I'm curious if acquisitions impact the rate of growth, Amazon has been on a buying spree these past couple of years.
Search and categories are complete garbage.
The listings that breakdown price (e.g. $/unit or $/Gb or $/ounce) are a inconsistent mess in the same search result, sometimes showing different units for the same product from two sellers!
and I still have no idea if a listing will be OK or from a seller worse thab ebay.
Of course it’s AWS printing their money, not retail.
I've actually gotten used to taking a weekend afternoon for errands or maybe in the evenings on the way home. It's a good way to just unwind from the computer for a while. There are times to avoid certain stores (Central Market on the weekend). Shopping online pretty much means being in front of the computer all of the time.
But yeah, when I die, I want Amazon electronics sellers to be my pallbearers so they can let me down one last time.
If you want to understand a product reliability is pretty simple. Just browse the 1 and 2 star reviews. You can find a lot of valuable information about the products in those reviews.
If you still want to use something to evaluate the reviews use Reviewmeta. It's much better and transparent than Fakespot.
I’ve told my older parent not to buy a product or use a svc based on Internet reviews.. go with personal recommendations only!
I really don't know how good or bad fakespot is, but I get a roughly uniform distribution - it's not at all rare for me to see an A or B, and I'm pretty sure I don't see more F's than those.
I'm consulting in the pet space, if you search for 'Mousr' which is the name of a very cool pet toy brand that they carry - you get PC mouses (ie 'Mouse'). Not the brand 'Mousr'.
So basically their 'spell correct' 'Mousr->Mouse' utterly and completely overrides the notion that the users wanted to in fact input 'Mousr' to get that product.
Instead, in order to find the product, you have to type the company name 'Petronics'.
FYI - if you've worked in NLP, you might recognize this problem: the edit distance from Mousr->Mouse is short, and the 'e' and 'r' keys are beside one another, and of course the word 'Mouse' is very common while the word 'Mousr' is very uncommon, leading the error correction algorithm to 'over correct'.
This is a problem in NLP where very uncommon things that are lexically adjacent to common things get suppressed.
Since 'Mousr' however, is 100% in the Amazon catalougue, you'd think they'd include it!
But no.
Consider the resulting 'lowest common denominator' approach: statistically speaking, it might be slightly more economically efficient to actually assume the error and show PC mice, besides, 'everyone' needs a PC mouse anyhow!
But the interesting, long-tail stuff gets suppressed.
This is to me the statistical epitome of not just Amazon, but of our new, globalized culture - the big brands have such power they just bulldoze through everything else - and the 'short run' economics are in their favour! They can point to 'the efficiency of the search' etc. to validate to everyone else in the equation (ie city councillors approving permits, banks giving loans, buyers - heck, even consumers) to validate their case on competitive, culturally secular grounds.
It takes quite some magic, a loyal following, a magic meme of attention, a super faithful investor, or deep rooted cultural behaviours (like Italians who could care less about the new 'Starbucks' because they already have good coffee).
Most big CEO founders have made some kind of 'big claims' or almost arbitrary statements about 'how it ought to be' and they stuck to their guns on those, I wish Bezos et. al. would think just one tiny bit about the 2cnd order effects of their businesses.
This is probably one of those products that can and should have its own website and sell from there or link directly to Amazon instead.
There are dozens of product results, there's no reason why one of the proper results wouldn't show up as well.
"Results for Mousr ..." "Results for Mouse ..."
etc
Seems more like poor naming on Mousr's part.
I pointed out in that I'm fully aware where the probabilities are, and the fact the 'keys are close to one another' - that's my whole point, it's a statistical suppression.
The 'short view' is to just show 'mice' - but there are problems with that:
1) 'over correction' is far worse error that 'insufficient correction'
2) There are tons of 'Mousr' products out there, i.e. products that overlap closely to more common things.
The likelihood of some random error being a product is low - and since 'Mousr' is an actual product, it makes sense to show it.
In this case, the rational product choice would be to show 'Mousr' first, and to not assume over-correction.
To take a more extreme example i.e. going with the 'highest probability case' ... it might make sense by that logic to show only the most commonly purchases products purchase, almost whatever is typed.
Most recently I was looking for a countertop hot plate/burner to replace one that finally broke (filament burned out after 15 years). Searching within the Appliances / Cooktops category, it is filled with random apparel like t-shirts and dresses for < $1. Try filtering by price to any interval and it's random kids' toys, arts and crafts supplies, tools, and sex toys. You're lucky to see one or two actual cooktops per page. Filter by 'Fuel Type: Electric' and there are only 5 actual results plus an impact socket even though there are hundreds of actual electric cooktops if you filter by brand name or surface material. It's insane.
Most of the sellers and brands have names that look like someone mashed their keyboard. At this point the only way to use the site is to already know the exact brand and model you want and search from google.
You can just use the search from the homepage. Typing "hotplate" gave autocomplete for "hotplate for cooking" which is hundreds of products: https://www.amazon.com/s?k=hotplate+for+cooking
Or if you want to browse, the category homepage has all kinds of filters: https://www.amazon.com/s?k=&ref=nb_sb_noss&rh=n%3A3741261
Amazons web site is a mess for sure, but finding a specific product is not one of those problems. The search is quite good...
Maybe it's true, but is there someone better ? with a better search ? more valuable reviews ? As wide selection ?