Service Provider buys voice recognition software and sets up complex maze of phone tree options to drive users away from the human support agents (even though the users can't solve their problem without human intervention - if you don't want to pay for enough support agents for your call volume, wouldn't it just be simpler to let me cancel my damn account online??).
Now user can deploy their own speech synthesis bot to wait on hold, with what is presumably a complex system of AI decisionmaking to be able to navigate the maze and find a human support agent to connect you with.
1. To make it difficult and frustrating to cancel by wasting the user's time.
2. To allow the company the opportunity to haggle the price of the subscription down, convince you not to cancel, or worse, upsell you on other services.
I can totally foresee some kind of Google Duplex detection in the future intended specifically to ensure that they're wasting the time of a human, not a pile of linear algebra in a Google datacenter. Or perhaps they play legal games with two-party consent states to try and find a way to sue Google into not offering the service.
It might not be as fast as one would like, but it usually does the trick.
I'm not sure what my threshold is for reporting a company to a bureaucratic institution in hopes for resolution, but I think it would have to be very high.
It's just unfortunate that mailing a physical letter is such a pain.
It could be worse: it was only 4 years ago I needed to cancel my HVAC installation with Sears Home Services and invoked my right to cancel the contract under WA law - and Sears required me to send an actual telegram (yes, the old-school, hand-delivered, all-caps short text messages) to their head-office address. I used this company: https://www.americantelegram.com/timedatestamp.php - to send it (and yes, it worked and I got out of the contract). But wow. I don't know how this can possibly be legal as it places an unnecessary undue burden on consumers unless there's some archaic legal carve-out for telegrams specifically?
In France you can send a registered letter from the post office's website for just a few euros. Upload your PDF, they print it, deliver it and email you a receipt when it's done. For a little more you can request a signature from the recipient. Since they keep a copy and record, it acts as a cheap form of notarization and is valid for many legal purposes.
So, if a company allows the client to create a contract by phone or Internet, it should also be easy to cancel it the same way.
no such thing exists; having to go to court at all is an expensive hassle
I’ve never heard of such a law before, but if one exists, it would be individual state law since corporations are incorporated at the state level. Do you have a link for this?
Correct. My understanding is that every state has an equivalent law. I can't give any citations really, though - my own state (WA) buries the responsibilities and requirements for registered business' registered or agent addresses in multiple separate sections in the state law website.
Nothing against php but I love it when technology from different centuries comes together.
As best as I can tell, Zipcar seems to be a shell of a company barely still functioning.
Yeah, Vodafone ... still sending invoices which deny I cancelled my contract half a year ago.
In Germany usually your contract says it will automatically extend for a year unless cancelled. This basically makes your contract open ended with cancellation option, which you need to exercise if you want to cancel. You can also cancel your contract in case the other party change anything related to it.
There's nothing wrong with that, you just need to read what you sign. And it's a common practice, not a surprise. If you want to be sure your contract will not extend, you can always hand over the cancellation request at the time of signing the contract.
I don't understand what distinction you're trying to draw there. It definitely feels as though the contract is automatically being renewed. Contrast with what I view as "a mere extension" - you contract for two years, and then after that two year period is over, it extends on a month-by-month basis and you can cancel at any time.
>There's nothing wrong with that, you just need to read what you sign.
This is a silly argument. The idea that large businesses and individual customers are equal has been disproved multiple times. It's like arguing that the American and European health care systems are equal. I can't modify the contract to have a fair and reasonable cancellation terms. But the business doesn't need to have unfair and unreasonable cancellation terms to provide a competitive product. It's just that they can say "These are the terms on which we will do business. Accept them, or accept the equivalent terms from another business."
In Germany, consumer rights basically don't exist and the playing field is heavily tilted towards big business.
I actually worked for a company that implemented an AI to try to recognized those email to automate the cancellation process (there was still a human at the end to validate the cancellation to avoid false positive).
Only after that, should a divorce occur.
The prenup is hardcoded.
That simple formulation is easier. “It must be as easy to cancel a service/subscription as to sign up”. Able to sign up online? Then it must be possible to cancel online
Gotta be careful with online service accounts though. Doing charge backs actually violates their terms of service and will probably get people's accounts banned/deleted. Not a good idea to do a charge back on something like a Steam account with over $2000 spent on it. The woes of DRM...
Their refund policy is generous. If you haven't played the game for less than 2 hours, you can get a refund, no questions asked. Normal customers generally don't need to apply chargebacks because of this policy.
Credit card thieves however, make purchases that are charged back eventually. Steam goes thermonuclear on such accounts, reducing the incentive for credit card fraud on their store.
Maybe in cases where the fraud is proven. After the consumer buys tons of stuff on an online store, the company shouldn't be able to just delete their accounts unilaterally just because of one chargeback. There's a huge power imbalance here and consumers need protection.
> Their refund policy is generous.
What they offer is the bare minimum really. And they had to be condemned by a court of law in order to implement that policy. They also suffered zero consequences for years of consumer exploitation with their sales plus lack of refunds strategy in order to trap impulsive buyers. It's not a coincidence that Steam sales became a lot less generous after the refunds policy was implemented.
For example, your Chase card pays Piracy.com. Some vendor makes an unauthorized charge on your Privacy card. Now you have to deal with Privacy's policy instead of your bank's policies to dispute
They can still send your account to collections or ding your credit record, but you can still refuse to pay; ultimately the only way they legally force you to pay is by going through the courts system but I expect that if you have documentation of your (unsuccessful) efforts to cancel the case should be in your favor (which is why this will never reach the courts in the first place).
Because legal consumer protections in the US are not very strong, and if you are wronged, you are instead supposed to go to arbitration, or file a lawsuit.
I shouldn't have to "cancel my gym membership". I just stop giving them money and my keycard stops letting me in. I shouldn't have to remain eternally vigilant with my credit card statements lest someone keeps pulling money from it, that simply shouldn't even be something that is possible.
So bizarre. This kind of one-way certainty is what I like about cash and thus Bitcoin. For example, it's kind of win-win that more shady websites (file lockers, mega.nz, porn, etc.) tend to use Bitcoin because it's also a weight off your shoulders.
I mean, I know it's a dreamland. It could come about if contracts between businesses and consumers are basically regulated into standardisation as they should be.
EDIT: looks like they settled for $1.25 million, a ridiculous salon on the wrist when their take home was many many times that
There may have been other suits... this one was limited to NY?
The only way to make sure someone answers when you call would be to have 10 times as many agents as you need. Instead, they try to find a balance and make sure the agents are more often then not on the phones (much higher efficiency), which ends up meaning long wait times on average.
High turnover (45%) in the call centers also makes the problem much worse, this means in practice that the centers are constantly understaffed
My company has maintained <60 second response times on our support 24/7 for a few years now, and although obviously it does require some level of additional staffing to cover peaks in demand, it's significantly less than you might imagine - I'd say maybe 1.5-1.7x.
We did optimise this by filling the lower demand periods with other tasks which can comfortably be completed more slowly - eg reviewing identity documents for KYC, etc.
I see it as being that the overall bandwidth you need being pretty close to the same, just choosing to have very low latency for specific tasks, and making sure constant context switching is manageable.
Edit: pulled some stats for another commenter - currently 51s first response, ~40s subsequent, 2.7 responses in total to resolution.
Of course a better fix would be a combination of the two. Have some agent read your original email sent through some secure internal site and build and classify a case in the system with email attached and then reply to let you know there is a secure reply with a chat link to the complete case detail with time frames reserved for high responsiveness.
But if I am having to stay on your website and use your chat widget I expect a faster reply!
I'm talking about how they bury the human agent help option, you often have to sit through and pay attention to minutes of different options to find the right path. E.g. "if you want to check your account balance, press 1. To pay a bill, press 2". Who are these absolute psychopaths that are calling the support line to check their bank balance or pay bills, instead of using the app/web site like a human being?
The only explanation for this is 1. There really are tons of people confusedly calling the help line for things they could trivially do online, and the automated options actually do handle this load or 2. The help options confuse people so much that they give up and just keep paying for the thing because it's too difficult to cancel, or to handle whatever their issue is.
I'm sure there's some of both, but I suspect there's a lot more of 2 going on and it has been increasing over time in recent years. More and more businesses have learned from the gym model that if you make interacting with their accounts difficult enough then people just won't be able to churn.
... a huge GCP bill for both sides of the call.
Here in NZ, I cannot imagine a company surviving if they made it so hard to contact them that systems like this were neccessary. I have waited a few minutes on occassion, but it's far more common to get a "enter your number and we'll call right you back" in that case.
Something is broken with US competition law if this is a commercially viable system.
You either go online, or set it up from the voice bot, and when it's your turn the provider just calls you back. We're at the point where phone calls are so cheap that I doubt it costs them more than the salary of the representative.
So what Google does is pretty much using AI to work around the fact that service providers don't provide this simple service.
Oh wow, I had not even taken that into consideration.
Checking prices on the first API-service that I know of, with standard prices (no need of any volume).
€ 0.064/min = € 3.84/hour
Salary cost would probably be about 10x that, and with volume I expect the calling cost to drastically go down.
Not without waiting for 40 minutes and using up your minutes and possibly paying for the air time!
To fix the problem, all the idiots have to do is turn off the annoying music so you can wear your headset until your call is served. That hold music really is the crux of the problem. If you could just hear a tiny bit of noise indicating that the line is up and a really brief "please continue to hold" message every five minutes, it wouldn't be so bad.
It really should be illegal especially if you already have an online account you log into not to be able to cancel. Idk if I was just stupid but I struggled to cancel my Scribd account so I cancelled it from the PayPal side cause I was so over the awful UX and it was probably me trying to find the payment stuff on the Android app which wasnt there cause I didnt subscribe through Android.
Companies do the most awful things. I want to make the world a better place by making it easier to cancel a simple subscription from any and all institutions.... Should be the pitch for a glorious startup.
But at the same time, I can't help but wonder if representatives will actually stay on the line to wait for you.
If you're accidentally on mute or they don't hear you, sometimes they'll hang up very quickly, within just a few seconds. (Not always though.)
I hope that because it'll play a message for them instructing them to wait for you, they'll wait... but I also assume they'll set an internal policy on how long they're allowed to. Will it be 30 seconds? 15 seconds? 60 seconds?
> Once a representative is identified, Google Assistant will notify you that someone’s ready to talk and ask the representative to hold for a moment while you return to the call.
Wouldn't be out of character a bit.
That's what I was thinking; here all these services let you wait for 20-40minutes and then they hang up if you don't IMMEDIATELY answer when they say; 'hello sir???????'.
I guesss we'll see if that message works or not... I guess it'll take a lot of tweaking per country and even regions of the country. Where I live they typically don't wait even if you don't speak the accent; they act like they cannot hear you and hang up.
9 years ago we announced an app that did this right here on Hacker News. https://news.ycombinator.com/item?id=2285594
How we did it: We hired the AT&T lady to make a recording that said, "Press 1 for your next caller". We played that on a loop for operators. When we detected a DTMF "1" we'd call your cellphone and connect the calls. We had most of the big companies in the app, but failed in the end because we couldn't find a business model. My first VC-funded startup. :)
Site is still up! (But not longer mine) http:/www.fastcustomer.com/
What I'm up to now: http://www.happymonday.com/ (helping people find work based on mutual values, priorities and functional work culture fit.)
I believe Path Talk did something similar too.
I'm curious about how you raised so much money?
We were solving a very real and emotional problem, received a lot of press coverage. TIME magazine named us one of their "Top 10 of everything" in 2011, for example. http://content.time.com/time/specials/packages/article/0,288... (look at that old iOS interface :)
Each time we were covered, we got more downloads, and were able to raise some more money. I don't think the money was misplaced. It was a good bet, sometimes they just don't work out.
Focusing on the Spanish market at the moment. Any chance we could have a chat on Twitter or somewhere else?
Here's the link btw: https://www.deicol.com/en.html
For instance, here are some features that I wish existed in the market with current automation phone techologies + respective APIs.
- Default voicemail, but having the flexibility to store, edit, record multiple voicemails that can be triggered based on context. If the phone number is not in my contacts / not recognized / no previous-history/interaction, play this voicemail. If my mom calls on the weekend, play her this one letter her know I'm out of the office. Spam number calls, automatically have automated voice say "I know you are a spam bot, reporting now.... please hang up as you are being recorded"
- grant auth to another individual to receive sms forwarding for a particular number. I.e. customer support text messages forwarded to my phone number, I would like to grant individual X to receive them, and when they reply back, it passes through my phone and to the end customer, who thinks they are interacting with me.
^^ just a few but I have a list somewhere. If this is interesting, I'd love to discuss/collaborate or minimally get your advice + thoughts.
Why did you need VC for this?
So I'm not sure how well this would work in practice. At minimum, I would be anxious the whole time I was waiting, ready to hit the return to call button at a moment's notice...
I think thats part of the playbook for most support call centers.
If I were writing an utopia, class action lawsuits would rain like fire and brimstone.
In 2015, the hotline got the busy signal for 29 million callers (Australia has 25 million inhabitants). 7 million were hung up during the wait time. I myself have been in their queue for hours at a time.
There are zero legal ramifications for this behaviour.
You've clearly never had to deal with a company in a monopolized industry like telecommunications. The call center needs to be there on paper, otherwise it will give people the option to cancel their contracts or even initiate lawsuits because something doesn't work and the company is unreachable.
The call center doesn't need to provide quality support however; as long as it is there on paper it is all that matters. In reality the call center is costing the company money and they'd rather not have you on the phone as long as you keep paying your bill (and most people will keep paying even if they receive subpar or lack of service because of the lack of consumer protection laws in the US, forced arbitration, the fear of a hit on their credit report or the lack of competition).
The people in this thread are acting like call centers are all bad when the real problem is the monopolized industry and not the customer service agent.
Also in more benevolent companies, providing support over the phone is the most expensive way to provide support. So as much of that that can be shifted to automated systems, better products, etc the better because it avoids the high cost of a call center.
While a call center employee may be paid by minutes on the line (though I think that is rare), they would be penalized for taking too long on a single call or helping too few customers. There's little incentive to spend time on longer more difficult calls when you could just focus your effort on solving many of the simple calls instead.
It's not just in monopolised industries, but the entire process has been labelled "low skill work" and left to fester.
Even in B2B, most commercial transactions are with an oligopoly, or monopsony, as one party to the transaction. Competition-driven customer service doesn't exist here. For instance, not even enterprise-level customers can get meaningful customer service out of the likes of Google.
In a healthy market this behavior would cause the company to lose marketshare to a competitor offering better support, which is why the only companies still using call centers as their primary (and often only contact) are those in monopolized industries like telecommunications. Every other industry has moved on to better solutions (async email, chat, etc) because consumers demanded them.
The real problem here is monopolies and misplaced regulation (harmful regulation that prevents competition combined with the lack of regulation that would force the incumbent to actually provide good service).
I'm not joking, this is an actual feature. I used to build callcenters and it surprised me when I saw it in the documentation. None of our customers ever implemented this but I can imagine some would otherwise it wouldn't have been in there.
Depends who you consider to be "the customer".
I suspect for a huge number of call centres, "the customer" is the bank/telco/insurance company that's outsourced to them, and the key metrics and incentives of their relationship with that customer _totally_ make "expectation management" make sense. The call centre company has no incentive to make life better for account holder or phone users or people who's house just burnt down - beyond extrememly narrow interpretations of the responsibilities in the contract they have with their customer - which is presumably as poorly written and as easy to game as any software project requirements doc you get from those kinds of clients...
In the company I work for now, they tried outsourcing but moved all their support back in house because they are a market leader so keeping customers is more important to them than gaining new ones. Hence more focus on good support than agressive sales. It makes a world of difference.
Roughly speaking in terms of (inbound) callcenters there's 2 kinds of attitudes.
- There's the callcenter type that actually cares about their customers and strives for the best metrics possible in terms of satisfaction (usually measured by after-call survey) and the shortest waiting time. This type of callcenter is often overstaffed to make sure they can handle peak times without pressure on the staff (which will reduce customer satisfaction because the staff will be stressed and the customers wait longer). In these callcenters it is common to see staff sitting 'idle' during which time they are supposed to do some elearning or support emails, or even kick back and relax. The metrics they care about will be customer sat ratings and waiting time (as in lower is better)
- There's also the kind of callcenter that is all about productivity. These are typically outsourced and get paid per call. They tend to have the minimum staffing so they reduce staff idle time: Hours not on call are 'lost' hours to them. Staff will be managed more on calls per day/hour and handling time per call (lower is better). Less time will be available for training (if any). These are the kinds of callcenters that want features like I mentioned. For one because it dissuades customers to call for minor issues and use email support instead, which is a lot cheaper to provide per customer interaction. And also to keep distance between the 'regular' and 'premium' support tiers, to make the premium waiting times shorter. This kind of callcenter will usually want to tweak the system to generate the best statistics for the customer that outsourced to them.
Obviously the first type of callcenter is much much better. Both for the customer and the employees. But the second type is very common too, especially for things that are commodity (as in easy to provide, not much knowledge needed), like number information services.
The only thing worse than the second type is the outbound sales-driven callcenter which are real sweat shops. Luckily the company I worked for didn't really deal with outbound.
Didn't look into it so no idea how it behaves, but sounds like a realistic solution to this problem and would have very interesting consequences!
The purpose of a hold queue is to reduce the number of customer service calls a company has to field. They will not take lightly to any service which defeats this.
Alternatively, they'll hire staff to answer the phone, say a few words to break out of Google's hold music detection algorithm, then put you back on hold, this time with ads instead of music. Repeat every few minutes to make sure you're still stuck on the line.
There's no limits to what's possible when you realize the purpose of anti-customer service is to harm customers rather than help them.
I’ve had Brands interactions on Twitter where they ask me to go into private, then ask for my phone number, call me and send me directly to a holding queue, i guess it would fair to respond in kind.
Exactly the sort of behaviour I'd expect from a company that considers "brand interactions" to be a thing.
It's probably a wash: their outgoing call to you costs as much or less per minute than their toll-free incoming lines. At least in North America where Calling-Party-Pays isn't really a thing.
In cellular plans that don't include long distance calling, it could cost you more to receive a call when you're out of your home area, while the same toll-free call is (usually) never long distance.
> somewhat common feature of customer service phone systems to have a way to do callbacks
Maybe on the sales side, but on the service side, many vendors (particularly b2c) prefer that you give up entirely and never call them ever again.
So double to triple the cost from some providers for just this feature sounds very expensive. But perhaps with large commercial pricing this difference shrinks.
That being said, are we not talking about AT-and-freaking-T? :) Surely they'll've been able to work things out. I think, anyway.
(Also, isn't there some weird thing in the US where cell phone users get charged for _inbound_ calls? They might even make money on these...)
I often wait 2-3 hours for a 5-10 minutes call.
Besides the obvious business incentives (eg Comcast wants to make their phone experience as bad as possible), I'd guess the main obstacle holding this feature back is the specific PBX system a business is running on.
It helps that they can link your caller ID conclusively to your account since it’s their own systems.
(Disclaimer: I work for Google, although on a totally unrelated team.)
Maybe the caller was a Norwegian death metal fan? ;-)
"What's that awesome track you were just playing? I tried Shazaming it, but it just came back with Beyonce. I can't find a death metal band covering any of her songs?"
The obnoxious part is where companies like Apple who actually choose and license nice music play it in such a way it sounds absolutely horrible on the phone. What a waste!
Modern digital is worse because it combines the frequency passband of analogue with combinations of codecs and bit rates that can't support the audio complexity of music.
In the age of Compact Disc, people are spoiled for audio quality ;)
HIPAA fines are fearsome. Medical IT is heavily weighted towards creating parallel systems that don't touch the rest of the world.
When I order food over the phone, the only people who need to take a cut of the transaction are the restaurant and maybe the credit card company if I don't pay cash. I can also be confident that my tips are going to the person whose hand I place them in.
When I order online, there are umpteen different companies taking my data and the restaurant's money, and they often obscure where fees and tips go. To me, it seems absurd to involve so many parties in such a simple transaction.
The experience is also usually fungible. Talking to someone doesn't take longer or introduce more error in my experience, so I usually prefer it because there are less externalities and fewer complexities.
Different strokes for different folks, I suppose.
Using computer agents avoids the middleman.
If the restaurant runs a website, there's no "umpteen different companies taking my data".
Also with a lot of these people, they can't really schedule, stuff is just a queue. This dude is definitely a queue.
Doing FIFO may be “fair”, but resulted in more time spent travelling rather than responding to calls.
And then you have restaurants own websites which if they exist, are super clunky and require typing payment details rather than paying on pickup.
In my estimation, the state of algorithmic music doesn't currently allow for the compelling listening that we're used to with most "human" music, but I can see it being good enough for situations like call-holding or similarly ambient situations like restrooms or lobbies. It could be somewhat bland, but it's content shifts subtly enough to avoid irritation over the course of ten minutes.
In my head, I hear something like the kind of minimal techno that Ricardo Villalobos is known for. Long, evolving, beatscapes that slowly transform over the course of an hour: https://www.youtube.com/watch?v=OZWdWzMdndc
I was once put on hold at a music equipment rental shop and heard a decidedly "non-muzak" drum and synthesizer loop. When the rep took me off hold they explained that I was hearing a live feed of some electronic instruments that they had running in the back room. He then put me back on hold in order to demonstrate him playing around with the synthesizers that were on loop. That really made my day.
Of course the real solution is to make phone support paid (with a refund if the problem ends up being the company's fault or something that can't be done self-service) as to discourage this behavior.
I reply "Thank you" when it says this. I mean, it's permission right? Saves me remembering to ask the person who picks up if it's OK for me to record our conversation.
Generally most companies I know have an opt-out option in the IVR now (IVR = voice menu). Or opt-in in the countries where that is required.
> * Every business’s hold loop is different and simple algorithms can't accurately detect when a customer support representative comes onto the call. Hold for Me is powered by Google’s Duplex technology, which not only recognizes hold music but also understands the difference between a recorded message (like “Hello, thank you for waiting”) and a representative on the line.
And yet they also want you to believe "Your call is very important to us!"