My two Echo Dots 3 had an aura of technical immaturity from day one, but I always viewed them as inexpensive minimalist audio players with some extra functions. They were on sale for like 20€ when I was looking for a simple Bluetooth speaker and I am happy I gave them a try.
* Spotify across the apartment due to the seamlessness of Spotify Connect. No complaints about sync or sound quality, works beautifully.
* Audiobooks or radio while unable to control smartphone (bathtub, exercise).
* Quick access to radio stations, news shows (the ones it understands at least :) ).
* Simple automatic routines like a brown noise playlist during sleep or wake up music, or turning the lights off at night.
* Elderly people in my family who don’t understand tech love theirs to request favorite music via Spotify, set kitchen timers, play digital radio, get a morning greeting. It’s an extended kitchen radio for them and in heavy use.
Would not call this a failed product, perhaps they should focus on core functionality, and improve the app (that I fortunately rarely use).
The "colossal failure" in all the recent headlines means "colossal [financial Return-On-Investment] failure" because customers are not shopping with Alexa voice commands.
A lot of consumers do like Alexa and millions have bought them because conveniences like turning on the lights and playing music are desirable -- but those "smart home" type of uses are not increasing revenue for Amazon.
Amazon was betting on more "Alexa, buy AA batteries" (i.e. spend money) instead of just "Alexa, turn on the light". With more sales enabled by Alexa, it would have justified the billions spent on developing the voice recognition infrastructure. In that monetization sense, it's a "colossal failure".
Amazon isn’t good enough for buying things on Alexa to seem reasonable. Many users barely trust the web interface to show legitimate products — how is voice control supposed to make any sense?
I would believe that Whole Foods could pull this off. Or maybe Target, or more generally any store that competently curates their inventory. That’s not Amazon.
This lets them shill for whatever gives them the best margin directly, without having to present the user multiple products.
I guess it turns out that people tend to want a bit more control than that. Especially considering that when I tell an SO or friend to buy something I can be confident they'll act in my interest, while I can be fairly confident that Amazon will act in Amazon's interests when told to buy me something.
Yeah this is it. I'm not sure how they actually attribute value and if "Alexa, buy batteries" counts to their P&L or the e-commerce side. Either way, I'd assume the same as you that they never intended to turn a profit on device sales but rather as a loss leader for ecomm and it's probably fallen short.
The listening capabilities are probably the most valuable and my grapevine says they're pivoting to buy up audio content (ie Wondery).
The failure is not in the product, it’s an excellent product.
Their problem is their high expectations to make it a big business. Same way Google kills good products willy-nilly because of unrealistic expectations or inability to spin it off into a self-sustainable project that builds good will with the parent brand.
Whenever Alexa tries to get me to add things to my cart or search products I roll my eyes (it’s never intrusive or annoying, they know better) so I can see why it might not tie in well with their primary business or any added revenue streams.
But they can make money selling the devices or on the App Store - they totally suck at promoting apps and educating users on new usecases. Especially if their AI isn’t going to improve like it hasn’t in the 5 years I’ve used it (if they are investing in the AI side heavily I haven’t seen the ROI).
I'd describe it rather as a failed platform, and I point the finger directly at the awful integration experience for developers and users alike. Alexa's skills ecosystem is painful to engage, being bureaucratic, riddled with misbehaviours, janky UX, and breaking API/data changes - all deficiencies whose consequences are pushed back on the consumer and integrator to handle.
The fact my US-made Rainbow Echo Dot stopped working entirely when moved overseas is a testament to the Alexa service team's pernicious gatekeeper mentality. Don't even get me started on the baffling inconsistencies in behaviour exhibited by Alexa when embedded e.g. trying to control my Sonos kit.
I am especially dissatisfied having been the recipient of a pre-production first-generation Echo courtesy of a visiting BDM whilst managing an AWS team in Melbourne. Super excited at the time, super disappointed now.
Perhaps I should've paid more heed to a key early warning of a restricted and rather parochial outlook: in the first year or so of operation it was impossible to register an Echo with a service address outside of the USA, and I could only set an overseas timezone by writing my own configuration front-end. Even for a MVP, in hindsight, that was a red flag.
That quote in the article, "Alexa is a colossal failure of imagination", sums up my feeling (I am not the former employee quoted).
In this time of too much complexity in apps and user-interfaces it is great that Alexa allows you to accomplish some simple things simply: Alexa play some jazz. Alexa set alarm in 10 minutes.
Those are the two simple things Alexa is good at. Then I also pay for Amazon Music. I don't know if the Amazon profits counts in profit from Amazon Music but without Alexa I would not be be paying for Amazon Music .
Agreed, I bashed the AI R&D but really I should be bashing the UX/dev environment. Adding new workflows and finding new apps should be 10x easier. Instead of asking me to add things to a cart they should recommend new ways to use the product.
Amazon has totally undersold the capabilities of their devices to users.
Idk if google is any better. I use Amazon because I love the firestick and I’m a big Audible user.
Indeed it is. What Alexa can do today was science fiction 20 years ago. I can say "Alexa, order Earl Grey tea" and a package of Earl Grey tea will show up on my front door step tommorow. Yeah, its slower than a Star Trek replicator but essentially the same thing.
I personally love it but I’m a nerd who is supposed to like it.
What sold me was hearing about my 9yr old niece asking Alexa random questions at 1am because she couldn’t sleep and then asking for rain sounds. And how everyone in the family knows she’s the weather expert because she asks Alexa the weather every morning.
That makes them failed products from a business viewpoint.
Of course, they would become successful from a business viewpoint if they could sell enough of them at (cost + reasonable margin), but the article doesn’t discuss whether that would be possible.
I barely trust that I'm going to get a decent, non-knock-off product when ordering from Amazon manually and carefully inspecting the listing. I can't imagine letting Alexa just choose for me. Even if I could set up a specific listing of X for it to buy when I say, "Alexa, buy X" I wouldn't be willing to assume that the listing would remain trustworthy over time.
Building a speaker with mic and some ML does not cost 10B dollars. Above is a PR piece to justify layoffs and restructure of Amazon. You want to make sure that engineers in AWS are not leaving when you want to fire some people in other places. You can trim fat without losing critical people.
I worked on a backend service for Echo for a bit so was exposed to the internal stuff. There are a ton of people behind this project, as in, lots of things were done manually. There is a PM for "dog facts", for example. He makes at least 250k in total comp per year. His cost to the company is at least about the same (for simplicity), so 250k/year is spent just on dog facts (and I'm guessing some other fact categories).
10B would then pay for 5000 employees for 8 years (echo was released in 2004). My impression is that there were more than 5000 people working on Alexa between devs, QA, PMs, TPMs, managers, designers, etc, so I'm not surprised by $10B at all. In fact, it seems low after doing the calculation above, but people do switch projects or are shared so maybe it just seemed like that to me.
The article claims that the devices are loss leaders. If so, device revenues are smaller than what it costs them to produce them, sell them to customers, and deliver them.
“loss leader” is not explicitly defined. It’s not clear at all if the devices are sold for less than their marginal cost of manufacturing or their total cost of manufacturing (ie including R&D).
There are more than 20,000 people working on Alexa in Amazon. That's more than 5B in salaries a year.
I worked with a guy that switched team, coming from an Alexa team dedicated to supporting jokes on the devices. It was dozens of people dedicated to just that, and that was a SMALL thing.
The problem is voice isn't a very good interface unless it's absolutely perfect. A standard GUI reminds a person the state that an application is in and allows someone to back out of a given decision. Another person can do that with voice but it involves being aware of where the person you're talking "is at". Even the most fluent voice interfaces aren't close to this.
I used to use the Alexa in my roommate's bedroom. By use, I mean that I would yell, "Alexa, add $RANDOM_ITEM to cart/wishlist" through their closed door and it would.
It was a funny prank but I was astonished to see that it wouldn't do any sort of voice detection and if I said "buy items in cart" I'm sure it would have done it without any biometric confirmation that I actually was who I said I was.
My aunt is blind. Alexa is a game changer for her. She doesn't do a ton with it, but asking questions, getting news and weather, and accessing her audio book library are some things she uses it for. I hope amazon doesn't pull the plug completely, for her sake. Or, if they do, someone picks up where they left off with a focus on people with disabilities.
I think this surfaces a bigger issue in that voice as an interface to our computing devices is a dead end for the majority of situations. Note - I’m not dismissing accessibility where there is huge potential. Also note - I’m talking about fully developed voice interaction on par with keyboards, touch screens, mice etc, not just “turn on the lights” or “play the Beatles”.
For a new technology to succeed it needs to be better than what already exists. Voice is absolutely not that. Compared to a keyboard or a GUI it is clumsy, in-precise and cumbersome. It is not convenient in a huge number of situations, for instance in places where other people are around, like offices or open spaces, or noisy conditions. It presents privacy issues that tactile interfaces do not.
In most cases voice acts as an additional input method to one that already exists, and as such adds cognitive load on the user, they now have two very different ways of achieving a given task, so there is an added initial choice, plus then having to think about how to achieve the task this way. A lot of tasks will not be possible via voice for a long time, so there is that added decision as well, can I even do this with voice.
People will say, ah but that’s not a problem with voice, it’s due to the immaturity of language processing. When we can more accurately understand what the user wants, voice will become a viable interface. There is something to that, but we are a long, long way from there, and it still doesn’t address a lot of the issues I raised above.
Apple are often chided as being way behind in voice. I’m beginning to think that they realised these problems a long time ago, and have decided not to put the effort in, apart from presenting it as an accessibility feature.
I don’t know about others, but our Alexas became really hard to use lately. Basic commands like “Alexa, turn on the lights” or “Alexa, play Frank Sinatra” get misunderstood and need to be repeated multiple times. Combined with unsolicited “by the ways”, we are generally unlikely to ever buy another one.
I think smart assistants could be a useful gadget eventually, but alas this isn’t it.
We have Alexa. The kids use it to control the television or set timers.
Among the challenges, as with Google Glass, is that the target audience that wants to live in a Jetsons house awash in gadgets is really not that great.
You do your job and are frustrated by tech all day.
Do you really want to come home and fight to get this Monkey Fighting puck to understand you? Why, no: no, you do not.
Why did they only focus on hardware/software control and totally miss the digital assistant line. Do you remember Bonzibuddy? Everyone seemed to love that, but Alexa has no equivalent. Also, those hologram AI displays been out in japan for a decade, and nobody made a US version for Alexa. https://www.gatebox.ai/en
I couldn't see using an echo when I had my phone/pc that does everything.
Now an entry into the home could have been a digital assistant on the PC, Hologram, tv or app. So many ideas for profit streams, open API for 3rd party digital assistants with fees attached, alexa tv's running an app with a visual digital assistant, run the app on a tablet and see a digital assistant.
Even if it was a novelty, a little animated person standing on the countertop is way more interesting than speaking to an echo.
It's amazing that Google, Amazon, and Apple, with all of their billions of dollars and research teams couldn't train a speech recognition model that works reliably well most of the time regardless of accent or background noise or even without any of those 'problems'.
Then OpenAI with its whisper model showed it is very much possible. The whisper model is so good, I sometimes feel like it's magic. I have tried all sorts of situations, non-native speakers, severe background noise etc, and it still does incredibly well both for English and Arabic (these are the only two I have extensively tested).
It's unfortunate, but I think the reason all smart speakers are failing to bring in profits is because IoT never really took off.
If everything in your house (curtains, standing desk, toilet paper roll holder, etc) was a connected device, think about the automations that Alexa could enable. Ask Alexa to auto raise your desk just before meetings start, to notify you if the toilet paper on the holder is getting low, and so on. Amazon could easily get creative with charging for these conveniences.
It doesn't matter how many were sold if they were all sold at a loss and they don't lead to increased revenue because people only use them as alarm clocks.
Have you ever actually used one? They're shit tech. A new gadget to market to American consumers who have no idea what to get their adult children for the holidays.
All the devices I know of in my immediate family are collecting dust. With the exception of my great aunt and uncle, who have one in the living room and constantly complain about it/at it.
I have a love hate relationship. Saves me time to turn off my space heater, or the living room TV when my kids are in trouble and keep turning on while grounded, or lights, etc, or google some mundane fact my wife and I are arguing about, sometimes getting a laugh when she TOTALLY gets the wrong thing in a funny way.
I curse her out more than anything but at this point she's almost a family member, I do wish they'd hurry up w/ a new generation using GPT3+dalle etc to create some much better ux.
I'd probably use a google home instead if I'd gotten that first, I'm not tied to the alexa brand, and hate Amazon as a whole as being a gluttonous monstrosity of industry, but I'm a big believer in global warming but still eat meat... Sometimes you just do what's convenient, I guess.
The Dot 4 speaker is a decent speaker. The voice recognition is also pretty impressive for a device that small. So, I disagree it's shit tech. But if it was my job to make money from it I would be sweating bullets.
I agree. I saw no data in these posts. Alternatives could be that they developed a deep learning algorithm to auto label intent and entity? Or possibly interest rate changes drove the stock down 50% and they need to get more free cash flow? Or they hit a glass ceiling in cpe?
"it would be simpler if there was a general language to learn, for how to communicate with it- but instead, it's just bad natural language processing around an undocumented collection of abilities, where you have a new learning curve for every different function"
I think the devices would have more adoption if they were open (hackable). I think because they have a locked bootloader - their use it limited by Amazon - when it could be 100x that by letting people figure out how they want to use them.
Not surprised at all, I only used one of them for a few months, the rest 4 along with that one used are all collecting dusts over the years, it's useless, my phone does everything alexa does well and better.
Voice control clearly has potential applications and merits, but in it's current iterations, does not seem to have a so-called 'killer app' or feature that would compel mass adoption.
If Amazon changed echo to a subscription service or a prime extra like prime video. Would that change the economics associated with it? Would you subscribe?
