filament bulbs. We're looking at bi-color white+white (flexible CCT) strips on top of the cable trays, illuminating the ceiling, for our renovated hacker space.

For reference: YUJI's "normal" "BC" double-phosphor 95 CRI LEDs are about 1 USD/W, sold in packages typically worth 50~150 USD.

Weird right?

You would assume that Amazon allows you to properly slice and dice their product catalog but i think that is something we do but the avg person doesn't.

When i need something and i know exactly what it is, i will search for it on amazon and other retailers to find a good price, most of the time with comparision services.

If i don't now exactly what im looking for i will research the market: What is a common producer/brand, who is known for good quality, what variations exists, what was the model a year before, whats the price difference, whats the feature difference. I don't think thats the normal behaviour.

People like to go to electronic markets to 'browse' and get advice.

I miss Amazon before 2010. Or the Internet. It was so easy to find great things, real reviews, etc.

The irony is that Google search does a much better job searching products on Amazon.com than Amazon does.

1000bulbs.com