Amazon Alexa is a “colossal failure,” on pace to lose $10B this year | Hacker News Reader

Amazon Alexa is a “colossal failure,” on pace to lose $10B this year

arstechnica.com

52 pointsarunbahl56 comments

The point about the vast majority of Alexa interactions being trivial and repeated things that aren't monetizable rings true with me.

When I've used one it's been for exactly two useful things: timers while cooking and weather, and one frivilous thing: jokes. It's valuable enough for those things to be honest, given the purchase price, but yeah there is no feasible way to make ongoing revenue from them.

I just can't think of anything else I want to use one for, if I could I probably would have.

I never understood the appeal of those things. Everybody who own one of those already own a smartphone whith the exact same voice assisted functions.

I think there is something to the idea of a beacon in common rooms rather than a cell phone.

When i'm at home I rarely have my phone with me. If I am in the kitchen and I need a timer initiated it's simply easier for me to speak the command out loud than it is for me to traipse to the next room for the phone.

Now, if the argument is "I don't know why we need so many competing services that do the same thing." , I agree.

I wish that there was an Echo equivalent that just acted as a dumb thin-ware style client connected to my mobile phone via wifi or whatever that was there only to act as a microphone bridge, with all the heavy work being accomplished by the mobile phone.

I use timer buil in into cooker or oven. Not voice activated, but super practical anyway.

Man, and here I am just looking at the wall clock every once in awhile. Honestly if you just stay engaged with the world around you and mindful while you're doing things you develop a strong internal sense of time-passing that makes external crutches unnecessary for most tasks. With a little practice you can get to a similar point with the weather too.

Or, I can set in timer and then go on happily daydreaming or solving puzzles or reading HN. Imo, I win here.

How is an overexpensive electronic waste device an improvement over a simple mechanical timer in your kitchen that will last decades or the builtin timer/programmation of your oven?

I mean if the only use of an amazon echo is having a timer, getting weather forecast and playing music this is a really bad financial, privacy and environmental decision.

Anecdotally, having raced my friend who uses Google Home extensively, it is much faster to set a timer on my Pebble by pushing a few 0-latency buttons than to navigate the verbal protocol demanded by these devices (trigger word, pause, "set a timer for <x>", pause, machine verbally confirms timer before starting it).

(On a laggy Bangle.js the race would swing the other way. Pebble got so much right.)

I like my bangle.js2, but I loved my pebble. If only the bangle had more physical buttons.

I used to change the lights sometimes.

Lights. Music. Fart noises.

Well, I worked at Alexa in the past. For the jokes you like, there was a team of about 10-12 developers plus 3-4 product managers, and BI infrastructure. They had their own speechlet. This group cost what, $3m/year to Amazon, more or less. From here you can start seeing why Alexa would never be a profitable enterprise.

My fondest memory of Alexa was when its VP, Tom Taylor, came to my site, pulled up a blackboard and started jotting down the math of how much it should make to be profitable. Then, to our astonishment, he solicited the crowd for ideas on how to monetize the platform.

This feels like one of those things where if they had just charged more per device and tried to make it as useful/unobtrusive as possible it would have been better than trying to use it as an ad/purchase machine and subsidize the hardware. I like my Echo and would have paid more for it, especially if it wasn't always trying to sell me things via Amazon Shopping and I trusted it to not record me for advertising.

This. Just make it a voice controller for music, lights, timers, and reminders. Everything else is fluff.

I threw out my Alexa. I tried, I really tried, but it’s just too fucking annoying.

Wake up, and then: “Alexa, good morning, what’s the weather like?”

Forecast follows, which is good, but then she unfailingly launches into DID YOU ALSO KNOW you can use the Alexa app on your phone to do this and that etc etc. it’s unbelievable. Why am I being served ads from an assistant in the first place? Why can’t you just tell me the temperatures and shut up?

Anyhow, I just ask Siri now, which does what I actually want, every single time.

Is it me or is it a bit embarrassing to use voice commands? Not just echo but also my phone and car.

It is. I don't think I've ever used voice commands in public.

$10 billion seems like a lot a money for an 8 year old product that hasn’t changed much.

Hardware is incredibly expensive, and Amazon is selling most of it below what it costs to make it.

But yes, more than $1B a year seems high.

Their hardware division seems cursed. I wonder how long that vacuum robot acquisition will last.

What's the objective of running these loss making businesses? If it's to funnel users to Amazon's e-commerce store, that doesn't seem like a great decision since the e-commerce store doesn't make money either. And I can't see enterprise cloud decisions being swayed much by Amazon's consumer business.

Spending money on a loss making product so that you can funnel people to another loss making business seems...counterintuitive?

Amazon (the marketplace) invests it's revenue aggressively, although this serves as a loss on paper there is still a benefit to generating that revenue in the first place as it serves as investment dollars for other parts of the business

Sometimes it's to capture the market: kill all competitors with a price they can't match and then jack up the price once you've achieved monopoly. Kind of a gross practice that is only possible when you've got a huge budget and no need to self fund.

It is the original playbook for many startup and it all started from Amazon?

I think Kindle might be the ideal situation for this kind of strategy, as I imagine ebooks have healthy margins, the store lock-in isn’t total or even that inconvenient but it’s enough to funnel sales their way most of the time, and I imagine the continuing development costs are minimal compared to a more full featured tablet or a program like Alexa. Unfortunately most things aren’t an easy commodity to work with like chunks of the written word.

Yeah I’d love to know where the money is actually going. Is it just an accounting thing? You could pay 40,000 people a quarter million a year!

I think they're including what they would charge someone else for compute and traffic, if someone else ran Alexa.

Thank you so I guess it’s partially an accounting thing and they aren’t actually shelling out 10B in cash on it.

That makes sense. The credit card invoice with nearly 1 billion dollars every month must be the deal. :-D

Just to train the ML models used in the NLU, ASR, and other ML-based platforms, I heard the number of $150m/year when I worked there.

> Google Assistant at 81.5 million users, Apple's Siri at 77.6 million, and Alexa at 71.6 million

That's a lot of scale before figuring out there's no way to profit!

Yeah. And Apple and Google didn't need exclusive hardware for it

I think it was an outburst of optimism? Or just finding new ways of selling to people

i think its FOMO

No one wants to lose market share because they are too late.

Everyone keeps telling about android and Iphone and how blackberry, nokia and windows lost

Voice assistants were really talked up for a while back then. Lots of future of computing claims, the big consulting companies were saying it’d be integral to business (the only way cubicle farms could get worse?), etc.

For Amazon, that’s something of a threat: Apple, Google, and Microsoft would be well positioned as OS vendors and there is certainly room to imagine that putting Amazon at a disadvantage similar to what non-default search engines faced (remember A9?), not to mention the value of making it as easy as possible to order things.

What I don’t understand is what all of these people were working on. The hardware isn’t bad but it’s not that kind of investment, and the software doesn’t seem to have improved in years (which seems to be true for the entire field - tons of people using them, mostly for the simple stuff Steve Jobs did in the first Siri demo a decade ago). I’m assuming there were projects which haven’t seen the light of day and quite likely never will, and that there’s probably a great inside story about strategic failures locked up behind NDAs.

Notice a difference between the first Echo from 2015 and today’s?

No?

Well they still had us rewrite the entire OS, not once, but twice, and maintain all three OS’s across 10+ different devices…

They did the same damn thing with the API to cloud servers. Every couple years, new deprecation and new way to do things. Promo driven development out the wazoo while ironically quoting “customer obsession”

Zero ability to influence direction as IC. All top down initiatives from directors planned a year in advance.

Interesting. Sounds like Google syndrome. Why do you need to rewrite the OS 3 times for a speaker with mic?

First OS was scrappy to MVP hardware and prove the market. But instead of iterating, they decided to fork Android so they could scale up a workforce who can only understand Java, not C++, to deliver features. Then they realized Android is too bulky for cheap chips, so back to custom C++, that just like the first is a trash fire for rando devs to deliver features fast under a whip.

Thanks for filling in those details - that makes sense for having so many people busy on things which aren’t customer-visible.

> Notice a difference between the first Echo from 2015 and today’s?

The later versions seem to be marginally less capable of understanding speech and far more annoying.

"Alexa, play romantic music!" "Here's 100 top romantic death metal hits! By the way, did you know I can produce fart noises? Just ask, Alexa, blah blah blah"

"Customer obsession" is a joke.

Any voice assistant should be a feature and not an independent product. I've never used Alexa but I use Siri daily for exactly two things - turning on/off my lights when I get home/leave and checking the weather. If Siri at any points tried to sell or advertise me anything I would simply stop using it. The problem with voice assistants is that they are too dependent on other devices for people to pay more money for them. Additionally, if you own a smartphone you already own one therefore you don't need to spend money on it at all.

Looks like MS saw this coming a mile away when they pulled back on Cortana. What was MS seeing that Amazon didn't, or was it pure luck?

The difference is Cortana was an inferior product!

So it’s worse because they didn’t sink $10MMM into it.

I think that shows they knew when to cut their losses and not throw more money down the drain.

>"We want to make money when people use our devices, not when they buy our devices."

Lack of vehicle integration is the colossal failure nobody is talking about. Selling ads for brick and mortar locations you are predicted to pass (e.g. coupons for food and fuel) are completely untapped revenue streams. Imagine Alexa parsing your shopping list after analyzing your GPS route. It announces, before you leave, where to stop for the largest discounts and where to stop for the fastest pickups. Basically, they need to add "navigation" on top of "assistant". Alexa as the "driver's navigation assistant" would be an overnight success.

Besides my phone, my echo is the only device I use every day. It's so useful that I would pay a monthly fee to use it. I suspect the must have subscription is coming as a way to monetarize it. I hope it won't be too high.

If you don't mind me asking, what do you do with it?

We have an echo, and it's got three uses:

- Alexa, add X, Y, Z to the shopping list - Alexa, set a 10 minute timer called "pies" - Alexa, turn the Bedroom/Living Room lights On/Off

We had a google assistant, but they fudged the shopping list in the mobile app. It was an on-special impulse buy to switch to an Alexa to get that feature back, but I can't imagine paying a subscription just for what we use it for.

Here's what I can think of right now:

turn on my coffee machine; which I setup to brew the previous night

listen to the news

listen to podcasts

listen to audio books or it can read kindle books

listen to local radio stations

add items to shopping list

listen to music

Control my TV

Control the lights around the house

Make phone calls and use it as speaker phone

get notifications on amazon orders delivered

get information about local businesses

look up wikipedia info

get word definitions

tell me the weather

look up recipes

Get traffic info

Turn a fan on and off

It's primary function is to listening to stuff. I can use my phone for all of this but using the echo is much more convenient.

I don't use any of the echo skills simply because I haven't taken the time to figure out what's available. I suspect there's useful stuff there.

How is it a "colossal failure"? Amazon has tens of millions of these devices which are a 24/7 microphone, and with the wireless transceivers even more data-collection capabilities.

THe data is worth more than $10-billion alone.

I think the problem is that they have no consent to collect or use any of that data, and if they were caught collecting/using it, they’d end up in massive legal trouble.

Just because it’s technically possible doesn’t mean it’s actually possible.

When you say 'any of that data', you are incorrect. Simply to function (voice recognition) the echo microphone is constantly on, in the same way that any input device needs to be on to function.

You're not only incorrect, but your comment goes on to be incoherent, did you think before writing this comment?: 'Just because it’s technically possible doesn’t mean it’s actually possible.'

I have a Google Home, and the only things I ask it to do are to turn on my TV when I’ve lost the remote, play music on my HiFi via a Chromecast Audio (which they don’t sell anymore) and play different radio stations.

Maybe the silver lining here is that we can finally get a good "home assistant" that doesn't phone home. I think there's huge potential with these types of devices, but they're also really creepy; they "accidentally" send recordings back to the server surprisingly often.

Maybe someone could build an assistant with trust at its core, everything happening on the device, and a hacker enabling ethos. These devices are a far cry from the independence we get from the old-fashioned desktop computer.

If you have a HomePod mini and a recent iPhone, it can do all the voice processing locally.

I believe Google is also moving to that model or has, but it’s been a while since I transitioned my devices over.

5 years ago, Amazon saw the sales figures and thought their smart speaker business was the bee's knees. The Alexa org had thousands of open roles.

There was all kinds of talk about how smartspeakers would be the next smartphones. They went all in.

There may be an important lesson here, to understand the limits of your markets. They could have made a killing if they had taken the win and developed it slowly, rather than doubling down and letting their bet ride.

It is hard to sell ads in voice/interactive interfaces so the scale does not matter, product has to be fantastic. But weather/music/timer is not revolutionary.

It may have a shot with a more clever AI that can do cool stuff, that you would be ready to pay $5/mo for.

It's been a build in progress since 2007 at least, according to public web scraper logs.

15 years without a stable, usable product makes this unsurprising.

Reply on news.ycombinator.com