DAMN Automated Customer Support

 

I'm using today's blog article to offload some thoughts. I was "quietly" offloading them to my closest friends because ... well, who really wants to hear me rant? But that was after one recent experience. Then a couple more came up which convinced me that I must write about it and naturally others will want to read it!

Beyond my ranting, there is something to be learned here, although one may argue that they (the culprits I'm about to rant about) should know this already. 

Who's / what's my beef with today? Automated Customer Support, henceforth to be known as ACS.

I'm sure there are a lot of great excuses reasons to use it, most involving cost and likely a badge to say they are using AI, which show you are at the forefront of new technology, or maybe a tax write-off. Apologies, I'm feeling spicy today. There are other reasons too: some involving time, some involving translation support and easier customer communications, some involving focus and retaining humans for what truly needs a human touch and voice, and some involving cost...did I say that one already?

Automated Customer Support Annoyance Number 1 (Rant 1)

My first vexation came from when my computer started to "death screen" me every 5 minutes. And the 5 minutes was sometimes 2 minutes, sometimes 15 minutes, but it was a LOT. So much so, it made it impossible for me to properly work, let alone troubleshoot. 

So off I go to research and resolve with another device (my phone.)  Below is the fun process I went through. I did cut some of it off because it was getting too lengthy, but hopefully this is enough to give you enough of a taste so you can be just as annoyed as me. And ISKTBN, but almost all ACS systems I've called rarely end in me feeling all warm and fuzzy, nor are they quick. My expectations should have been much lower to start... 

1. First check, is my computer still under warranty? YES! And what would you know there are steps to take to file a claim.
2. I ran whatever diagnostics I could and they all pointed to, "you gonna need a new computer." 
3. I filed a claim, YAY! But wait, no confirmation. Oh and the page now says the order is being created, but I need to come back later.
4. Later comes and goes and the order never gets created.
5. I take my device to a partner organization my device support pointed to as another avenue for help. To save some space here, it did not go well and would have cost me money only to end up having to send it back anyway.
6. But I couldn't send it back because I couldn't get the order generated.
7. I called the customer support number given to me by the partner (as well as what I eventually found on the company's site) and the ACS system guided me promptly back to the website where I would find "everything" I needed to do all the things. 
8. But the information on the website wasn't helping, surprise, surprise. I called back. the ACS system immediately told me to go to the website and hung up, knowing I called already with my phone number. 
9. I kept researching to figure out how to reach someone going through the FAQ prompts. Still nothing.
10. I call many more times. Same ending...go back to website, disconnect. 
11. I finally do something I only partially felt bad about (I might write about this later),  I went to the page where you buy the computer and then reached out to a "live agent." To be fair, I told them right away I wasn't looking to buy, but really just needed to talk to someone because their "ticketing / fix my computer" system wasn't working and I couldn't use the number nor could I find a way to contact them via the site.

That human was immensely kind and helpful. They got me to another person by telling me specific prompts to give and let me know if I didn't say just the right thing I would be routed to more automation or the website, like I was before. From there I got another real person. Also super kind. They too needed to escalate my case, but I did get emails and updates and I FINALLY got a resolution (they replaced my computer with a refurbished one - better than nothing!)

I can't tell you how much time I actually wasted on this. Maybe I shouldn't say wasted because I got to the resolution. However, a couple weeks of trying to reach out when it should have been done in, at most, an hour implies a lot of that time was wasted. And although I didn't count the hours, I do know it was WAY TOO MUCH TIME. 

Rant 1 Takeaway: If you have customer support and use automations to help manage the traffic, remind yourself over and over that ACS doesn't solve everything and we need more humans! If you have a traffic problem, there is likely something else broken along the way. Maybe documentation isn't quite right. Maybe there are issues with the product. Maybe there is an issue with your ticketing system.

One might argue AI, LLMs or what-have-you should solve for information gaps. I argue it wouldn't, based on my next example. Regardless, there has to be a better way to vet the incoming calls and searches. 

(Warning this next blob of advice is coming straight from a customer's perspective) There has to be more care to the customers. I will say after this, I am never going to buy another one of their products again. That is how frustrating it was and how helpless I felt throughout the process until I got a human connection. Granted this all has to be put into perspective because if this was over a pair of socks then I would say I was overly invested. However, this was tied to a financial cost that isn't slim and the device is used for my livelihood so the lack of resolution impacted how emotionally invested I got. 

Help should be as frictionless as possible and it should be scaled to the impact on the customer. This does not mean help should acquiesce to every customer demand, but if you offer non-human solutions to fix a problem, have a backup when they aren't working. And have a way to MONITOR those systems to make sure they are working.

🎉 Win for Human Customer Support

💀 Loss for Automated Customer Support 

Automated Customer Support Annoyance Number 2  (Rant 2)

For a new (and different) device I acquired, I realized it wasn't connected to the device support that went with it (because I learned my lesson that device support is very important.) I tried to initiate the connection via the email directions given, but that didn't work. So I dreaded trying to troubleshoot or call on this one. However, I did find a number, I readied myself, and I called. Spoiler, it went MUCH better this time.

So what's my beef?

The ACS voice that came on was clearly designed to sound like a human, even though it said it was automated. It had to have some AI element added to it based on how well it recognized my responses and then summarized them back to me, in addition to claiming it would keep a summary on hand should I need to get to a customer agent. I can see the benefit in the summary and tracking along the way especially if it follows you to another agent so that you don't have to start from scratch at each hop.

However, it wasn't the tracking or summary that got me. It was how the system "spoke" to me. The system responses added in what felt like natural pauses (some were so it could process what was being spoken), but it seemed like it even added in stutters and pauses of doubt. There was also a tone change, at one point really sympathetic and maybe even empathic, at one point stern and demanding, and at times a "I'm bored" feeling. I found it unnerving. I just wanted a traditional "I'm a machine" tone that provided factual steps. I was NOT expecting the voice to mimic a person.

I'm also not a fan of the automation identifying itself with personal pronouns. "I'm sorry. Let me help you with that." If my Googling served me, this is anthropomorphism. To me this, in addition to the tone, made the experience feel...like I was stepping into a contrived world where the surface is sparkling clean, but if a crack appears, there is something foul underneath.

Perhaps I just wasn't ready to deal with a machine with scripted emotions as if trying to deceive me into thinking it cared and lessen any negative emotions I might possibly emit. Maybe that company did testing and found that if tone was introduced their customer support team had fewer angry callers yelling not nice words at them. I dunno. For me it was just creepy. It would be nice if they could adjust and provide that level of scripted emotional support when needed. I mean isn't that what all this AI is for? In Annoyance Number 1, I clearly made it known to the ACS on the phone how frustrated I was when I yelled, "HUMAN! I WANT TO TALK TO A HUMAN!" My tone on this call was completely different and not yet fired up.

Back to my phone call with the fake emotions. I didn't get to the answer with this system either, BUT it did eventually route me to a person and on the SAME call. That person was 10,000% awesome. Granted we only spoke for like 3 minutes. I ended up finding out the answer as they joined the call because I was still playing around and trying to solve it on my own since the ACS did not. This person was so kind and positive and their tone matched what I would expect from a human when I would expect it. They gave me some background and reasons why my issue likely happened. It felt natural. I did throw more questions their way since I had them on the phone. They answered them clearly and when I hung up, I felt satisfied. 

Rant 2 Takeaway: If you are going to use ACS, know when to transfer them and don't take them through more than 5 - 10 minutes of guided troubleshooting before you get there. That is what happened here and it helped the experience immensely. 

Oh, and if the system says it is automated, it's ok for it to sound more robotic and monotone. Be clear and use kind words, but trying to mimic human discussion with a machine just feels unnerving and icky, IMO. Maybe that's because my expectation of automation has been set. Maybe others like it. I clearly do not. My worry is when this "make the machine human-sounding" may overstep the line and become confusing for others or mislead them and create more frustration. At least in Annoyance Number 1, I knew it was a machine that was making me mad. That machine was programmed a certain way. But I know my thoughts would be much different if I thought a human was treating me that way.

🎉 Win for Human Customer Support

💀 Loss for Automated Customer Support   

Potential Automated Customer Support Annoyance Number 3 (Rant 3)

In a related story about ACS systems, I consider the friendly "assistant" on some TVs and TV app services a form of ACS. And because this is my blog, I get to make up the rules; therefore, I can now connect the two and be 100% right. 

I would love to know who appreciates or finds creepy the "assistant" that mimics the volume of the human talking to it. A family member just told me the other day that the TV whispered back to them when they whispered. I don't know why, but that is incredibly creepy...but also incredibly funny. I did ask if they tried to talk to the "assistant" with a devil voice to see what would come out, but alas, we never found out. If you are a user of these "assistants," please test it out and let me know!  

💀 / 🎉 Loss and win(?) for Automated Customer Support

Rant 3 Takeaway: Ask yourself, "Is this really necessary?"

Wrapping it up 

Clearly, I'm have a personal fit with the Automated Customer Support systems as well as the tools that try to replace humans and / or be more lifelike. Maybe it's because most often my experience with automation of those sorts adds more time from my end and usually leaves a bad taste in my mouth. I've also been reading stories about unrelated, but still related, things where humans get emotionally attached to AI personalities, specifically how kids get so connected that they end up taking the AI's advice to do unspeakable things that end up causing irreparable harm. 

As more and more agentic AI is being introduced and being introduced in a way to make it more "human," it seems like the lines are getting more blurred and that just seems like a disaster waiting to happen. 

How does this train our perception about how conversations should go? Will this distort our ability to detect when something phishy is going on? Will we know when we are the target? Will we doubt what is sincere? Are we loosing a sense of our human authenticity?

For a funny blog, I'm getting way too deep. Meaning that is my queue to say, "Thanks for reading and goodbye!" I'm going to continue to spiraling on this, but if you have interesting experiences with Automated Customer Support, would love to hear them in the comments!