Latest / The Innovation Show / AI Value Gap | Why AI Investments Fail to Pay Off | David Rogers

Transcript

AI Value Gap | Why AI Investments Fail to Pay Off | David Rogers · The Innovation Show

0:00
a value driver is a lever that when moved will change the profitability of the
0:06
business in a way that matters, right?
0:08
And so it depends, what industry you're in, what's your business model maybe,
0:11
where do you sit within, that business.
0:14
Looking at an industry like oil and gas producers,, physical plant
0:18
operations is incredibly important.
0:21
So minimizing any downtime will reduce the cost of production, right?
0:26
So this is a commodity kind of business, so cost of production becomes critical.
0:30
And so that's where you start to see AI solutions that really drive value.
0:36
So one real-world example some of these companies are doing, and have just
0:40
started to do, is AI drone applications, where You have drones using AI and
0:44
machine vision to inspect, physically inspect, different parts of operations
0:49
and plants on a continuing basis.
0:52
And this is enabling a better kind of preventative maintenance, that is
0:56
really showing up in the numbers, right?
0:59
They are actually cutting their downtime, you know, in a big plant,
1:02
a production facility of some kind, sometimes thirty, even forty percent.
1:06
That saves hundreds of millions of dollars because that,
1:09
again, that's the bottleneck.
1:10
That's absolutely critical to them.
1:12
if you find the problems that really matter to your business, right?
1:16
If you put your money where the value is in your business, that is gonna
1:22
start pointing you towards that real change in the way you do that business,
1:26
Before we begin, I wanna thank our sponsor, Kyndryl, who runs
1:29
and reimagines the technology systems that drive advantage for
1:32
the world's leading businesses.
1:34
With a unique blend of AI-powered consulting built on unmatched managed
1:38
service capability, Kyndryl helps leaders harness the power of technology for
1:42
smarter decisions, faster innovation, and a lasting competitive edge.
1:46
You can find out more about Kyndryl and the Kyndryl Institute
1:49
at K-Y-N-D-R-Y-L, kyndryl.com.
1:52
When an AI initiative fails to show up in financial results, leadership
1:56
instinctively points fingers at technology bottlenecks or cultural resistance.
2:02
They assume the algorithms were not advanced enough, the enterprise
2:05
data was too messy, or the frontline resisted changing its workflow.
2:10
Today's guest will show us that those assumptions actually miss the mark.
2:15
The shortfall is almost never an engineering flaw or a
2:18
change management failure.
2:20
We are once again graced by the presence of the author of The Digital
2:25
Transformation Roadmap and The Digital Transformation Playbook, and of
2:29
course, The Network Is Your Customer.
2:31
His hit books, his hit albums.
2:34
He is David Rogers, the OG of digital transformation.
2:38
Welcome back
2:39
Thank you.
2:39
Thank you, Aidan.
2:40
It's great to be back on the Innovation Show.
2:42
Great as always to talk with you
2:44
I'm delighted to have you back, man.
2:45
Any excuse I can get.
2:46
And I saw a brilliant article you published, and it was
2:50
with a co-author this time.
2:51
It was a HBR article, but it was also on your Substack, which I highly
2:55
recommend, and we always recommend from the Innovation Show Substack as well.
3:00
But I thought we'd share the concept of that, this idea of the AI value
3:05
gap, and this concept of the value gap actually comes up in your work as well.
3:09
So maybe you'll give us a bit of context in that, and then we'll dig into some
3:13
of the core concepts from that article
3:15
So the AI value gap became apparent to both myself and my co-author,
3:19
Krishnan Sankaranarayanan, through our experiences, you know, working with
3:25
and talking with a lot of executives.
3:27
You know, myself and my teaching and my advising really across a lot of different
3:32
sectors and industries, financial services, you know, professional services,
3:36
industrial, companies of many kinds.
3:38
Krishnan has a deep experience over two decades in specifically oil, gas, and
3:43
chemicals, as an executive, sort of really making… You know, industry where key
3:47
investment decisions really pay off or don't in the matter of, you know, hundreds
3:51
of millions or even billions of dollars.
3:53
So the conversation we kept seeing is this kind of awkward conversation now where
3:58
companies have been investing, they've been spending money on AI projects, right?
4:04
It's not just sort of a, a, a little show and tell.
4:07
Like, they're putting real money in as these token budgets increase and
4:10
as their, you know, distribution of the tools through large companies
4:14
kind of really takes hold.
4:16
yet there's a point where you're talking with, the CFO, right?
4:21
And the question is, "Okay, what are we getting for this?" Like, what's
4:24
our, what's our measurable return?
4:27
Not like, "Oh, people are using it, and we've got some good feelings," and, "Oh,
4:31
we have a use case we can point to." But what's the measurable financial return?
4:36
And it often gets really awkward, right?
4:38
At that point.
4:39
So then there's some kind of hemming and hawing and like, "Well, we're
4:42
gonna see that in a little bit," or, or, you know, "That's not the
4:45
stage we're at," or something.
4:47
But everybody knows that is actually the goal, and I think we have been
4:52
losing a little sight of that.
4:53
So that was really the focus of the, the article, was looking at can we
4:57
understand why this is happening.
4:59
And, you know, our thesis is very simply that this problem is a strategy problem
5:06
it's so like the idea of digital transformation, and
5:09
I called you the OG there.
5:10
That's because you wrote the world's first book on digital transformation.
5:14
And we saw the same thing.
5:16
I, I think I told you at the time, I was working as head of a digital in
5:19
a media company, and the CEO at the time asked me for a digital strategy.
5:24
And I went, "I'll give you a digital strategy if you give me a company
5:27
strategy." Obviously, obviously I wasn't very popular for saying that.
5:33
But it's, it's, consistently the challenge, and in absence of that,
5:39
everybody's busy, but they're not busy even pointed in the same direction.
5:43
And I was reminded by, a regular listener to the show, a guy called John Sheridan,
5:48
shout out to you, John, of a joke that was told by David Allen, who is the
5:52
author of the book, Getting Things Done, a hit book, like your own books as well.
5:56
And he, now pointedly because I'm an Irish man, but I'm, I'm weirdly an
6:01
Irish man who doesn't drink very much.
6:03
Yes.
6:05
this joke goes as follows.
6:07
So the Irish man gets a wish from a genie, and his first wish was
6:11
a never-ending pint of Guinness.
6:13
So he drinks that pint of Guinness, and it refills magically once again.
6:17
He drinks it again, it magically refills.
6:20
He's astonished.
6:21
He thanks the genie, and the genie, impressed with his
6:25
manners, grants him a second wish.
6:27
So the Irish man looks at the bottomless pint of Guinness that he
6:30
has in his hand, and he goes, "I'll have another one of these, please."
6:36
And, , the reason I- it come to mind for me was in, in the age of AI and the
6:42
access to these state-of-the-art tools that we have, we consistently use them
6:47
for trivial problems, problems that actually don't really move the dial.
6:52
And again, in absence of that strategy, we- we're trying to get things done,
6:57
but we're not getting them done in a, wi- with meaningful impact.
7:01
And you say in the article that this is organizations deploying state-of-the-art
7:07
tools against trivial problems, and you see it in enterprises routinely
7:11
pouring resources into things like smoothing out internal administrative f-
7:16
frictions that irritate senior leaders but barely move any to- operating costs.
7:21
You say again, rolling out marginal project, product enhancements that
7:25
leave customers completely indifferent, they don't even notice, copying
7:29
generic use cases marketed by ent- enterprise software vendors simply
7:33
because competitors are doing so, and you give a list of other reasons.
7:37
And that's how so many of us are using AI, me included, at the start, and we
7:44
make the excuse, the cognitive dissonance of, well, I'm just get, figuring it out.
7:48
But when it comes to actually meaningfully, meaningful usage of AI,
7:53
we stop, and I'd love you to share that idea, and then we'll give, we're gonna
7:58
give some solutions today, so we're not just gonna moan about the challenges here.
8:02
A-a-absolutely.
8:03
Never, never, never stop with the moaning.
8:06
I mean, o-o-one of the ways we, we, we talk about this, and one of the ways I
8:10
think about this that seemed to resonate with folks who read the article, we got
8:13
a lot of great feedback, not just from you, but from many other, uh, readers,
8:18
is, you know, think about, think about a lot of applications of AI right
8:22
now, and they're about time-saving, and yet the context is absolutely
8:27
critical to whether value occurs or not.
8:30
You know, because a lot of time-saving, if you think of using the technology, and
8:34
it can save time in a lot of places, a lot of tasks, but if you apply it to a routine
8:39
task and say you have one minute, right?
8:41
That a worker is saving on a task, what typically happens, right?
8:45
In most business cases, they save a little bit of time,
8:48
and it just disappears, right?
8:50
It's reabsorbed into sort of their slack.
8:53
it's not even that they're consciously saying, "Oh, I'm gonna go take
8:56
a coffee break," or something.
8:57
It just, it just kind of vanishes.
9:00
Um, and so there's no measurable result in terms of the firm level, right?
9:04
That task was just something that was being done, wasn't on a critical path.
9:07
It didn't really change something significant downstream.
9:11
But in a different business, you could use the same AI model, save one minute
9:16
of time, and see a really significant, a transformative even business impact.
9:21
And that is because you chose a place to apply it that sits
9:25
at a bottleneck, something that is gating throughput, right?
9:28
Loading an extra product onto a truck for shipment, or maybe it is a task
9:33
that is repeated of times a day, right?
9:37
Or maybe it's a one minute saved when a worker is exposed
9:40
to a hazardous environment.
9:41
You know, there's certain places where even one minute can actually
9:45
create a measurable difference in your business, but those are really
9:47
unique, special places, and unless you are thoughtful about where you
9:52
are applying the technology and the work to build an application and a
9:56
solution, this is where we see the value.
9:59
We see what we call activity without return, meaning you're doing a lot of
10:03
stuff with it, and it seems to be kind of worth doing, uh, when you look at the
10:08
individual kind of task or tool or, oh, the agent's doing something, and it's
10:12
doing it accurately, and yet there's no measurable return to the business
10:18
in the article at a high level.
10:19
You give the example of Netflix, Nike.
10:22
You give the example of Merck and Nestle.
10:24
Maybe you'll give us some meaningful impacts that you have seen out there with
10:28
people using AI so we have a contrast
10:31
Sure.
10:31
So, Nike, one of the ones you just mentioned, is a great example.
10:35
Um, you know, Nike, of course, is a… We're all familiar with the,
10:38
the, the basic business model.
10:40
It's a, you know, retail, consumer goods manufacturer and, and increasingly
10:44
to a degree retailer of its own products through digital channels.
10:49
one of the key things as more and more is sold through digital channels, a, a
10:52
really important economic driver for its business has to do with inventory, right?
10:58
Holding inventory, carrying inventory, and striking this balance.
11:01
Because if you are holding more inventory than you need, you are tying up capital,
11:06
and it is expensive literally just to actually house it, you know, h- a-
11:11
a- have people keep track of it, and, you know, pay for the, the, the, the
11:14
humidification, you know, the HVAC of, of the rooms it's being stored in.
11:18
But also, you're simply sinking your capital, getting it tied up, sitting
11:21
on, on the shelves, so to speak.
11:23
On the other hand, if your inventory is too low, the right product at the right
11:28
time is not available for the right customer, and then there's a delay, you're
11:32
losing out on revenue, you're hurting the customer experience, you're reducing
11:36
customer, loyalty and lifetime value.
11:39
So it's a really critical driver that impacts cash flow
11:42
and customer satisfaction.
11:44
So that, I think, is an example of a really smartly chosen
11:48
problem to focus on, right?
11:50
And that is, I wanna end… Before you even think about, oh, what was
11:53
the data and the model, and how do they integrate it, and how do they
11:56
integrate that in then into the systems within the fulfillment centers and all
11:59
these things, absolutely important.
12:01
But before you even get there, the critical thing is, I would say the
12:04
most important thing is Nike first did what I call, uh, i, i, i- in the
12:09
Roadmap book, uh, picking a problem that matters, where I say picking
12:13
the problems that matter most, right?
12:15
And so I think - Nike did a good job of picking one of the problems that
12:19
matters most to their business, where they might apply artificial intelligence.
12:23
I, I constantly remind people about that, because if you can show
12:26
some type of small win, you create momentum inside the organization and
12:31
confidence in you if you're driving transformation inside the team.
12:35
But to help people with their, particularly their AI portfolio,
12:38
Krishnan and yourself developed a simple diagnostic framework and a
12:42
little bit of mathematics here for people, but it's at a very high level.
12:46
So this is value equals technology by adoption, by strategic alignment.
12:51
Maybe you'll explain what this means, help people get inside it
12:55
Yeah, this was just sort of a way we wanted to, to convey sort of the high
12:58
level of concept here, and I think it, it has resonated with folks.
13:02
Again, that's the eq-- what we call the AI value equation.
13:06
Value equals technology times adoption times strategic alignment.
13:12
And the point here is all three of these factors are critical.
13:16
The really important point here, why we do it as an equation, is
13:19
these factors multiply, right?
13:21
They don't sum.
13:22
You don't add them up, they multiply.
13:23
So if any of these factors, as it approaches zero, the value collapses, and
13:29
basically you're seeing no business value.
13:31
And, you know, most folks are pretty aware that if the technology is not
13:35
working, right, if the model is not accurate or it's hallucinating, or it
13:39
doesn't have the right data in order to make a prediction or, you know,
13:42
accurate classification, whatever it is, that it's not gonna work, right?
13:46
There's a lot of attention to that.
13:49
I would argue that there's been a lot of attention in recent, you
13:52
know, uh, uh, you know, year or two to the adoption piece, right?
13:56
We're becoming very aware of the fact that there are reasons why employees,
14:00
customers, you know, whoever is the user of an AI model may or may not have
14:05
trust in it or, you know, believe in it or see value in it, um, that we--
14:09
that it's really important to think about we solving for the adoption?
14:14
Are we actually integrating this effectively into the organization?
14:17
Again, a topic I've written about a lot in my, you know, prior books.
14:22
but where we see the highest leverage and the most neglect right now, kind
14:28
of at this moment in history, is the strategic alignment part, people are
14:33
not paying enough attention to that.
14:35
And the fact is, you can have a brilliant model with all the right training and
14:39
all the right data, and you actually get people to use it, but again, the problem
14:43
you chose is simply not high enough value and really you're still stuck with
14:49
what we'd call activity without return.
14:51
You're, you're, you know, keeping folks busy.
14:54
And yes, you can say, "Look, we're now using AI," but you can't really
14:58
say, "Here's how this is showing up on the income statement of our business.
15:02
This is where we see it in our financial numbers at the end of the quarter."
15:07
So you, you talk about that strategic alignment as the most important aspect.
15:11
If that gets skipped, the rest are pretty much meaningless on their own.
15:15
They help, but not that much, and you point out a few common reasons for that.
15:20
Maybe we'll talk about those four that you point out inside the article
15:25
Sure.
15:25
Well, I mean, the one I would mention first is, is ownership.
15:30
I think this might be the most common one.
15:32
So ownership of AI, AI projects within … A-and really this is
15:37
all talking about sort of a larger established companies, right?
15:40
Not AI in your personal use or AI in maybe a small business or a startup, but
15:44
sort of bigger established companies.
15:46
What tends to happen is the ownership, who's gonna build some new thing
15:49
with AI, is very often given to some kind … E- either an IT team or
15:54
maybe some sort of innovation team, or maybe you call it an AI team.
15:58
Whatever name you have it, but that's sort of how it's structured.
16:01
But more importantly, that team is disconnected from P&L responsibility.
16:06
They themselves do not have P&L responsibility.
16:09
And so in a sense, it's kind of back to the old, you know, IT problem of
16:13
build it and throw it over the wall.
16:15
It's like, "Oh, we're gonna build something for the business," and
16:18
then give it to them and say, "Hey, you, you run with this.
16:21
Look at this thing we built for you." Um, that I think is, is, is probably the most
16:26
common reason why we see this happening
16:29
Before you go onto the others, and I'll remind you of the rest of them, the, that,
16:33
that one about ownership, what I have often seen in being this person as well,
16:38
and many of our audience of the show are, have been this person is, you have, you're
16:43
a leader inside an organization, and you go, "Who, who will we give digital to?"
16:47
Or, "Who will we give innovation to?" Or now, "Who will we give AI
16:50
to?" And somebody's like going, "Oh, there's this kid in sector 7G, really,
16:55
like, high potential, loves reading all about this innovation crap.
16:58
He's read Roger's book." You know?
17:00
And then they're like going, "Oh, yeah, bring him in here." And then
17:03
they bring in that p- that guy or girl
17:05
great, but you don't wanna pick of those people and then send them off to build
17:10
it on themselves . You know, you need to sit them down next to people from that
17:16
business unit, and you need someone who owns the P&L to actually sign off on
17:21
b- what problem you are solving and why you're solving it, and just as important,
17:25
how you're gonna measure the results.
17:27
You know, you've gotta keep that ownership of the process and really particularly the
17:32
problem selection with the PL- P&L owner.
17:35
It doesn't mean the P&L owner needs to actually be building the model,
17:38
'cause yeah, right, they probably are.
17:39
You- that's where you want these, you know, brilliant, whatever,
17:43
young or not young man or women or group of people coming in.
17:46
Uh, but the ownership of the problem has to, has to really
17:49
stay with the business itself.
17:51
And the lack of poli- political nous, uh, is a huge challenge there because
17:56
often if you're gonna make a big impact inside the organization, it's gonna
17:59
mean meaningful change to the structure, and they don't have the teeth for that
18:04
Yeah.
18:04
Yeah, yeah.
18:05
They're like, "Well, of course, if you really want it to pay off, you actually
18:08
have to change the way that this department, uh, you know, interacts with
18:11
that department, or the way we, you know, uh, make certain decisions, or the way
18:16
this process, you know, connects from the customer back to, uh, uh, the business
18:20
and, and, and returns to the customer."
18:21
And, uh, y- nobody's, nobody's ready for that, right?
18:25
And they're not saying… They're like, "Yeah, yeah, yeah. Okay, fine, fine.
18:27
Just give us the, give us the widget, give us the tool." Uh, and you're,
18:30
you're not gonna see… You're gonna get some nice-looking technology,
18:33
but you're not gonna see, you know, a, a, a payoff on the bottom line
18:37
One of the other reasons you point out here, so reason two for, for the
18:41
skipping of ali- alignment is that leaders assume that a powerful enough
18:45
model connected to their data will spare them from making those tough choices.
18:49
Kind of aligned to what you were saying there.
18:51
But also, if they have the data at all, so they're, they're trying to get
18:56
these AI to, to create new business models or create new ways of doing
19:00
things, but they haven't got a clean data set at all in the first place
19:05
We do, and this is also, I mean, it's an interesting-- Right now there,
19:10
there, there are folks who are so bought into the, the hype around
19:14
AI, uh, which is a little untethered from the reality inside the business,
19:20
where they actually believe this idea, 'cause you'll hear this told to you by
19:24
the frontier labs, "Oh, don't worry.
19:25
If we can just get all your data organized and readable to the AI,
19:30
you don't have to choose, right?
19:31
You don't have to pick which problems matter most.
19:33
The AI will just-- It'll tell you what's important to your business.
19:36
It'll solve all your problems.
19:38
It's like this magic wand." And that's not true, right?
19:42
These tools are very powerful, but even if you do have your data organized,
19:47
uh, uh, effectively, and you do have a really powerful model, I mean, for
19:52
one thing, don't wanna be using the most powerful model on most of your
19:57
problems, 'cause you're gonna lose money simply because you're overpaying.
20:01
So it's a much more nuanced, subtle thing than people think.
20:05
It's just sort of like, "Give me the, you know, most, most horsepower," and like
20:08
point it at, at my, my, my operations, and voila, magically everything will be fixed.
20:14
No.
20:14
It's really critical to make these trade-offs and these choices.
20:17
I mean, strategy is all about choices.
20:19
It's about choosing what not to do even more than it is about choosing what to do.
20:23
So we're at a point where we have to reflect and reconnect with that kind of
20:30
baseline truth and be really thoughtful about what are the places we are going to
20:35
really double down and really apply this AI effectively and really make sure we do
20:41
it well, because this is where it's gonna have a direct impact on really the unique
20:46
economics of our business, which is gonna be very different than someone else's.
20:51
I was thinking there about my, I ha- so I pay for the top frontier models on,
20:56
say, Claude or Anthropic and ChatGPT, and my kids got access to my login.
21:03
And they were using the l- the top model to do stuff like take a picture
21:08
of my wife and they, "Turn this lady into, uh, Shrek." And I was like,
21:14
"You've, you've used up all my credits.
21:15
You've used up all my credits." Anyway, uh, but another one, speaking
21:21
of that, is point three, which is often leaders think AI is a, is framed
21:27
as, or it is a skill to be learned, and therefore success is measured in
21:32
training completed or tools adopted.
21:34
And we saw that with some of the consulting companies, where people were
21:37
actually given bonuses or recognized on how many credits they'd used, and
21:42
they were using them for all crazy things, like Shrekifying people
21:46
Yeah, yeah.
21:48
you get sort of mis- misalignment of scent- incentives in the extreme cases.
21:51
I think Meta briefly fell into that trap when they said, "Well, let's stop
21:55
putting up a leaderboard for how many tokens did you use." not the goal.
21:59
But I think more broadly, companies have sort of gotten to the point where
22:04
they think, you know, h- let's have…
22:05
You know, we paid for some license, from, you know, , whatever, Copilot
22:09
or, or, or Claude or, or, or ChatGPT.
22:12
, A-and now the main thing is just to get as many people using it as possible.
22:16
And sort of it's devolved s- purely into kind of a training.
22:19
It's like, well, we need to get people comfortable and show them how…
22:22
I mean, you know, a couple years ago it seemed like the key to getting
22:25
value out of AI was knowing how to write these arcane prompts, right?
22:29
And we were literally hiring people as prompt engineers.
22:32
The train has left the station.
22:34
We're not there anymore, thankfully.
22:35
You know you don't need to be a prompt engineer.
22:38
The problem is not training and adoption primarily at this point.
22:42
The bigger problem is, you know, creating value in the firm.
22:45
The last point that I, we mentioned in the article was the, you know, just
22:49
kind of the mindset from leadership.
22:50
The, the en- the whole effort around AI is being very solution-focused,
22:54
starting from the technology, right?
22:56
Well, we just have to do AI.
22:58
And a lot of this is what we'd call, I could say, you know, it's a FOMO
23:01
strategy even more than an AI strategy.
23:04
It's like, just saw a presentation by another company, and they're
23:07
doing this, and they're doing that, and we have to do something similar.
23:11
And you know, there's a lot of evidence now, I mean, Gartner had a
23:15
recent survey this year where they, you know, they said the finding was
23:19
basically that the most popular use cases in your industry are usually not
23:24
the ones that create the most value.
23:26
So just saying, "Let's do what others are doing," is really not helping
23:32
Again, very like what happened with digital transformation is that
23:36
the stuff that usually made the headlines was not often the most
23:39
impactful stuff to the business.
23:41
Didn't even make revenue for the organizations most of the time.
23:43
It was, it was digital transformation theater, just like we see
23:47
innovation theaters, just like we're seeing AI theater today.
23:50
So you do give four steps to strategic alignment, and the nerd that I am,
23:56
David, I know you know I am, I went and I cross-referenced these four points with
24:02
your books behind me there, and I'll point out where where, where that happened.
24:08
So there are four kind of disciplined steps here that you can take.
24:12
So first we talked about this idea of don't always start with the technology.
24:16
Start with a challenge that you have.
24:18
Pick a problem.
24:19
But David also says, firstly, define your value drivers,
24:23
and that rang a bell for me.
24:25
I was like, "Um, I've heard that before," and indeed it did come from the Digital
24:28
Transformation Roadmap from a section that David calls Pick Your Value Drivers.
24:34
And it is such an essential and, and somewhat simple step to do,
24:38
a great e- a great exercise for any team to do, actually, even for
24:41
themselves within an organization, is what are our value drivers?
24:45
So maybe you'll, you'll tell us what they are, what, what are value drivers, and
24:49
then what they are in the context of AI.
24:53
Just a value driver is a lever that when moved will change the profitability of
25:00
the business in a way that matters, right?
25:02
That, that's, that's how I would, , define it.
25:04
And so it depends, you know, what industry you're in, what's your
25:07
business model maybe, where do you sit within, that business.
25:10
But, you know, looking at an industry like oil and gas producers,, physical
25:15
plant operations is incredibly important.
25:18
So minimizing any downtime will reduce the cost of production, right?
25:23
So this is a commodity kind of business, so cost of production becomes critical.
25:28
And so that's where you start to see AI solutions that really drive value.
25:33
So one real-world example some of these companies are doing, and have just started
25:37
to do, is AI drone applications, where they are-- You have drones using AI and
25:42
machine vision to inspect, physically inspect, different parts of operations
25:47
and plants on a continuing basis.
25:50
And this is enabling a better kind of preventative maintenance, that is
25:54
really showing up in the numbers, right?
25:56
They are actually cutting their downtime, you know, in a big plant,
26:00
a production facility of some kind, sometimes thirty, even forty percent.
26:04
That saves hundreds of millions of dollars because that,
26:07
again, that's the bottleneck.
26:08
That's absolutely critical to them.
26:11
So another industry might be something very different.
26:13
It might be procurement, right?
26:15
might be, you know, employee retention or onboarding in a very
26:20
talent-dependent business, something like, you know, professional services.
26:25
It's gonna be very different things, a retailer, right?
26:28
Pricing optimization, inventory management, those are gonna be
26:31
critical value drivers, right?
26:33
So you really need to figure out, is our number one value driver… If I'm Netflix,
26:37
customer retention is much more important even than customer acquisition, right?
26:42
Sure, you wanna do that well and keep the cost down.
26:44
But even more important to a subscription business is customer retention.
26:48
That's why Netflix winds up spending so much of its AI efforts focused
26:52
on things like co- effective content recommendations, right?
26:57
Because they find that the more they're continually finding something that you
27:01
didn't know you wanted to watch, but you do wind up watching, that actually
27:05
greatly reduces the chance that you will unsubscribe in the next thirty days.
27:11
And that is why they continue to beat, year after year, quarter after
27:15
quarter, all of their peers on that critical metric of customer retention.
27:19
It is not just that they have more subscribers.
27:22
There are some big numbers now for total average subscribers for a, a
27:25
Disney or, Amazon Prime, but Netflix is still ahead of everyone on customer
27:31
retention, that sort of loyalty measure, which is such a value driver
27:35
for that subscription business model.
27:37
So that's just an example of, of, like, what does a value driver look
27:40
like, and it's really different from one company to another.
27:43
So value driver number one, value chain is number two.
27:47
And, uh again, I was like going, "We've covered that before." So in The Digital
27:52
Transformation Playbook, you say about this concept of a competitive value train,
27:58
and you say the overriding competitive goal is to gain more leverage in the value
28:03
train than an organization, so you don't become a commodity business, essentially.
28:08
But point two of this article is know where you sit in
28:12
your industry's value chain
28:13
Exactly.
28:14
Right.
28:14
It's not just that, what industry are you in?
28:16
Okay, you're in oil and gas.
28:18
Oh, you're, you're in, uh, media.
28:20
Oh, you're in consulting, whatever it is.
28:23
Where do you sit within that industry?
28:25
Again, that's something I've been writing about for, for some time now in terms of
28:28
competitive dynamics in the digital era.
28:30
But here it's really understanding, and we contrast two different companies that are
28:34
both, , manufacturers of chemicals, right?
28:37
So they're in this- very much in the same industry, but they
28:40
sit at different positions.
28:41
So Westlake, we give as one example, that's a commodity
28:44
manufacturing company, right?
28:46
So they're producing things like chlorine and polyethylene, which
28:50
the world needs a lot of, right?
28:51
And yet chlorine is chlorine.
28:53
There's not a lot of differentiation within the product, right?
28:56
So for them, the way you win, if you're at, sitting at that part of the
29:01
industry, you know, fairly, you know, upstream, if you will, is you've got
29:06
massive processing facilities, right?
29:08
You know, Krishnan used to run one of these in his own…
29:10
Not at Westlake, but, you know, at his own firms.
29:13
Steam crackers, things like this that are producing these, for the world
29:17
market, these, at scale, these, you know, commonly used, chemical inputs.
29:22
And so there, it's kind of like the, the, the maintenance example we gave.
29:27
You know, things relating to operational efficiency,
29:29
downtime, uptime, and so forth.
29:31
That's where you're gonna see a powerful impact with value drivers.
29:35
Givaudin in this example, they're sort of at the other end of the spectrum.
29:39
So they are specialty, right?
29:41
So they are creating chemicals, but they go into things like fragrances, right?
29:45
And so their whole business model is much more oriented towards R&D, for example.
29:51
Uh, they're spending, you know, five to 10 times the R&D spend as a, as a share
29:55
of revenue versus a firm like Westlake, even though they're in the same industry.
29:59
So for them, that is much more where they're gonna real- really get an impact,
30:04
uh, by investing in how can we use a new technology like AI models of various
30:09
kinds to, you know, create intellectual property, to shorten the innovation cycle,
30:14
to integrate our development of new, uh, products into, uh, the work of our, the
30:22
manufacturers who we serve, so that we are the, not even just the preferred,
30:27
but the sort of essential partner of choice for these particular inputs.
30:31
So really radically different value drivers, even though you
30:34
might sit in the same industry.
30:36
That's kind of the key point we're making.
30:38
Lovely.
30:38
And then point three is to trace opportunities directly
30:41
to your income statement.
30:42
And again, in Digital Transformation Roadmap, I remember a line you said,
30:46
"Will your financial return come from selling more, reducing costs, generating
30:52
new sources of revenue, or some mix of these?" And it's something that we always
30:56
miss with these new technologies, is that oftentimes we can have huge savings.
31:01
So you can have an innovation team that perhaps brings new savings
31:04
to the organization, but they're not as recognized as creating new
31:08
value or new profits or new services or products for the organization.
31:13
But they're equally as important because that saving can be
31:15
reinvested then in R&D, for example
31:20
Absolutely.
31:21
And this is so we talk about, in the article about income statement, you
31:24
know, starting from the income statement, because it's good discipline, right?
31:29
look to whenever you're thinking of a new AI application, this is one of the sort
31:33
of most practical points of advice that I-I hope we get across in this article
31:38
is, you know, be out there experimenting.
31:41
We're not saying, you know, don't be ready to try unproven things.
31:45
Absolutely.
31:46
You know, uh, uh, don't be risk-averse, but be focused, right?
31:50
So this is a discipline to help focus.
31:51
Where should we be looking to experiment?
31:54
So start from the income statement, and if you're looking at AI at the
31:58
enterprise level, that's gonna be the company-level income statement.
32:01
If you're within a business unit or a function, you're gonna be looking at
32:04
the financial statement, uh, for that department or for that, you know, business
32:08
unit, however y-you're accounting.
32:10
But you're gonna look at these different items.
32:12
Maybe it's revenue, maybe it's cost of, of goods, you know, COGS, maybe
32:16
it's your operating expenses, and you're gonna start with a hypothesis.
32:20
We believe that if we, uh, apply AI effectively to this value driver, to this
32:28
activity that we do, to this part of our job, it ought to-- if it really works, it
32:32
ought to show up in this number, right?
32:34
Point to the number on the income statement, right?
32:37
That is-- Or start from the income statement and say, "Hey.
32:41
All right, this is the number we wanna g-do.
32:43
We wanna drive down these expenses, or we wanna, uh, uh,
32:46
expand, uh, uh, this revenue.
32:48
Or maybe revenue number is fluctuates too much.
32:51
We wanna create more stability in this," right? So think
32:54
from that and then say, "Okay.
32:56
What are the parts of our business w-where AI might help us to achieve
33:00
that goal that is gonna show up there on the income statement?"
33:02
So that's, that's really that piece of advice we give there.
33:05
It's like don't start from the use case, which is where ninety-nine
33:09
percent of companies start from.
33:10
They say, "Hey, where is somebody else using AI?" Or, "Look at my company and
33:15
tell me some places where I could use it." start from the income statement.
33:21
And an important aspect of that is the last point here, which again comes
33:25
up in digital transformation roadmap, where you said the most important
33:28
part of measurement is engaging in a thoughtful debate, first of all,
33:32
about what your metrics should be.
33:35
And as we know, you cannot measure the new the same way you measure the old.
33:39
The existing is not the same as the new, and this is point four,
33:42
is to decide how you'll measure financial return in the first place
33:47
Yeah.
33:47
A- and that's really, I would say, the most essential kind of lesson
33:52
of, of, of the whole article.
33:53
The, the most important kind of practical step is just before you start anything
33:58
new with AI, as we're all being, you know, many of us being told we've gotta,
34:04
you know, apply AI to this part of the business, your business unit, et cetera.
34:08
you start that project, that pilot, that experiment, define your
34:13
measures of financial return, right?
34:15
Define at the beginning, how are we gonna measure this is working,
34:18
and w- how are we gonna measure the ultimate financial payoff?
34:21
Doesn't have to be a quick win, right?
34:23
You may say, "If this works, it's gonna take some time, but within 12 months,
34:27
16 months, this is where it should show up in the financial numbers," right?
34:32
Now, you may have some intermediate metrics.
34:34
It's not… But you've really gotta start with that discipline of saying, what
34:38
will be the ultimate final financial test of, is this really creating,
34:44
again, what we call measurable return?
34:48
Fantastic.
34:48
And before we, we wrap it, David, I have a question here.
34:52
So, I, earlier on today before preparing for the show, I just, I
34:56
came up with this idea of a cartoon.
34:57
I'm gonna show it for people on the screen here who watch on YouTube.
35:00
But, I asked a question and I, I share, I said to people,
35:04
" pick a question for David.
35:06
I'm interviewing him." So I only did it a couple hours ago, so I didn't really
35:09
have much time, but I shared this image, and for those who just listened to us,
35:13
it's a bunch of computer scientists standing around a computer with huge
35:18
data center behind them powering that computer, and it's this idea that
35:23
they're created this beautiful AI for the company, and a senior executive
35:26
comes over and says, "Ask it to find a meeting slot that works for everybody."
35:31
And there's the thought bubble coming from one of the scientists, "Why did
35:35
I take this job?" Another one says, "I think I'm gonna puke." Another one says,
35:38
"I'm glad I have a PhD," sarcastically.
35:41
But then on the questions, a lady called Andrea Rubick, PhD, she's a fractional
35:48
CMO, strategic marketing advisor.
35:50
She asked a question for David.
35:52
So a question for Da- David.
35:53
I don't know Andrea, so thanks for your question, Andrea.
35:57
Are organizations focusing too much on AI adoption and not enough on org change?
36:03
We measure tool usage and efficiency gain, but rarely whether AI has fundamentally
36:07
changed how people make decisions, collaborate, and work and operate.
36:11
How should leaders measure that shift and its impact on business performance?
36:16
Question for you, David
36:17
Great.
36:17
So thank you, Andrea, for the question.
36:19
And, and yes, I agree.
36:21
As we said before, many folks are really just focusing too much
36:24
on sort of adoption and not what is the ultimate business impact.
36:28
And, and as Andrea points out, if we're gonna see a real impact, we're gonna
36:33
actually have to do things differently.
36:34
It's not gonna be simply applying, uh, you know, faster technologies to
36:40
do run our business in the exact same fashion it happened before, right?
36:44
And this is something I think a lot of folks have kind of come to
36:46
that realization in the last year.
36:49
But a lot kind of gets stuck.
36:50
Many people I talk to, they're like, "Oh, yes, we, we, we
36:52
need to reinvent our workflows.
36:55
We need to redesign work.
36:56
We can't apply new tech to the old way of working.
37:00
Um, how do I get started on that?" Right?
37:03
"How do I actually go about changing the way we're working?" And again, this
37:07
is where I would point back to that, that value equation, as we call it.
37:12
The value is gonna come from technology times adoption times strategic alignment.
37:16
So are you already have people focused hard on making sure the technology works.
37:22
And, and if your data is not in the right state, uh, today, then folks are
37:27
probably starting to try to prepare that.
37:30
has had a lot of attention, but where we have neglected, I really would stress
37:35
this word neglect, is strategic alignment.
37:37
And if you figure out, if you find the problems that really
37:41
matter to your business, right?
37:43
If you put your money where the value is in your business, that is
37:48
gonna start pointing you towards that real change in the way you do
37:52
that business, to Andrea's point.
37:54
You know, the strategy, I guess I would end by saying, you
37:58
know, the strategy is your moat.
38:01
you can link AI to your particular economics, that is where you get
38:06
real and even advantage over others.
38:09
If you're just applying the good technology to smart operations in the
38:16
same kind of ways, you know, you know, that your vendor suggests, just like all
38:20
of your peers are doing, at most you're gonna be sort of standing still, right?
38:24
That's kind of the best-case scenario.
38:26
So that is my advice at this moment, as companies are trying to figure out,
38:29
how do I go about actually starting to change the way my business works?
38:34
Start from really getting a deep understanding of what
38:36
your business is, right?
38:38
How are the economics of our business different?
38:40
How does our strategy differ from others who are even in the same industry as me?
38:46
And then really closely aligning your AI efforts to that.
38:50
And then I think you'll start to see, through actually building and applying and
38:55
integrating new solutions, you will start to redesign the way that the work is done
39:02
Beautiful, David.
39:02
Mic drop, man.
39:03
Mic drop.
39:04
So for people who wanna find you, I have two of your books there.
39:07
I see you have multiple languages behind you now, so congratulations
39:11
on the expansion of your work as well, and I'd love to come back and
39:15
do Is the Network Your Customer?
39:17
I do wanna do a series on networks and understanding networks that
39:21
includes David's book, The Network Is Your Customer, and indeed, this
39:24
book, Winning the Right Game, Ron Adner as well, which is a great book,
39:28
similar kind of concepts to David.
39:30
And we're gonna do a series on that coming in 2027.
39:34
Can't believe it's 2027 already, David.
39:37
It's coming up.
39:38
And, and I know you've a,
39:39
fast,
39:39
you've a new book in the pipes as well.
39:41
Uh, maybe we'll tell our audience
39:44
Yeah, so I mean, it's not out, but just that is what I'm working on
39:47
next, is a book that's going to be specifically thinking about, really
39:51
addressing how do we create value and how do we transform our businesses.
39:55
I mean, transformation, I've learned, has really, since I started writing on the
39:58
topic and then now so many others, I would argue, become a core, uh, leadership job.
40:05
I mean, you made the point of companies saying, "Oh, who's gonna run digital?
40:08
Who's gonna run this?" For a while it was like, "Who's gonna be the
40:11
transformation person?" And now we realize that's a discipline for every leader.
40:15
Every lead- it doesn't matter if you're the head of procurement, part of your
40:19
job is finding talented people, right?
40:22
Part of your job is defining strategy for what are we focused on in this,
40:26
in this function, and part of it is leading an ongoing transformation
40:30
of what does, know, procurement look like or marketing or whatever
40:34
part of the company that you're in.
40:36
So that's really kind of the focus of my work and, and really the next book
40:40
is about how do we think about these challenges in the context of the AI era.
40:46
So it's not so much that it's not really gonna be a technology book, but it's
40:49
very much is how is the current era and what we're dealing with in these
40:53
new capabilities and c- really the accelerating pace of change that we
40:57
are seeing across so many industries., How do we make sense of that?
41:01
How do we stay on top of that?
41:02
How do we lead through that so that, again, so that AI is not just something
41:07
we do, so we can check a box and say, "Look, we've got a bunch of AI
41:10
projects," but we can really know why and where we're applying it and see and
41:17
measure the impact that it's having.
41:19
And speaking of finding talented people, where can people find you and your work?
41:23
I know you do keynotes, you do workshops, you work as an advisor all over the world
41:28
Sure.
41:28
I'm easy to find on, LinkedIn, David Rogers, you know, probably
41:33
just David Rogers, certainly if you search digital or transformation.
41:36
Davidrogers.digital is my website, so that's an easy place to find me.
41:40
And also my Substack is davidrogersdigital.substack.com.
41:45
You can find there this latest article on the AI value gap, call it,
41:49
the title there is The AI Value Gap Is a Strategy Problem, as well as a link from
41:54
there to the HBR, the original, article it published in Harvard Business Review.
41:58
And to the point of the book that's in development, there's gonna be
42:01
more articles like this coming as I'm working on the book, and the research.
42:05
So many interesting case studies and executives I'm, meeting with on a almost
42:10
weekly basis, and just sharing some of those stories and what I'm seeing
42:13
about really what does work in this fascinating time that we're in today.
42:19
It's always a pleasure.
42:20
Author of The Digital Transformation Playbook, The Digital Transformation
42:24
Roadmap, The Network Is Your Customer, and this new book
42:27
coming down the pipeline as well.
42:29
David Rogers, thank you for joining us
42:32
Thanks so much for having me, Aidan.
42:33
Always a pleasure
42:34
As always, thank you to our sponsor Kyndryl, who runs and reimagines the
42:38
technology systems that drive advantage for the world's leading businesses.
42:42
With a unique blend of AI-powered consulting built on unmatched managed
42:47
service capability, Kyndryl helps leaders harness the power of technology for
42:51
smarter decisions, faster innovation, and a lasting competitive edge.
42:54
You can find out more about Kyndryl at KYNDRYL, kyndryl.com