Suumit Shah, the CEO of Bengaluru-based Dukaan, said the chatbot answered customer queries in 2 minutes — a task that took the humans over 2 hours.
I’ve worked in customer service software companies for the last 30 years, and one thing I can tell you is that average handle time is not a good metric to decide your success or failure on.
Having a low average handle time is easy. Just hang up on the customer. Or show them quickly that you won’t do shit for them so that they hang up on you.
How about showing us those customer satisfaction and first call resolution scores?
deleted by creator
This is so true. The ability for human beings to game any metric you put out there is pretty legendary. I’ve seen it in action so many times. Measure people on a single metric and they will sacrifice every other aspect of your business to make that metric look good.
Reminds me of the story about the British India Company putting a bounty on snales to combat an infestation. People started breeding snakes to claim the bounties. When the government caught up to that, they stopped giving bounties for snales heads. So people released all their snakes into the wild, making the infestation worse than it was.
Do you mean that GDP is not the most important measure of a success of a country?
Isn’t the GDP itself a roll-up of dozens or hundreds of other metrics, making it nigh impossible to game? Gaming the metric is the problematic part of identifying metrics to track performance being referred to.
No, GDP (gross domestic product) is an amount of money produced by states’ economy. That’s why you can have falling wages and lots of people in poverty yet if the stock market is doing well the government can boast how everything is just fine.
show them quickly that you won’t do shit for them so that they hang up on you.
I do believe this is the reason why AI is so much faster than humans at this guy’s company.
Reminds me of my local Rally’s switching to an “automated drive thru assistant.” The jank thing doesn’t even respond when you talk to it, just reminds you every 60 seconds that it’s ready when you are. First time I went I drove off. Went a second time thinking it might have been a fluke and I’d get actual human service. Nope. Guess I’ll be finding a new burger place for my hangovers
If the majority of the service requests are that quick, then it’s probably something you can automate or by providing a knowledge base. It’s the complex problems that require a human and I see us needing that for the foreseeable future
Yes well we will never know any of that based on the metrics he’s using to define success.
We’re going to be watering plants with Gatorade soon, we’re on our way.
It’s what plants crave!
It’s got electrolytes!
Water? Like out of the toilet?
Did I miss the part about customer satisfaction? Guy could have just moved from solving customer issues in 2 hours to aggravating and loosing customers in under 2 minutes.
Yeah, like, how do you even help someone in two minutes?? They probably just see “oh, it’s a bot” and leave
Without any ratings for customer satisfaction. I might as well sack the entire support staff, don’t bother with AI and I’ll get a answered query to F off in 0 minutes and 100% savings.
I mean this is what teslas PR email does, or is it Twitter… it’s one of those lol
I think it’s twitter. Journalists often contact twitter because of controversies, and they try to highlight that Twitter always says nothing useful
This fails to say by what metric the bots are more efficient. Unless it’s just time-to-first-response. That’s the only metric referenced and it’s a stupid one if it’s the only metric.
I’m pretty sure most of those layoffs that are contributed to AI are just dumb CEOs that a) buy into the hype that AI makes human workers superflous (which is just completely wrong at this point) and b) just needed a reason to fire a few people to get a bonus.
It was interesting that the stats they were talking about were time to respond and time to close, which are both key customer service stats. I’d be interested to know what the customer satisfaction rating was.
If I message, and someone answers immediately, but I figure out it’s a bot and I’m not getting anywhere after a minute, I stop and leave a bad review. From a time standpoint, the interaction looks great. When you integrate the CSAT score, it’s terrible. A quick response contributes to a good interaction, but it doesn’t make it good outright, unless you don’t actually care about whether customers are helped.
Another paragraph masquerading as an article. Ironically, probably written by AI…
AI is not the enemy, if we humans have to work less that’s fine, just increase social programs so we can relax at home while the computer works.
And someone else gets paid.
Imagine he goes to his own support bot just to get told to F off
I like how that free market suddenly starts working now with technologies allowing idiots to kill themselves that easily.
It just has to stay more agile than the big guys and this really does work.
Why is he getting roasted? Seems like he improved efficiency
Because he’s only measuring how long the interactions were, not whether anything was accomplished for the customer or whether the customer was happy.
You know the easiest way to have a short interaction with a customer? Just hang up on them. Or piss them off quickly so they hang up on you.
Because according to Reddit and Twitter, CEO = Bad. More than likely people are just knee-jerk reacting to the headline without reading the headline that says 90% of a company’s staff lost their job to a chatbot.
The guy employed 60 people, almost half of which were support staff (26 people, less the 23 that got laid off). I’m assuming this was a text-based support link, so yeah if this guy could get his customers better support (2 min via chatbot vs 2 hours via live person), and the results are satisfactory, I’d do the same thing.
Some people on the internet seem to think that companies are supposed to be job charities, while failing to realize that put in the same position, they’d do the exact same thing.
Edit: Clearly this didn’t go over very well. I’m not trying to say that letting those people go is a win for society, or that I welcome our AI overlords to displace our jobs. My point is that from a business standpoint, if an effective solution saves the company money, any company would be somewhat foolish not to implement it. And truth be told, this is only the beginning of the labor market in the coming years. It would be in everyone’s best interest to develop skills that are much more difficult to be replaced by a computer.
job charities
So can we then dispel the capitalist notion that they are “job creators”?
Clearly this is not the case as we see here, you even seem to cheer this sacking jobs on as a good thing.No, I do agree that capitalists are job creators, for better or worse (and I do wish our society wasn’t constructed the way it is, but that’s a different thing all together), but that isn’t the point I was intending on addressing. My implication on this particular instance is that, as a business owner, the guy found a solution to improve customer response time (2 min vs 2 hours), and that it was apparently successful enough that he felt that those jobs weren’t necessary. Don’t get me wrong, it’s a major bummer to those 23 people who now have to find new jobs, nor would I ever celebrate someone losing their income. But if they’re were that easily replaced by a computer, it’s hard to say those jobs were in fact necessary. Almost a bullshit jobs conundrum (David Graeber). Hopefully they find something more fulfilling. I know this sounds very anti-human, and that is not my intent. I’m saying all this from an objective business standpoint.
😐
I believe you’re simplifying the situation a bit & overlooking all aspects of ‘job performance.’ I do agree w/ your sentiment that jobs are/should not be a given, but from what I know about AI I’m struggling to believe it’s ready for nuanced roles in customer service.
It’s not ready for prod.
I was thinking technical questions about the software or UX, like “where is setting X” or "how do I achieve LMNOP, or perhaps questions with billing and such. I did a customer support job years ago after high school, and most of the questions were basic RTFM type things that people had either overlooked or just didn’t read the manual, and left me bored out of my mind. Sure, for more technical or in depth questions, it makes sense to talk to a person, and I’d assume that’s why the guy kept 3 people. Clearly my previous comment ruffled some feathers.
