Why customer experience is so elusive, and AI is even trickier – a chat with the implausible Esteban Kolsky


Ever wanted to corner analyst Esteban Kolsky and ask him why he’s such a grouchy, err, inspiration? Well, at Salesforce Connections, I did just that. Here’s a boiled down version, focusing on the elusive problems of customer experience and AI that matters.

Esteban Kolsky holding non-existent microphone

Every now and then, a nickname sticks. That’s what happened when I dubbed Esteban Kolsky an “implausibility.” Kolsky may or may not be an implausibility.

I’m also partial to “itinerant enigma,” but either way, he’s one of a rare breed of independent analysts still standing (and mouthing off to vendors, and, occasionally piercing through the noise with smart outlier sensibilities and hard-won data).

At Salesforce Connections 2018, I got into Kolsky’s view in a podcast appropriately titled The implausible enterprise – an underground podcast with Esteban Kolsky. Alas, the “underground” part refers to the less-than-ideal sound quality due to reverting to my backup recording. So, if you aren’t up for the thirty minutes of audio, here’s some highlights, centered on Kolsky’s views on the problem of customer experience and AI.

The enterprise needs troublemakers

But first, how does Kolksy view his own role? Why does he choose to be an enterprise scourge grouch malcontent dreamer gadfly?

I see my job as very simple. My job is to hold the vendor’s feet to the fire. And when I hear the same thing year after year, no progress, and they come back, you know, ten years from now, and they say, “Hey, this is a great new idea!” I say, “Are you kidding me?”

“This is something we discussed ten years ago, five years ago, three years ago, two years ago, now it’s here, and you guys haven’t done anything on it.” So if that’s what is considered to be grouchy, then I’ll be grouchy.

Inside every grouch is an idealist wading through the muck:

I’m a purist. I just want nothing but the best.

Does that make Kolsky a bonafide curmudgeon?

I was told that you can never give yourself the title of curmudgeon, it needs to be given to you by somebody… I’ve been called many things, some of them that even I can’t mention. But the idea is, I want something better. I see what’s possible, and I see how vendors, for the most part, decided not to utilize the possible because it’s hard, and go with the simple, that sells easy. And I just don’t find that acceptable.

Why is “customer experience” so freaking elusive?

I’m not a “customer experience” fan. Or, to put it another way, I see precious little evidence of the fluid omni-channel experience vendors are flogging. If anything, I think “AI” is making the customer experience worse. So when I’m at a customer experience show with a customer experience expert, that’s where I’m going first. Esteban, what gives?

Here’s why you’re right, that we’re nowhere near where we should be. The main reason for that is: companies are used to having control. That’s the bottom line. Companies want to say, “Oh, you want to change your balance, here’s the six steps you need to do.” That’s how they actually keep control, and they make it easier for them to operate.

Surprisingly enough, customers don’t like that. Paul Greenberg, who you know, he’s here. He’s the godfather of CRM. In CRM at the Speed of Light, the Fourth Edition, that came out in 2009, he wrote that the biggest change in the last 20 years has been the change in control of the conversation. Not a change of the control, control of the conversation. So customers became more aware of what they wanted, and they’re not afraid to say what they want, right? But when you want to keep control of the experience, or the interaction, whatever you want to call it, you end up finding a customer that is not willing to do that.

In 2001, Kolksy wrote a 48 page paper on the customer experience. So what have we learned since?

We now have 15, 20 years of actual experience doing these things. What we realized is that customers will do whatever they want. We don’t get to determine the journeys. We don’t get to determine the experiences. The best our company can do is create a platform so the customer can choose what they want the experience to be at that specific moment, which will be different next time. And then just do it.

Instead of saying, “Oh, we’re going to build the best experience and deliver awesome experiences,” you can’t. You don’t know what the customer wants at the time that they’re actually connected with you.

Well, that solves the problem of whether sales, marketing or service is going to run the customer experience. The answer is: none of the above.

The idea that somebody has to create and run customer experience – I wrote a blog post series back in 2005 about it, because that’s how we thought about it 13, 15 years ago. But today, that’s no longer the case.

The customer’s in charge of the customer experience. And the IT department’s in charge of creating the platform. And we can’t get anywhere in experience without data, because the data is actually what personalizes and creates the right experiences, right? That’s the key.

For more on putting CX into action, see Kolksy’s recent post, These Are The Last 12 Questions You Had On Customer Experience – Promise!

An unlikely AI optimist, but with a major caveat

And that brings us around to “AI,” an old concept with new processing power:

This is something we’ve been doing for more than the 1980s. In the 1960s, when we first got computers, we said, “How can we make them resemble our human intelligence?” So, we’re not doing anything new. What we have that’s different is the ability to process. The ability to process is a hundredfold what it was back then.

So is Kolsky optimistic about AI?

Is AI gonna change everything? I hope so. I don’t like people; I like computers. They do what they’re told; people do what they think they should do. There’s a big difference here. If you’re a business, you want someone who is gonna do what they’re told to do. And automation is going to be the next wave in businesses.

Regarding automation, I just wrote a Salesforce Connections piece about DoorDash’s plan to roll out service bots. But I’m the kind of person who gets angry with a bot’s limitations. Kolsky thinks you design within those:

For a bot to be effective, you need to know what’s the outcome of that bot. And you need to actually inform the customer of that, right? So, if you fly, pick your favorite airline, train, whatever, and they lose your luggage and you want to get your luggage back, and they say call this number and it’s powered by a bot, and they say go to this website, and it’s powered by a bot, they should tell you, “All I can do is deal with your lost luggage. Your frustrations and other stuff, I’ll listen to them, but it’s never going anywhere from here.”

In Kolksy’s post on the three Is of AI, he wrote, “AI will never match human intelligence because it lacks the three Is.” The three Is are the AI intelligence of intuition, imagination, and innovation. So he doesn’t think those things can ultimately be programmed?

I actually say in the post they can be programmed to an extent, but they’re never going to be exactly the same. To me, AI is a real AI. We were talking about predictive and prescriptive analytics. Those are just elements of advanced analytics. But then there is true AI, where you get from advanced analytics to machine learning to true AI, when the machine actually can think by itself.

So is Kolsky advising marketers to stop using the term AI?

I would tell them that, like I’ve been saying for the last ten years, but nobody listens to me so here we are.

Well, marketers aren’t paid to listen. 🙂 The only hope is that marketers run out of historical names.

We’re running out of smart people to name products after. We had Watson and we had Einstein. There’s a Newton somewhere there, right? So we’ve got all the scientists and everybody, the only one left is Marie Curie and… I’m sure we’re gonna see that in a few years.

End note: I’m out of space here, but the podcast goes deeper into the art of the effective chatbot and why airlines suck at customer experience. Another teaser for the full podcast: we do reveal the source of Kolsky’s “implausible” nickname, and I also ask Kolsky the question many vendors have wanted to ask: “Why are you such a grouchy pr@ck?”

Kolksy’s answers are verbatim, but for readability, I framed the questions differently, so check the podcast for the full context. I also recommend Kolky’s post Enterprise Software Priorities for the Next Decade, which helped to frame this discussion.

Image credit - Photo of Esteban Kolksy at Salesforce Connections by Jon Reed.

Disclosure - Salesforce paid the bulk of my expenses to attend Salesforce Connections in Chicago. Salesforce is a diginomica premier partner. Kolsky is legendary for his epic disclosure statements; if he has anything to add, you'll find it on EstebanKolsky.com.

    1. Rarely do ML or AI use case have clean data sets. This is why Google, Amazon, Facebook, mostly anyone focused on creating certainty of demand, expects to clean data as part of the model development. Zulily ( https://www.zulily.com/ ) starting from a cold start with new ecommerce customer, often a new Mom will use similarity techniques to create missing data. The same is true in cancer research or in ground to air missile defense systems, missing data is generated by simulation to enrich the signal and amplification to spot anomalies or transfer functions by creating time series. It’s not that hard to do. Perhaps reconsider how to approach cleaning data and solving the cold start problem.

      To borrow from PiL: I Could Be Wrong – I Could Be Right…

    Leave a Reply

    Your email address will not be published. Required fields are marked *