March 26, 2025 • 33 mins
In this episode of Let’s Talk Pricing, PPS President Kevin Mitchell sits down with Matthew Knaggs, Senior Business Value Lead at Zilliant, to explore the evolving relationship between AI and human expertise in pricing.
As AI continues to reshape pricing strategies, where do human pricers fit in? Matt shares insights on how strategic thinking and interpersonal skills can enhance AI-driven decisions, real-world examples where human intuition outperforms algorithms, and how to build trust in AI across cross-functional teams.
📌 In This Episode, You’ll Learn:
- How AI is transforming modern pricing strategies—and why human expertise is still essential
- Real-life examples where human insight fills AI’s blind spots
- Practical steps to align AI outputs with broader business goals
- How to foster confidence in AI tools across your organization
💡 Want to dive deeper? Matt will be leading a workshop at PPS profitABLE: Dallas called “Pricers Assemble! Defending Humanity in a World of AI.” Don’t miss your chance to learn directly from him and other industry leaders—register today at pricingsociety.com/ppsdallas25.
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
(00:00):
Hello, everyone. Welcome to Let's Talk Pricing,
the official podcast of the Professional Pricing Society.
I'm Kevin Mitchell from PBS, andI'm thrilled to have you with us
today. In this episode, we're going to
dive into a topic that is very relevant as artificial
intelligence, AI continues to shape the business world and the
(00:21):
world of pricing and revenue management and related fields.
As businesses increasingly turn to AI for pricing decisions, for
business decisions, the questionwill arise, where does human
expertise fit into this equation?
And today, I'm very, very happy that we are going to be joined
on Let's Talk Pricing with Matt Nags.
(00:44):
Matt is the senior business value lead at Zilliant.
Matt, thank you so much for joining us.
It's great to have you on the show.
Yeah, really excited to be here,Kevin, and excited to talk about
a topic that I'm super passionate about.
All right. And of course, we have a lot of
questions, a lot of information coming in from all over the
place about AI. So, Matt, thank you very, very
(01:09):
much. I know that you're an expert in
this area, so we are going to ask you to clear some things up
for us. We'll start with a couple of
discussion, discussion questionsfrom our audience here.
So, Matt, first of all, to kick things off, can you tell us a
little bit more about the role of artificial intelligence AI in
pricing today and how it's reshaping traditional pricing
(01:33):
strategies? Yeah.
And to cover that, I think it's important to talk about the
context of kind of today throughthe lens of what happened before
today, because AII think we're starting to see where people,
when they say AI today, what they're typically referring to
(01:53):
is generative AI, right? Ever since ChatGPT exploded on
the scene. And since then, we've got all
these other transformer models, deep learning models, and these
these new generative AI tools. When we say AI, that's what
first comes to mind. But there are all these
(02:14):
different categories of AI, manyof which have been around for
decades and decades, right? Technically going all the way
back to the late 1940s, early 1950s.
And in the world of pricing, we've seen machine learning and
algorithmic pricing for decades.That's that's not new in the
(02:36):
world of pricing. What's starting to change,
what's getting more exciting is the ability for people to
partner with generative AI in their day-to-day to look at
pricing within their organization to identify
opportunities in ways in which previously they've been unable
(02:58):
to do so. So with that traditional pricing
approach through machine learning AI, that is what we
would consider deterministic AI,meaning we've got calculations
and algorithms and rules that are going to ensure reliable,
(03:19):
consistent outcomes over time. So we build pricing models based
on that approach and we can expect the same results over and
over again given identical inputs.
Now, when we shift to generativeAI, we're no longer in that
realm of deterministic, but we're in probabilistic AI.
(03:43):
What that means is that these generative AI models, what
they're doing is they're taking all of the available context, so
data that they've been trained on, and then they are coming up
with a output based on what theybelieve to be the most probable
correct answer. And so that is going to lead to
(04:06):
a lot of what we're talking about, Kevin, in terms of where
do humans come in? Because given the natural
language abilities of Gen. AI, Gen.
AI can be very convincing, very believable.
Like, oh man, that's a great answer.
I'm just going to run with that.And there's a risk in doing that
(04:27):
without having some level of human interaction to validate
and inform what the actual finaldecision should be relative to
pricing. And really there are still
limitations and gaps when it comes to generative AI in
pricing. If you think about it from a
(04:47):
data-driven approach, we can only use the data available to
us in order to make a decision. So if we're feeding data into
any kind of pricing model, whether that be the traditional
type that's deterministic and rules based or this more
generative approach, probabilistic even, so there may
(05:11):
be other data points that would be what we would call blind
spots if we're thinking about external influences.
Right now, tariffs are a hot topic, right?
Do you actually have data flowing into whatever model
you're using that's going to leverage the current information
available around what's happening on the tariff side of
(05:34):
the world in order to inform what your prices should be?
And more often than not, in fact, I, I have not yet found a
company that is saying, yeah, we, we've got real time
information coming in right now where we can incorporate that
readily, right? That's where humans come into
play, as one example, because they have the additional
(05:56):
context, experience and intuition that is going to help
them make the right decision fortheir business.
Understood. And that's a very key point and
thank you for your description of how we've gone from
deterministic to probabilistic with some of these models.
And Matt, you probably know, butone of the great old pricing
(06:19):
examples of a more deterministic, you know, more of
an algorithm within pricing is there was a textbook, I think it
was called The Life of the Fly or something like that.
And if you've been around pricing, you may have heard this
example. Or classic.
Story. A classic story, but basically
this ordinary textbook that probably a few 100 people on the
(06:40):
planet even care about at one point was priced at several
$1,000,000 on Amazon. Because Amazon had a model that
basically took the Barnes and Noble or a competitor's price
and said we're Amazon, we're going to go 1 or 2% over that
price. But that competitor had an
algorithm that said we're going to match Amazon's price.
(07:01):
And of course it became this self fulfilling, self
perpetuating thing where the next thing you know, this
textbook that probably was worthfifty, 100 bucks at most went
for two point whatever, $1,000,000.
So it's interesting how we've seen the development in AI and
things there, but we still obviously, as you mentioned, do
need those human guidelines and rules.
(07:24):
And we know game theory, we knowhow other humans may react and
also all of this comes into play.
So that's a very good description there.
And Matt, I want to ask and alsowant to mention one other thing.
So the next PPS conference in Dallas from May 6th, very, very
(07:45):
excited you're going to join us for a keynote where you're going
to talk about some of these things here.
So in your upcoming presentationwith us, in your upcoming
keynote, you're going to talk about how human Pricers, how we
as human beings can complement AI tools using strategic
thinking, interpersonal skills, game theory, some of these human
(08:06):
intuitions and human knowledges that the AI driven pricing
decisions probably can't do on their own.
So can can you give us some examples of how we as human
beings, how this human element that we possess can enhance AI
driven pricing decisions? Yeah, definitely.
First I got to, I got to go backto your textbook example and I'm
(08:30):
going to use that as kind of like a double pun, right,
Because it's a classic example and it's literally about a
textbook. So a shameless plug for our
ethics and bias in Gen. AI course that was recently put
to the PPSAI certification program as well that I did.
(08:50):
And I used that exact example, right?
And at the heart of that exampleis one of the things that's
really important moving forward in this ever changing world of
AI. And that is the need for what's
called now because people want to sound cool and use acronyms
HITL, which means human in the loop, right?
(09:12):
And the idea there is that even with the best conceived ideas
around what an AI model should be doing, and we think we've set
up all the correct parameters and boundaries, we still have to
ensure accountability for what comes from these models.
And so still we need, even with as fancy as AI is getting, we
(09:37):
need a human in the loop. And the idea there is that in
that textbook example, a human, had they been aware of what was
going on, would have said, Oh mygosh, we can't sell a book for
$1,000,000 or I think it got up to 17 million or something like
that, if I recall, before they actually caught the issue and
(09:59):
stopped it, right. So ensuring that humans maintain
decision accountability and steer the ethical usage of AI
and set the correct boundaries. That even means once I've got a
model set up, I still need a human to oversee that and ensure
(10:20):
it's actually doing what I expect.
Now, not only that, but humans have a specific set of skills
that at least today, AI does notyet have.
Now AI can imitate some of thoseskills with the the latest model
from ChatGPT Open AI, by the way, 4.5, they're even touting
(10:45):
that it's more empathetic, right?
But even that empathy that comesthrough the model is still
programming, and it's not actualreal human empathy.
It's impressive the way these models are maturing and
developing over time, but we have not yet reached a point in
(11:05):
our collective society where we're saying, yes, that empathy,
for instance, is enough. It's valid if it's coming from
the machine. I'm going to say that carefully
because I don't want 20 years from now a machine to point to
this interview. And, and how dare you, Matt?
(11:25):
But but just acknowledging our reality today, we haven't
evolved to the point of embracing that yet as a
replacement for human interaction.
And so when we talk about empathy as an example, it's a
soft skill. You know, many years ago, people
might have said, ah, you don't need empathy in business.
(11:47):
But what's often referred to as the last mile and pricing, which
is how we actually execute basedon here, here's our pricing
strategy, here's the pricing that we're putting in front of
our sales team. Sales team is then expected to
carry that across the line to the customer.
Well, when my sales team comes back to me and they say, you
(12:08):
know what, yeah, you, you've given us this price.
But here's why in this particular situation, it's not
going to work if I can't respondto that with my actual humanity,
right? Even if ultimately we're going
to have some debate and I'm going to say no, we are still
going to go with this. That's going to do a lot to
(12:31):
erode the trust that's there, right?
And that's that's critical if wewant our pricing strategy to
actually be able to be realized in the day-to-day.
We've got to ensure that there'strust and trust only comes from
being able to have those real human interactions where we're
going to have an honest conversation.
(12:53):
We can show that we understand the challenges being brought to
us. And I think I said at the start
of this, if not I meant to, if we get to the point where we
say, well, I hear you, Mr. Salesguy, but AI said this is the
price. So it's the price.
Well, then we've completely goneoff the rails.
(13:16):
And I, I don't see a point ever where we're going to be able to
take humans completely out of the situation to where that
level of validation, both from an emotional empathetical
standpoint is no longer required.
And also validation to be able to say, well, based on my
(13:38):
experience, here are the things that I'm seeing, not what Jen AI
is telling me, but here's what'shappening in the market.
Can we agree on those things? OK, here's what's happening in
your particular case with this customer.
Let's talk about how that impacts the prices that we're
putting in front of them and being able to show that I'm not
just somebody taking what was presented to me and saying copy
(14:04):
paste that into reality. But I can at least explain it,
justify it, validate it, or say,you know what, maybe you're on
the something. Maybe there's an additional
piece of context here. And we need to either fine tune
the model, we need to make a rules based exception for this
particular case and that can be incorporated into the model.
(14:27):
But that level of understanding that you're only going to get
from a human pricer with actual experience and strength of those
human characteristics, that willcontinue to be relevant.
Yeah, absolutely. And some key things that you so
eloquently put out in that is that as pricers, we have to be
(14:51):
artists and scientists. It's never one or the other.
We also have to use empathy and sympathy and talk with our
internal teams as you mentioned,talking with sales teams in a
language that they can understand and the AI said the
price is 2.3 million, 16,000,000for this book is not a language
that anyone with any real knowledge could understand.
(15:13):
So we definitely have to have that in there.
We have to be able to relate with our colleagues, be they
inside our company, be they our senior management, be they our
marketplace, our customers or whomever on terms that they can
understand, that they can relatewith.
We have to use those human nuances, those ways of basically
(15:39):
using our language in order to make that connection with them.
Another question for you, you are going to discuss some other
things in your presentation withus.
So you're also going to talk about how to build trust in AI
recommendations across cross functional teams.
So what are some ways that companies can foster confidence
(16:02):
in these AI tools and get different teams to align on
pricing strategies? Yeah.
Today I see a lot of organizations taking many
different approaches or almost taking what I'll call a no
approach approach, right? Because there's so much
uncertainty around not only what's out there today, but also
(16:25):
what tomorrow looks like. It's almost every day it seems
like there's news dropping aboutsome huge advancement in the
world of AI or quantum processing or all these things
which, you know, are developing at a pace unlike anything we've
seen in the past. And so one of the things that's
critical to establishing trust is first adopting an AI strategy
(16:50):
as a company and communicating that to everyone, right?
Because for many people, there is already this fear of AI is
coming for my job, right? That that has been stated for
quite some time now. You're, you're hearing more
people say, no, AI is not comingfor your job.
(17:11):
Humans who leverage AI are coming for your job.
OK, there may be some truth to that as well.
But the idea is, if I want to create trust, first I've got to
have confidence and lack of fearon my own part.
So where does that come from? It comes from starting to get
(17:31):
exposure. If I'm not using AI today, then
part of my strategy organizationally should be how
do I begin to train my employeeson what I think they should be
learning from an AI perspective and getting them more
comfortable with it? Because we often fear what we
(17:52):
don't know or what we don't understand.
And what can work really well tobuild trust is to start to find
opportunities for quick wins with leveraging some of these
newer AI tools. Now, I would recommend starting
with lower risk quick wins, right?
We don't want to start out the gate with, hey, let's create a
(18:16):
brand new pricing model using generative AI and you know it'll
be safe AB testing, we're only going to do half of our
customers on that new model. Don't do that, right?
That's not OK. What I mean is really starting
small, and that might be throughanalytical capabilities, that
(18:37):
might be through brainstorming. I mean, Gen.
AI in particular is great for what's often referred to as
thought partnership. I've got this hypothesis that
there may be a better way to do something within my pricing
approach. So I'm going to download
everything that's in my brain into this Gen.
(18:59):
AI tool and have a brainstormingsession to either validate or
invalidate my theory as much as I can.
Now again, part of the training needs to focus on understanding
the difference between Gen. AI and deterministic AI.
So that means that even in thesebrainstorming sessions, which
(19:22):
can be very powerful and useful,it's still a probabilistic
response that you're getting. And these responses do have a
high degree of accuracy, but there's also the potential for
what's called hallucinations from these Gen.
AI models where the Gen. AI model doesn't really know,
(19:44):
but it's been programmed to givea convincing response.
And so it gives you something that sounds great.
And without validating whether or not that is something you
should run with, there's a risk.So understanding as a part of
training that. Even Gen.
AI, as as great as it is, still has its gaps and still has
(20:06):
hallucinations, although to a lesser degree today than two
years ago. That's something that's that's
really important. Outside of that, looking for
organizational usage through, I'll say cross functional
alignment. So we see in some companies
(20:28):
where they're starting to have like an AI council and there's
representation from different functional areas talking about,
hey, what are the opportunities for us to leverage new AI tools
within our respective areas? And then also potentially cross
functionally. And the more exposure we start
(20:48):
to give to the rest of the organization around what we're
planning to do, how it's going to impact individuals within the
company and how they can reskillor upskill.
So that way they're ready for these changes that come.
That'll do a lot to improve the trust overall, I'll say, in
(21:09):
their company's approach. But then there's still the trust
in AI itself. And that only comes from actual
adoption and usage and importantly, explain ability.
That explainability piece looks really different if we're
talking about machine learning algorithmic AI models or if
(21:31):
we're talking the probabilistic Gen.
AI models. And so to improve the trust in
those, it's still relatively thesame approach.
And that is a before and after session, we'll say where, hey,
here's what the model was predicting.
So here's how we responded and here were the results.
(21:54):
So when we look at that over time and see the intended
consequences actually happening,that does a lot to build the
trust of the AI tools that we'reusing as well.
Now, that's not always going to happen.
It may be here was the intended consequence that we were after.
(22:15):
It didn't happen. And without an understanding of
why, then there is risk of trusterosion in the AI.
So that goes back to that training and knowledge on the
tools that you're using and the types of technology you're
using. Because more often than not, if
(22:36):
the outcome you're after didn't happen, if someone is really
experienced both in the realm ofAI and also we're talking
specifically around pricing, they're going to be able to say,
look everybody, it's not that this AI tool is broken.
Here's what we missed. We didn't include these very
(22:57):
important pieces of context thatwould have helped us arrive at a
different conclusion or there were these other external
factors, you know, huge supply chain disruption or a new entry
into the market that wasn't a part of what we built into this.
(23:17):
And so the explainability portion will always come down
to, I think, a human validating and being able to explain,
here's what worked, here's what didn't work.
If we don't have that, you know,you can, you can scroll YouTube,
Twitter, LinkedIn for Gen. AI sucks.
(23:40):
Look at the question I asked andlook at what it gave me, and
you'll get nerds like me saying here's why you got that
response. Here's how you could prompt
better to get a better response.But that's an example of a skill
set that is going to be growing more and more in importance.
(24:01):
The more that we're looking to leverage this technology, the
more that we're going to have tobe aware of the ways in which we
can best interact with the technology.
That's very interesting and I like how your your descriptions
about AI recommendations and howwe can get everyone aligned on
(24:21):
AI driven pricing decisions. A lot of that matches up pretty
closely with some of the well known pricing strategies and
tactics that we've been talking about for a long time.
And many of these will not be a surprise to people in our
audience. In particular when you talked
about you don't start with an ABtest where you're going to have
(24:42):
50% doing A, 50% doing B and andcompare and contrast there.
That's very similar to the classic pricing strategy of when
you have a brand new strike, a pricing incentive or a pricing
strategy. Basically, to use an American
example, we might start geographically with Rhode Island
(25:03):
and see how it works in a smaller territory and then
extrapolate that, multiply that and see what the effects might
be for California, Texas, for a larger state, for our friends in
EMEA. I'll make up an example.
We'll start testing this in Luxembourg and then we'll try
the EU 5, France, Germany, the UK, Italy and Spain and so on
(25:24):
and so forth there. So it's very interesting how
even with this new tool and withthis new knowledge, a lot of as
how you've mentioned a lot of the human intuition, the human
empathy, sympathy, that thought processes and things like that
still apply when you're going todo testing and when you're going
to try to use this as a tool in order to help with your
(25:47):
strategies and tactics. There.
We are running a little short ontime.
So we will have, how about one more question here.
As businesses look to implement AI in their strategies, what are
some of the key steps they should take to ensure that the
AI outputs align with their broader business objectives and
(26:08):
their long term success? So there there are a couple of
things that that I would recommend and, and one is
ensuring that, you know, we're not doing AI just for the sake
of AI, right. Right now, you're seeing where
it's almost a mandate in any kind of, you know, 10K quarterly
earnings call, anything like that where companies have to be
(26:32):
calling or you know, calling that out that that is something
strategically that they're tackling or else their investors
will be in a tizzy, right? So if we're doing AI, we have to
have a clear purpose for it and ensure that it aligns to the
overall strategic goals of our organization.
(26:53):
Not only that, but there's a lotof challenge right now and being
able to actually measure the theKPIs around whatever we've done
from an AI perspective. So it's just as important that
not only are we aligned to our organizational strategic goals,
but that before we set out on implementing some new AI
(27:16):
approach that we've aligned on what are the metrics by which
we're going to measure success. Because just to do it and to be
able to point back and say we did it is not going to be very
helpful, right? It doesn't confirm for us that
it was something that we should have done.
So we have to start with the outcome in mind.
(27:37):
And that's going to help us to identify the measures, the KPIs
that will help us determine how effective we actually were.
We also need to really differentiate between the
various AI approaches we can take.
Are we going to use deterministic AI?
Are we going to use probabilistic AI?
(27:59):
Or will it be a blend? Because the reality is we've got
well established history that shows there's a lot of validity
in that deterministic AI, that traditional algorithmic based
approach to pricing is still valid.
So we don't just throw that out the window and say Gen.
(28:20):
AI is so exciting and it's new and it's great, so it's better.
The truth is it's not better in all aspects of where we might
apply AI. So you have to be very measured
in your approach on where do I use which one?
And then how do I ensure that human accountability factor
(28:42):
ensuring that I've got a human in the loop for any type of
process that I'm using? Because that's ultimately going
to also impact my governance to if we're talking within the
context of pricing, my ethical usage, right?
The monitoring of, you know, risk or, or even the risk of
customer perception if we just let a new pricing model run
(29:04):
unconstrained. And then the customers that we
have look at the prices we put in front of them and say, oh, I
don't know that that was the right approach to take.
We can actually have a really negative impact on our brand
perception if we are just running hands free without
humans in the loop. So, yeah, all of that and then
(29:28):
just underscoring continual learning and training, right.
That's, that's where it really needs to begin.
I've, I've heard some companies say we're waiting to start until
things level out and there's more certainty about what's
coming. Those companies will be at a
disadvantage in the future because things are changing.
(29:50):
They're not going to slow down. They're not going to stabilize
anytime soon. So even for those companies that
are saying we're waiting, well, let's say whatever that future
point is, is 2 years from now and they say, OK, now we're
ready. They're at A2 year disadvantage
because in addition to using whatever's available today,
they're going to have that lack of skills from having interacted
(30:12):
with those tools, from those approaches that they could have
been upskilling themselves on upuntil that point.
So the biggest take away is don't wait to get started, dive
in, become familiar, get experience and continue to learn
about what is available because this is an ever changing field
(30:36):
and there will be new use cases,new application where human
creativity is going to say, you know what, nobody's talking
about this, but we could use this particular type of AI to do
this. And at the end of the day, that
human expertise, creativity, critical thinking is going to
(30:59):
drive, I think, the most impact for how we can use AI moving
forward. Understood.
And Matt, thank you so much for our insightful conversation.
I would imagine that some of ourlisteners are going to want to
hear a lot more from you on this.
So everyone please reach out to Matt Nags at Zillion For more
(31:20):
information. And also we look forward to
seeing you very soon in May at the PPS Profitable conference.
We're going to be in Dallas fromMay 6th to 9th.
And Matt, your keynote is entitled Pricers Assemble,
defending humanity in a world ofAI, so we're looking forward to
that as well. Can you tell us briefly about
(31:41):
some of the things that you're going to be talking about with
us when you join us in Dallas inMay?
Yeah. So we'll be looking at the the
current landscape and what some of the predictions are just
looking at five years ahead of now.
The World Economic Forum took place in January.
A lot of the discussions there are around the economic impact
(32:03):
and the impact to the jobs landscape as the result of
everything going on within the realm of AI.
So we're going to be looking at some of those trends we're going
to be talking about. How does that actually apply to
pricing professionals specifically, and what are the
key things that pricing professionals need to know to
(32:25):
navigate the ever changing world?
So I'll be using as many practical examples as I can, as
well as personal examples that hopefully the audience will find
relatable with stories from my own world as well.
And then plenty of, I'll say probably dad jokes and sci-fi
(32:46):
movie references and hopefully alittle bit of humor as well to
keep things fun and engaging. Yeah, we're looking forward to
some great information about AI,about humans can partner
essentially with AI and how we are still needed for a lot of
things that we talked about earlier.
And of course, we're looking forward to a couple of great dad
(33:07):
jokes and sci-fi references. So everyone, please join us at
PBS Profitable in Dallas from May 6th to 9th.
You can always reach out to me with more details about this
conference, about our other offerings, about the AI
certification that Matt and his team have helped us develop as
well in conjunction with quite afew other experts in the field.
(33:30):
And for our listeners, thank youso much for tuning in.
Make sure to subscribe, make sure to keep pricing smart and
stay ahead of the curve. And we are looking forward to
seeing everyone soon. Thanks everyone.
Hello, everyone. Welcome to Let's Talk Pricing,
the official podcast of the Professional Pricing Society.
I'm Kevin Mitchell from PBS, andI'm thrilled to have you with us
today. In this episode, we're going to
dive into a topic that is very relevant as artificial
intelligence, AI continues to shape the business world and the
(00:21):
world of pricing and revenue management and related fields.
As businesses increasingly turn to AI for pricing decisions, for
business decisions, the questionwill arise, where does human
expertise fit into this equation?
And today, I'm very, very happy that we are going to be joined
on Let's Talk Pricing with Matt Nags.
(00:44):
Matt is the senior business value lead at Zilliant.
Matt, thank you so much for joining us.
It's great to have you on the show.
Yeah, really excited to be here,Kevin, and excited to talk about
a topic that I'm super passionate about.
All right. And of course, we have a lot of
questions, a lot of information coming in from all over the
place about AI. So, Matt, thank you very, very
(01:09):
much. I know that you're an expert in
this area, so we are going to ask you to clear some things up
for us. We'll start with a couple of
discussion, discussion questionsfrom our audience here.
So, Matt, first of all, to kick things off, can you tell us a
little bit more about the role of artificial intelligence AI in
pricing today and how it's reshaping traditional pricing
(01:33):
strategies? Yeah.
And to cover that, I think it's important to talk about the
context of kind of today throughthe lens of what happened before
today, because AII think we're starting to see where people,
when they say AI today, what they're typically referring to
(01:53):
is generative AI, right? Ever since ChatGPT exploded on
the scene. And since then, we've got all
these other transformer models, deep learning models, and these
these new generative AI tools. When we say AI, that's what
first comes to mind. But there are all these
(02:14):
different categories of AI, manyof which have been around for
decades and decades, right? Technically going all the way
back to the late 1940s, early 1950s.
And in the world of pricing, we've seen machine learning and
algorithmic pricing for decades.That's that's not new in the
(02:36):
world of pricing. What's starting to change,
what's getting more exciting is the ability for people to
partner with generative AI in their day-to-day to look at
pricing within their organization to identify
opportunities in ways in which previously they've been unable
(02:58):
to do so. So with that traditional pricing
approach through machine learning AI, that is what we
would consider deterministic AI,meaning we've got calculations
and algorithms and rules that are going to ensure reliable,
(03:19):
consistent outcomes over time. So we build pricing models based
on that approach and we can expect the same results over and
over again given identical inputs.
Now, when we shift to generativeAI, we're no longer in that
realm of deterministic, but we're in probabilistic AI.
(03:43):
What that means is that these generative AI models, what
they're doing is they're taking all of the available context, so
data that they've been trained on, and then they are coming up
with a output based on what theybelieve to be the most probable
correct answer. And so that is going to lead to
(04:06):
a lot of what we're talking about, Kevin, in terms of where
do humans come in? Because given the natural
language abilities of Gen. AI, Gen.
AI can be very convincing, very believable.
Like, oh man, that's a great answer.
I'm just going to run with that.And there's a risk in doing that
(04:27):
without having some level of human interaction to validate
and inform what the actual finaldecision should be relative to
pricing. And really there are still
limitations and gaps when it comes to generative AI in
pricing. If you think about it from a
(04:47):
data-driven approach, we can only use the data available to
us in order to make a decision. So if we're feeding data into
any kind of pricing model, whether that be the traditional
type that's deterministic and rules based or this more
generative approach, probabilistic even, so there may
(05:11):
be other data points that would be what we would call blind
spots if we're thinking about external influences.
Right now, tariffs are a hot topic, right?
Do you actually have data flowing into whatever model
you're using that's going to leverage the current information
available around what's happening on the tariff side of
(05:34):
the world in order to inform what your prices should be?
And more often than not, in fact, I, I have not yet found a
company that is saying, yeah, we, we've got real time
information coming in right now where we can incorporate that
readily, right? That's where humans come into
play, as one example, because they have the additional
(05:56):
context, experience and intuition that is going to help
them make the right decision fortheir business.
Understood. And that's a very key point and
thank you for your description of how we've gone from
deterministic to probabilistic with some of these models.
And Matt, you probably know, butone of the great old pricing
(06:19):
examples of a more deterministic, you know, more of
an algorithm within pricing is there was a textbook, I think it
was called The Life of the Fly or something like that.
And if you've been around pricing, you may have heard this
example. Or classic.
Story. A classic story, but basically
this ordinary textbook that probably a few 100 people on the
(06:40):
planet even care about at one point was priced at several
$1,000,000 on Amazon. Because Amazon had a model that
basically took the Barnes and Noble or a competitor's price
and said we're Amazon, we're going to go 1 or 2% over that
price. But that competitor had an
algorithm that said we're going to match Amazon's price.
(07:01):
And of course it became this self fulfilling, self
perpetuating thing where the next thing you know, this
textbook that probably was worthfifty, 100 bucks at most went
for two point whatever, $1,000,000.
So it's interesting how we've seen the development in AI and
things there, but we still obviously, as you mentioned, do
need those human guidelines and rules.
(07:24):
And we know game theory, we knowhow other humans may react and
also all of this comes into play.
So that's a very good description there.
And Matt, I want to ask and alsowant to mention one other thing.
So the next PPS conference in Dallas from May 6th, very, very
(07:45):
excited you're going to join us for a keynote where you're going
to talk about some of these things here.
So in your upcoming presentationwith us, in your upcoming
keynote, you're going to talk about how human Pricers, how we
as human beings can complement AI tools using strategic
thinking, interpersonal skills, game theory, some of these human
(08:06):
intuitions and human knowledges that the AI driven pricing
decisions probably can't do on their own.
So can can you give us some examples of how we as human
beings, how this human element that we possess can enhance AI
driven pricing decisions? Yeah, definitely.
First I got to, I got to go backto your textbook example and I'm
(08:30):
going to use that as kind of like a double pun, right,
Because it's a classic example and it's literally about a
textbook. So a shameless plug for our
ethics and bias in Gen. AI course that was recently put
to the PPSAI certification program as well that I did.
(08:50):
And I used that exact example, right?
And at the heart of that exampleis one of the things that's
really important moving forward in this ever changing world of
AI. And that is the need for what's
called now because people want to sound cool and use acronyms
HITL, which means human in the loop, right?
(09:12):
And the idea there is that even with the best conceived ideas
around what an AI model should be doing, and we think we've set
up all the correct parameters and boundaries, we still have to
ensure accountability for what comes from these models.
And so still we need, even with as fancy as AI is getting, we
(09:37):
need a human in the loop. And the idea there is that in
that textbook example, a human, had they been aware of what was
going on, would have said, Oh mygosh, we can't sell a book for
$1,000,000 or I think it got up to 17 million or something like
that, if I recall, before they actually caught the issue and
(09:59):
stopped it, right. So ensuring that humans maintain
decision accountability and steer the ethical usage of AI
and set the correct boundaries. That even means once I've got a
model set up, I still need a human to oversee that and ensure
(10:20):
it's actually doing what I expect.
Now, not only that, but humans have a specific set of skills
that at least today, AI does notyet have.
Now AI can imitate some of thoseskills with the the latest model
from ChatGPT Open AI, by the way, 4.5, they're even touting
(10:45):
that it's more empathetic, right?
But even that empathy that comesthrough the model is still
programming, and it's not actualreal human empathy.
It's impressive the way these models are maturing and
developing over time, but we have not yet reached a point in
(11:05):
our collective society where we're saying, yes, that empathy,
for instance, is enough. It's valid if it's coming from
the machine. I'm going to say that carefully
because I don't want 20 years from now a machine to point to
this interview. And, and how dare you, Matt?
(11:25):
But but just acknowledging our reality today, we haven't
evolved to the point of embracing that yet as a
replacement for human interaction.
And so when we talk about empathy as an example, it's a
soft skill. You know, many years ago, people
might have said, ah, you don't need empathy in business.
(11:47):
But what's often referred to as the last mile and pricing, which
is how we actually execute basedon here, here's our pricing
strategy, here's the pricing that we're putting in front of
our sales team. Sales team is then expected to
carry that across the line to the customer.
Well, when my sales team comes back to me and they say, you
(12:08):
know what, yeah, you, you've given us this price.
But here's why in this particular situation, it's not
going to work if I can't respondto that with my actual humanity,
right? Even if ultimately we're going
to have some debate and I'm going to say no, we are still
going to go with this. That's going to do a lot to
(12:31):
erode the trust that's there, right?
And that's that's critical if wewant our pricing strategy to
actually be able to be realized in the day-to-day.
We've got to ensure that there'strust and trust only comes from
being able to have those real human interactions where we're
going to have an honest conversation.
(12:53):
We can show that we understand the challenges being brought to
us. And I think I said at the start
of this, if not I meant to, if we get to the point where we
say, well, I hear you, Mr. Salesguy, but AI said this is the
price. So it's the price.
Well, then we've completely goneoff the rails.
(13:16):
And I, I don't see a point ever where we're going to be able to
take humans completely out of the situation to where that
level of validation, both from an emotional empathetical
standpoint is no longer required.
And also validation to be able to say, well, based on my
(13:38):
experience, here are the things that I'm seeing, not what Jen AI
is telling me, but here's what'shappening in the market.
Can we agree on those things? OK, here's what's happening in
your particular case with this customer.
Let's talk about how that impacts the prices that we're
putting in front of them and being able to show that I'm not
just somebody taking what was presented to me and saying copy
(14:04):
paste that into reality. But I can at least explain it,
justify it, validate it, or say,you know what, maybe you're on
the something. Maybe there's an additional
piece of context here. And we need to either fine tune
the model, we need to make a rules based exception for this
particular case and that can be incorporated into the model.
(14:27):
But that level of understanding that you're only going to get
from a human pricer with actual experience and strength of those
human characteristics, that willcontinue to be relevant.
Yeah, absolutely. And some key things that you so
eloquently put out in that is that as pricers, we have to be
(14:51):
artists and scientists. It's never one or the other.
We also have to use empathy and sympathy and talk with our
internal teams as you mentioned,talking with sales teams in a
language that they can understand and the AI said the
price is 2.3 million, 16,000,000for this book is not a language
that anyone with any real knowledge could understand.
(15:13):
So we definitely have to have that in there.
We have to be able to relate with our colleagues, be they
inside our company, be they our senior management, be they our
marketplace, our customers or whomever on terms that they can
understand, that they can relatewith.
We have to use those human nuances, those ways of basically
(15:39):
using our language in order to make that connection with them.
Another question for you, you are going to discuss some other
things in your presentation withus.
So you're also going to talk about how to build trust in AI
recommendations across cross functional teams.
So what are some ways that companies can foster confidence
(16:02):
in these AI tools and get different teams to align on
pricing strategies? Yeah.
Today I see a lot of organizations taking many
different approaches or almost taking what I'll call a no
approach approach, right? Because there's so much
uncertainty around not only what's out there today, but also
(16:25):
what tomorrow looks like. It's almost every day it seems
like there's news dropping aboutsome huge advancement in the
world of AI or quantum processing or all these things
which, you know, are developing at a pace unlike anything we've
seen in the past. And so one of the things that's
critical to establishing trust is first adopting an AI strategy
(16:50):
as a company and communicating that to everyone, right?
Because for many people, there is already this fear of AI is
coming for my job, right? That that has been stated for
quite some time now. You're, you're hearing more
people say, no, AI is not comingfor your job.
(17:11):
Humans who leverage AI are coming for your job.
OK, there may be some truth to that as well.
But the idea is, if I want to create trust, first I've got to
have confidence and lack of fearon my own part.
So where does that come from? It comes from starting to get
(17:31):
exposure. If I'm not using AI today, then
part of my strategy organizationally should be how
do I begin to train my employeeson what I think they should be
learning from an AI perspective and getting them more
comfortable with it? Because we often fear what we
(17:52):
don't know or what we don't understand.
And what can work really well tobuild trust is to start to find
opportunities for quick wins with leveraging some of these
newer AI tools. Now, I would recommend starting
with lower risk quick wins, right?
We don't want to start out the gate with, hey, let's create a
(18:16):
brand new pricing model using generative AI and you know it'll
be safe AB testing, we're only going to do half of our
customers on that new model. Don't do that, right?
That's not OK. What I mean is really starting
small, and that might be throughanalytical capabilities, that
(18:37):
might be through brainstorming. I mean, Gen.
AI in particular is great for what's often referred to as
thought partnership. I've got this hypothesis that
there may be a better way to do something within my pricing
approach. So I'm going to download
everything that's in my brain into this Gen.
(18:59):
AI tool and have a brainstormingsession to either validate or
invalidate my theory as much as I can.
Now again, part of the training needs to focus on understanding
the difference between Gen. AI and deterministic AI.
So that means that even in thesebrainstorming sessions, which
(19:22):
can be very powerful and useful,it's still a probabilistic
response that you're getting. And these responses do have a
high degree of accuracy, but there's also the potential for
what's called hallucinations from these Gen.
AI models where the Gen. AI model doesn't really know,
(19:44):
but it's been programmed to givea convincing response.
And so it gives you something that sounds great.
And without validating whether or not that is something you
should run with, there's a risk.So understanding as a part of
training that. Even Gen.
AI, as as great as it is, still has its gaps and still has
(20:06):
hallucinations, although to a lesser degree today than two
years ago. That's something that's that's
really important. Outside of that, looking for
organizational usage through, I'll say cross functional
alignment. So we see in some companies
(20:28):
where they're starting to have like an AI council and there's
representation from different functional areas talking about,
hey, what are the opportunities for us to leverage new AI tools
within our respective areas? And then also potentially cross
functionally. And the more exposure we start
(20:48):
to give to the rest of the organization around what we're
planning to do, how it's going to impact individuals within the
company and how they can reskillor upskill.
So that way they're ready for these changes that come.
That'll do a lot to improve the trust overall, I'll say, in
(21:09):
their company's approach. But then there's still the trust
in AI itself. And that only comes from actual
adoption and usage and importantly, explain ability.
That explainability piece looks really different if we're
talking about machine learning algorithmic AI models or if
(21:31):
we're talking the probabilistic Gen.
AI models. And so to improve the trust in
those, it's still relatively thesame approach.
And that is a before and after session, we'll say where, hey,
here's what the model was predicting.
So here's how we responded and here were the results.
(21:54):
So when we look at that over time and see the intended
consequences actually happening,that does a lot to build the
trust of the AI tools that we'reusing as well.
Now, that's not always going to happen.
It may be here was the intended consequence that we were after.
(22:15):
It didn't happen. And without an understanding of
why, then there is risk of trusterosion in the AI.
So that goes back to that training and knowledge on the
tools that you're using and the types of technology you're
using. Because more often than not, if
(22:36):
the outcome you're after didn't happen, if someone is really
experienced both in the realm ofAI and also we're talking
specifically around pricing, they're going to be able to say,
look everybody, it's not that this AI tool is broken.
Here's what we missed. We didn't include these very
(22:57):
important pieces of context thatwould have helped us arrive at a
different conclusion or there were these other external
factors, you know, huge supply chain disruption or a new entry
into the market that wasn't a part of what we built into this.
(23:17):
And so the explainability portion will always come down
to, I think, a human validating and being able to explain,
here's what worked, here's what didn't work.
If we don't have that, you know,you can, you can scroll YouTube,
Twitter, LinkedIn for Gen. AI sucks.
(23:40):
Look at the question I asked andlook at what it gave me, and
you'll get nerds like me saying here's why you got that
response. Here's how you could prompt
better to get a better response.But that's an example of a skill
set that is going to be growing more and more in importance.
(24:01):
The more that we're looking to leverage this technology, the
more that we're going to have tobe aware of the ways in which we
can best interact with the technology.
That's very interesting and I like how your your descriptions
about AI recommendations and howwe can get everyone aligned on
(24:21):
AI driven pricing decisions. A lot of that matches up pretty
closely with some of the well known pricing strategies and
tactics that we've been talking about for a long time.
And many of these will not be a surprise to people in our
audience. In particular when you talked
about you don't start with an ABtest where you're going to have
(24:42):
50% doing A, 50% doing B and andcompare and contrast there.
That's very similar to the classic pricing strategy of when
you have a brand new strike, a pricing incentive or a pricing
strategy. Basically, to use an American
example, we might start geographically with Rhode Island
(25:03):
and see how it works in a smaller territory and then
extrapolate that, multiply that and see what the effects might
be for California, Texas, for a larger state, for our friends in
EMEA. I'll make up an example.
We'll start testing this in Luxembourg and then we'll try
the EU 5, France, Germany, the UK, Italy and Spain and so on
(25:24):
and so forth there. So it's very interesting how
even with this new tool and withthis new knowledge, a lot of as
how you've mentioned a lot of the human intuition, the human
empathy, sympathy, that thought processes and things like that
still apply when you're going todo testing and when you're going
to try to use this as a tool in order to help with your
(25:47):
strategies and tactics. There.
We are running a little short ontime.
So we will have, how about one more question here.
As businesses look to implement AI in their strategies, what are
some of the key steps they should take to ensure that the
AI outputs align with their broader business objectives and
(26:08):
their long term success? So there there are a couple of
things that that I would recommend and, and one is
ensuring that, you know, we're not doing AI just for the sake
of AI, right. Right now, you're seeing where
it's almost a mandate in any kind of, you know, 10K quarterly
earnings call, anything like that where companies have to be
(26:32):
calling or you know, calling that out that that is something
strategically that they're tackling or else their investors
will be in a tizzy, right? So if we're doing AI, we have to
have a clear purpose for it and ensure that it aligns to the
overall strategic goals of our organization.
(26:53):
Not only that, but there's a lotof challenge right now and being
able to actually measure the theKPIs around whatever we've done
from an AI perspective. So it's just as important that
not only are we aligned to our organizational strategic goals,
but that before we set out on implementing some new AI
(27:16):
approach that we've aligned on what are the metrics by which
we're going to measure success. Because just to do it and to be
able to point back and say we did it is not going to be very
helpful, right? It doesn't confirm for us that
it was something that we should have done.
So we have to start with the outcome in mind.
(27:37):
And that's going to help us to identify the measures, the KPIs
that will help us determine how effective we actually were.
We also need to really differentiate between the
various AI approaches we can take.
Are we going to use deterministic AI?
Are we going to use probabilistic AI?
(27:59):
Or will it be a blend? Because the reality is we've got
well established history that shows there's a lot of validity
in that deterministic AI, that traditional algorithmic based
approach to pricing is still valid.
So we don't just throw that out the window and say Gen.
(28:20):
AI is so exciting and it's new and it's great, so it's better.
The truth is it's not better in all aspects of where we might
apply AI. So you have to be very measured
in your approach on where do I use which one?
And then how do I ensure that human accountability factor
(28:42):
ensuring that I've got a human in the loop for any type of
process that I'm using? Because that's ultimately going
to also impact my governance to if we're talking within the
context of pricing, my ethical usage, right?
The monitoring of, you know, risk or, or even the risk of
customer perception if we just let a new pricing model run
(29:04):
unconstrained. And then the customers that we
have look at the prices we put in front of them and say, oh, I
don't know that that was the right approach to take.
We can actually have a really negative impact on our brand
perception if we are just running hands free without
humans in the loop. So, yeah, all of that and then
(29:28):
just underscoring continual learning and training, right.
That's, that's where it really needs to begin.
I've, I've heard some companies say we're waiting to start until
things level out and there's more certainty about what's
coming. Those companies will be at a
disadvantage in the future because things are changing.
(29:50):
They're not going to slow down. They're not going to stabilize
anytime soon. So even for those companies that
are saying we're waiting, well, let's say whatever that future
point is, is 2 years from now and they say, OK, now we're
ready. They're at A2 year disadvantage
because in addition to using whatever's available today,
they're going to have that lack of skills from having interacted
(30:12):
with those tools, from those approaches that they could have
been upskilling themselves on upuntil that point.
So the biggest take away is don't wait to get started, dive
in, become familiar, get experience and continue to learn
about what is available because this is an ever changing field
(30:36):
and there will be new use cases,new application where human
creativity is going to say, you know what, nobody's talking
about this, but we could use this particular type of AI to do
this. And at the end of the day, that
human expertise, creativity, critical thinking is going to
(30:59):
drive, I think, the most impact for how we can use AI moving
forward. Understood.
And Matt, thank you so much for our insightful conversation.
I would imagine that some of ourlisteners are going to want to
hear a lot more from you on this.
So everyone please reach out to Matt Nags at Zillion For more
(31:20):
information. And also we look forward to
seeing you very soon in May at the PPS Profitable conference.
We're going to be in Dallas fromMay 6th to 9th.
And Matt, your keynote is entitled Pricers Assemble,
defending humanity in a world ofAI, so we're looking forward to
that as well. Can you tell us briefly about
(31:41):
some of the things that you're going to be talking about with
us when you join us in Dallas inMay?
Yeah. So we'll be looking at the the
current landscape and what some of the predictions are just
looking at five years ahead of now.
The World Economic Forum took place in January.
A lot of the discussions there are around the economic impact
(32:03):
and the impact to the jobs landscape as the result of
everything going on within the realm of AI.
So we're going to be looking at some of those trends we're going
to be talking about. How does that actually apply to
pricing professionals specifically, and what are the
key things that pricing professionals need to know to
(32:25):
navigate the ever changing world?
So I'll be using as many practical examples as I can, as
well as personal examples that hopefully the audience will find
relatable with stories from my own world as well.
And then plenty of, I'll say probably dad jokes and sci-fi
(32:46):
movie references and hopefully alittle bit of humor as well to
keep things fun and engaging. Yeah, we're looking forward to
some great information about AI,about humans can partner
essentially with AI and how we are still needed for a lot of
things that we talked about earlier.
And of course, we're looking forward to a couple of great dad
(33:07):
jokes and sci-fi references. So everyone, please join us at
PBS Profitable in Dallas from May 6th to 9th.
You can always reach out to me with more details about this
conference, about our other offerings, about the AI
certification that Matt and his team have helped us develop as
well in conjunction with quite afew other experts in the field.
(33:30):
And for our listeners, thank youso much for tuning in.
Make sure to subscribe, make sure to keep pricing smart and
stay ahead of the curve. And we are looking forward to
seeing everyone soon. Thanks everyone.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
New Heights with Jason & Travis Kelce
Football’s funniest family duo — Jason Kelce of the Philadelphia Eagles and Travis Kelce of the Kansas City Chiefs — team up to provide next-level access to life in the league as it unfolds. The two brothers and Super Bowl champions drop weekly insights about the weekly slate of games and share their INSIDE perspectives on trending NFL news and sports headlines. They also endlessly rag on each other as brothers do, chat the latest in pop culture and welcome some very popular and well-known friends to chat with them. Check out new episodes every Wednesday. Follow New Heights on the Wondery App, YouTube or wherever you get your podcasts. You can listen to new episodes early and ad-free, and get exclusive content on Wondery+. Join Wondery+ in the Wondery App, Apple Podcasts or Spotify. And join our new membership for a unique fan experience by going to the New Heights YouTube channel now!