Emotionless chatbots are taking over customer service
theconversation.com
theconversation.com
IMO, most chat bots out there are just digital reincarnations of phone VRU systems. I don't see them taking over customer support anytime soon. Bots should not be the front and center of customer support, but should be assistants on the sides of both customer and the support agent to provide sideline help while a conversation is ongoing, instead of the bot taking over the role of the support agent.
For example, when I start a conversation with a support agent, I could be prompted to enter my email address if I havent, or select a category (shipping, refunds, etc), all while waiting for the agent to get back to me, or as a starter to the conversation. On the agent side, a bot could be helping me by suggesting responses to the customer's question and letting the agent pick a response, instead of the bot answering the customer directly.
The human experience is crucial when dealing with a business. We're not there yet with AI to have bots take over the entire conversational experience between a customer and a business, so if you're building a chat bot, please don't bring the horrible phone VRU experience to the internet.
Don't get me wrong on bots though, they're awesome when it comes to certain situations and tasks that don't require human to human interaction and are usually structured, like shopping. Take Shopify's Facebook Messenger bot where a person could shop on a business's Facebook Page via their Messenger bot, with the bot taking on the role of navigating the catalog for the user.
What AI's are there that attempt information gathering and understanding? All I can think of is IBM Watson on Jeopardy.
Of course, if the rules say "if the customer asks for a refund, stall for 5 minutes and then transfer to a different department, unless they threaten with legal action", you can't really do that. So I'd guess that most companies won't want to expose themselves that way.
So far, in my experience... they don't work all that well. Although to be fair, it might be entirely psychological. When you're speaking with a human, you feel as if there is a possibility for improved understanding. When you're talking to a robot which doesn't understand you, that's a pretty hard barrier.
That requires actual intelligence to be able to interpret what it browsers -- which we are nowhere near close.
[0] https://aws.amazon.com/contact-us/
[1] https://support.stripe.com/questions/do-you-have-a-phone-num...
...of course, that does nothing for getting to an intelligible human, with the heavy outsourcing that's happening today. In that case, maybe emailing might be the easiest way to communicate.
I am interested to knows if there are non-profane keywords that are programmed into text-based systems. I would assume that using profanity trips a flag that indicates that the user is hostile. This might cause a human rep to enter the chat with their guard up, less inclined to want to help the user than if the request was delivered in a neutral tone.
I thought the system may have simply detected the volume of my voice, but when I heard about profanity detection it made a lot of sense.
CALL TERMINATED
Better than the drug store I called and had to sit for 3 minutes while their phone bot read me the entire FAQ before I could even "press 1 for...".
Store hours, pharmacy hours, "We now offer" specials, "Did you know?" ads... jeez.
I unsubscribed to an email newsletter recently and was shocked it only took one click. I didn't have to check boxes or "update my profile".
In a time of fake on-line Houston Relief charity sites and endless massive account hacks it made me suspicious it was so easy.
Exception is if they have me by the balls i.e. a service I need to get some emails from. Then I'll log in and figure out how to "update my email preferences". Which are usually not my preferences at all but some silly defaults.
The online fax company that sends me fake random floating point numbers that represent how many dollars I owe them. They get a special rule!
At the time, they were making fun of automated phone trees and the joke was that they would get so good people would prefer them in the future.
I think many people already think that day is here. How many folks prefer to use an online form or chat instead of a phone call? Most of those systems are just automation with a human checking their work.
Soon the human will be out of the loop.
a recent experience with a plumbing equipment company let me upload photos of my issue. Something I can't do on the phone.
I also don't have to sit there and wait on hold.
Online form/self-service: a few clicks. Done.
Calling my carrier: explaining what I want, turning down the offer to upgrade to a more expensive plan, hearing about a promotion I'm not interested in, etc...
Somehow, though, they managed to stop my service when I cancelled my card.
I think Demolition Man was fortelling on several jokes!
> Company bots refuse legally required request.
> Open phone app, click a few buttons.
> Law firm drafts letter of complaint, lawyer skims it for 30 seconds.
> Law firm threatens to sue 100,000+ times across 50 jurisdictions unless company settles all pending cases at 10% premium of expected outcome.
> You receive payment minus $2 for legal expenses.
I actually suspect that automating both sides of that exchange will shift the balance of power in favor of consumers, since companies already have a massive amount of liability they're ignoring because it's difficult to extract payment over. If you could automate that process, especially by filing a distinct case for every contract violation or legal issue they refused to honor, I suspect many companies would collapse.
If lawyer costs fall to the point where you can automate the suits over such business contract violations, I expect you'd have the ability to legal DDoS a bunch of abusive companies into submission.
I think a lot of modern AI misses this where they just make a better version of the automated phone redirecting thing (Dont remember the name for this). They should include more ways on actually designing the whole experience and make the users actually enjoy using the AI.
All in all this makes me think of Manna, as customer support was pretty much already about humans following scripts with no allowance for deviation...
Moreover, customer support will get more automated in the future and the differentiating factor will be how human or nice it feels to use which will in turn make the company stand out.
I'm a good example. I hate it when I phone banks and they give me a lot of automated phone waiting and redirecting. It makes me not want to keep phoning in the future and also dislike the bank because the experience sucked.
The caveat is of course now you have to put more work into it to create that experience but I believe its worth it. I wouldnt train customer support stuff before automation to be emotionless so why should they be emotionless after automation.
I would claim that game AI almost seems natural to some people even though some are basic FSMs but thats because they study the user and the experience first and add game design before creating the program. A lot of what people do now is create the AI program, feed data, train and then think about the user or not even think about the user and instead talk about recall or metrics etc.
Human speech and language itself is significantly more nuanced and also evolving over time.
I'm not suggesting fun general NLP-based AI but to make those constrained to a certain problem space better.
Edit: I noticed I didnt clarify this before but I meant AI as in chatbots that were mainly for customer support or inquiries etc.
It feels a bit like moving the car aircon controls to a digital touchscreen display, sure it sounded like progress, but it really just made things more ambiguous and harder to use.
Having a tone dial menu works as long as there are only a few places to route you, and as long as all situations only have a handful of answers.
Furthermore, you can have tricks like looking for keywords in your human speech to make sure the guy you get patched to has the expertise to help you. If you say "my router won't turn on" you won't be sent to a password reset monkey.
At least, that's the theory.
Of course, you do not always get that from humans either, especially from companies that treat customer service as a cost to be minimized by hiring people to do nothing more than emulate a chatbot by following scripts: "Through some serendipity, I was connected to a human representative, but before I had time to utter my predicament, he told me “I shall put you through to the booking system” – and the infernal loop resumed."
I'm looking for something that will solve my problem, not something that will pat me in the back and ask if I want to talk about it.
On top of that, a lot of customer service calls boil down to wasteage. People call because they want something for free or a reduced price. With a commodity, there's no need to build loyalty for products that are primarily sold based on initial cost.
The quintessential commodified consumer product is Google search. Chatbots would be a step up. But that is not the direction customer service is heading...working in a bookstore now requires following a script for pitching savings cards and magazine subscriptions, not selling books.
I think they're just abused by angry customers so much it's easier to just appease emotionally than solve unreasonable requests
Reading your story it actually seems like it'd be hard not to feel patronized. Which I hadn't considered before. Usually it slips by me, masked behind my basic assumption of good intentions, and my tendency to just ignore anything scripted. Like at the onset of "Wow yikes ouch too bad I'm sorry that happened for you" my brain says
--- 2017-09-03-122634 -- irrelevant data stream -- waiting ---
If I'm going to the trouble to call or email support, I have a question that I couldn't answer with a simple search or with a help article. Wasting my time with frustrating chatbot doesn't get me any closer to a solution.
The vast majority of customers ask basic onboarding questions covered by documentation, questions entirely unrelated to the product because they don't understand anything about it ("hello Ford dealership, I'd like to report a pothole near my home. You deal with things related to cars, right?), or are things that could be avoided by making the product more user-friendly if engineering weren't focused on new feature development instead of addressing technical debt.
Chatbots (or the rough equivalent) are a necessary evil to deal with the bulk of customers whose first inclination to solve any problem is to call support.
In assistance they are going to replace worthless call center people that know nothing of the matter at hand and/or have no power/autonomy to decide with these chatbots.
It won't be that much difficult to make something better than the awful experience of talking with a clueless human, but from this to have something actually useful for the customer there is a loooong way.
The good news are that the underpaid worker at the call center (while now starving because of no income) will not be upset anymore by the angry reactions of the calling customers to their nonsense, and customers will feel much less that sense of impotence they experienced most of the time when they attempted to report an issue, as hitting a mechanical/automated hard rubber wall is more acceptable than having to fight for hours with humans in (often vain) attempts to get a proper answer or to talk with someone understanding/in charge.
I'm not surprised you have such a sour taste in your mouth when your belief is that customer service people are worthless.
* who cannot understand any 10 or even 5 lines of text properly;
* who do not answer the single question you asked and finish their answer with "I hope I answered all your concerns";
* who asks you for information that you already provided (in the very mail to which they are replying);
* who provides you with information that is unrelated to the case;
* who gives instructions which are inappropriate for the case;
* who gave you already 3 different and conflicting versions, assuming you are stupid enough not to notice it;
* who has demonstrated several times her lack of basic knowledge of the products;
* who blatantly lies;
* who repeatedly refuse to transfer the case to another of their workmates, despite your several requests;
I am entering my 6th week of struggle with such an individual (and we didn't get to half of the process), for a problem that could have been sorted in a week. It has taken a lot of my time already, and I do not get paid for that time, unlike her. It's also been damaging for my nervous health.
I formed a (kind of) theory (because that's far from the first time this happens, even if this one is the worst of them):
We used to have plenty of jobs for people with low intellectual capabilities. As workers, skilled workers, sometimes craftsmen. As long as the work was framed, directed, with many repetition, they could become efficient and even develop fine skills in their trade.
But those jobs became scarce. Unemployment rose a bit, but not so much in fact. The same kind of people are now kept in the school system, to keep them busy and not send them increase the unemployment figures. Many fancy educational streams have been created, they are given degrees, even small higher education degrees.
And they end up in the millions of tertiary jobs, they are service employees. But they are still the same people, and despite having been given degrees for the time they spent at school, those degrees have almost no value, for those people lack intellectual capabilities, abstraction, adaptable intelligence, you name it. They cannot read and comprehend basic very short texts in their mother language.
(A few years ago, I was back into studies for a while, and I have seen people who were given some kind of Bachelor degree, who were like that. The degree was not much more than an acknowledgement of the fact that they were present at school. But, for example, after 10 years of English lessons, they could not spell A-B-C-D-E...)
The class of employees used to be higher than the one of the workers, employees were better paid and better considered. Now it is the opposite. They are not more brilliant than workers, and the problem is that their job would require some intelligence and knowledge, but they don't have it, thus they suck at their job. It wouldn't have been a big problem if they had become workers, they would have been suitable after a while, but they chose, or they were made to believe so, to pursue 'studies' and become employees. Now they are less paid than workers, they cannot do their job properly, and customers have to undergo them.
In short: intellectual abilities still present the same repartition amongst the population as 40 years ago, except that there are much less low-skilled jobs available for the low-skilled people, so those people were artificially moved to higher-skills jobs, and it doesn't work because they don't have sufficient capacities for those jobs.
I can assure you that I am among the most polite and patient people you can find.
And I don't - in the least - put the blame on the (usually very nice and polite) guys/gals that answer the phone, they are (normally) very good willing and courteous, and I always treat them with the utmost respect.
It is the management that often (not always) puts in the front line clueless people or people constrained to follow a given (often wrong) procedure that rarely brings satisfying results.
The Authors of the (flawed) procedures AND the call center responsibles that blindly and senselessly implement them in extremely rigid ways (and hires, for the sake of the low wages, largely inadequate people) are the ones to blame, not the poor workers (the ones that will lose their already scarce income thaks to these chatbots).
There is no reason to believe that because of the chatbots the procedures will be bettered, nor that they will introduce more flexibility.
Reading your posts (this one included), it seems to me that you are not even among the most polite and patient people of this thread (neither I).
I'll happily pay extra to be able to call and speak to a normal person who's fluent in my language. In fact I go out of my to avoid companies that don't meet that criteria when it comes to important stuff like power, phones etc. Bonus points if they have a physical storefront of some sort that I can actually walk into.
Example of the weirdness: https://imgur.com/RPyhtkH
Its certainly adding to the experience one usually has with customer service- like drunken personal, constant advances to sell stuff to you- or to put you in the problem accelerator, by handing you around.
If the customer is king, and this is what is flung at the king, one may safely assume, that the rattling cart below is going to the Place de l'hôtel de ville.
The other role of the chatbots might be discouraging people from doing actions that the company does not like - like taking refunds etc. I guess there will never be a solution in this area that would be welcomed by both customers and companies.
By the way I have heard that in debt collection people prefer to talk to a chatbot than to a human, because it is more impersonal.
To help another person you have to hire another person.
A dedicated server chat-bot can be helping 500. Most companies don't want to be customer service companies.
I remember 20, 30 years ago, when there was not infrequently a well thought out and well-written manual, that actually taught you and answered many of your questions.
When "help" was real help, that actually knew something and presented it in a useful and authoritative fashion.
I just spent 33 days with ATT working to change an aspect of a service I purchase. Each time I called in, I had to re-explain the whole situation to whoever I got. They had different approaches to resolving it. There was no formal nor permitted mechanism for them to check back that the problem had actually been resolved. A couple took my email address and told me that their supervisor or manager would check back with me. That never happened.
33 days later, someone in support transferred me to someone who -- I learned near the end of the call -- was actually in sales but knew their way around the systems and was helping out with the Hurricane Harvey related call-volume spike.
I finally happened to get the right person. We went through the "official" process one last time, to document it, while simultaneously escalating to the appropriate team -- which team they happened to know of.
The next day, my service issue was resolved.
WHY wasn't that process documented internally, so that the first person I contacted, approximately 30 days earlier (after the initial request had obviously failed) knew what to do? Why couldn't the process be documented externally, to me, so that I'd know where I was at and what I needed to request?
I don't need a fucking chatbot. Just give me a page that clearly describes what's going on and what I can do about it.
One good technical writer who researches it, writes it, clears it with management (who may also want it cleared with legal), and publishes it.
Don't game me. Answer my fucking question. Solve my fucking problem. Leverage the abilities of a smart, capable person by making their research results clearly and concisely available.
If your process changes frequently, do you think it's going to be any easier to maintain the "spaghetti" help of disjointed web pages and now chatbot programming, that we've now come to?
No... it's that person in sales -- or wherever -- who is interested and self-motivated enough to keep up with things, who actually solves my problem
Give them a bonus.
P.S. Another person mentions the "swearing" trigger. I've actually started using that to get past the interminable "tree" of automated help, when/once I've learned that it has no feature/leaf that addresses my problem.
I had to use that with ATT, to get past some nigh-permanent automated system loops between tree leaves, to actually talk to a person. Plus, to "fucking" talk to them without wasting minutes bouncing around the tree, with the risk of ending up in one of the many nodes that simply hang up on you or put you in a hold "limbo" that never gets through nor ends.
I think chatbots can be great but their design and implementation are often terrible.
I'm not sure if the issue was actually emotionless chatbots or if the issue was outsourced employees who were just phoning it in by copying from a likely question in the FAQ. I highly suspect the issue was chatbots because of their inability to use context.
-----------------
ORIGINAL MESSAGE:
-----------------
I will be closing my checking account before the end of the month. Recently I received $50 back into my account because a bill pay check was never cashed. My question is this: If I close my checking account and then cash is supposed to be credited back to me, would I receive that cash by a check in the mail?
Thank you,
Novia
---------------------------------------------------------------
From : Customer Service
Sent : Received Date: 08/30/2017
Subject : Re: Account questions or requests
Dear Novia:
Thank you for contacting [Company Name]. My name is [Name #1], and it is my pleasure to assist you today.
When a merchant issues a credit to an account that is no longer open, the credit will be applied to an open checking, savings, or prepaid account.
If you do not have any open checking, savings, or prepaid accounts with [Company Name], the credit will be returned to the merchant. Please contact the merchant for more information.
Credit for returned card purchases usually post within 5 business days but may take up to 15 business days to post to your account, depending on when the credit is processed by the merchant. Returns will display on your statement as "Purchase Return."
If we can be of further assistance, please respond to this email by clicking on the "Reply" button or call us anytime at 1-800-###-####.
On behalf of [Company Name], thank you for your business. We are happy to have you as our customer and appreciate the opportunity to assist you today.
Sincerely,
[Name #1]
[Company Name]
-----------------
NEXT MESSAGE:
-----------------
[Name #1]....
Your response did not address my concerns.
I am asking about a "stale dated check" that would be returned to my account because it was never cashed. Please look at my account history and see the $50 credit that was applied recently because of exactly this situation. If my account had been closed, what would have happened to that $50?
The merchant is not the one initiating the return. It is happening automatically on [Company Name]'s end because a certain amount of time has elapsed since the check was issued.
Thanks,
Novia
---------------------------------------------------------------
From : Customer Service
Sent : Received Date: 08/31/2017
Subject : RE: Re: Account questions or requests
Dear Novia:
Thank you for replying to our recent email. My name is [Name #2], and it is my pleasure to assist you today.
I understand your concern regarding a Bill Pay check that was not cashed.
I realize the importance of this issue.
I would like to inform you that whenever a Bill Pay item is remitted by physical check and that check has not cleared after 90 days, the check is stopped and the funds are automatically credited back to your account.
As the Bill Pay check has not been cashed, we can place a stop payment on the check. There is no charge for this stop payment. If you would like to request a stop payment on this Bill Pay payment, please click the "Reply" button to submit your request. Please be sure to provide us with the payee name, the amount of the payment, and the date the payment was debited from your account. If we are able to stop the payment, your account will be credited within five business days.
If you have any further questions or concerns, please respond to this email by clicking on the "Reply" button or call us anytime at 1-800-###-####.
On behalf of [Company Name], thank you for your business. We are happy to have you as our customer and appreciate the opportunity to assist you today.
Sincerely,
[Name #2]
[Company Name]
-----------------
NEXT MESSAGE:
-----------------
Please forward my prior two emails to an actual human so they can respond to my inquiry. The past two responses have sounded like they were generated by robots.
Novia
---------------------------------------------------------------
Subject: RE: RE: Re: Account questions or requests
From: Customer Service
Received Date: 09/01/2017
Timestamp: 05:53
Dear Novia:
Thank you for replying to our recent email. My name is [Name #3], and it is my pleasure to assist you today.
I apologize that our previous emails did not correctly address your concern. I would like to take this opportunity to do so now.
Novia, I understand you are planning to close your account. Also, the Bill Pay check was never cashed and you would like to know if you close your account and the cash is suppose to be credited back to your account, if you would receive the cash by check in the mail.
Yes, when a Bill Pay payment is returned and the original funding account has been closed, a check will be sent by U.S. Mail to the most recent address on record. Please allow up to 10 business days for delivery.
If you have any additional questions or concerns, please call us anytime at 1-800-###-####.
Novia, ensuring that you have the best possible banking experience is of the utmost importance to me, as you truly are a valued customer.
On behalf of [Company Name], thank you for your business. We are happy to have you as our customer and appreciate the opportunity to assist you today.
Sincerely,
[Name #3]
[Company Name]
> If I close my checking account and then cash is
> supposed to be credited back to me, would I receive
> that cash by a check in the mail?
Yes.But then there's a typo in the reply ("is suppose"), so maybe not...
[Introduction with name]
[Problem Summary]
[Solution]
[Closing with phone number]
[Optional Extra Line]
[Signature]
If parts of that template are already filled in when handed to an actual human, it would make it easier for the company to get a chatbot to fill in the middle sections [Summary of Problem] and [Solution] by pulling from the FAQ.
If they are trying to cut costs by pulling stuff like this, is it any wonder that I would want to close my account?
It is not bad news. Automated registration of user complains should drive product development. Few people complained to a bot and problem should get to kanban board of developers.
If you need service, you got a bad product, good products need no customer support.
Bots are good scalable way for constant improvement based on feedback. It's a temporary crutch.
A common mistake that people make when trying to design something completely foolproof is to underestimate the ingenuity of complete fools.
Douglas Adams