From Learning and Adaptation to Intelligence (248)

This week’s podcast is about learning and adaptation as a core capability. Which is becoming much more important with AI.

You can listen to this podcast here, which has the slides and graphics mentioned. Also available at iTunes and Google Podcasts.

Here is the link to the TechMoat Consulting.

Here is the link to our Tech Tours.

6 Levels of Rate of Learning and Adaptation.

• Tactics: Such as marketing, inventory and assortment, store layout, pricing, churn
• Customer retention, customer segmentation and other more complicated questions.
• Adapting current products and services. Such as personalization. 
• New products and services. And technologies.
• Internal processes.
• Business models.

4 Rate of Learning and Adaptation Capabilities:

• Signal Capability
• Experimentation Capability
• Organization Capability
• System Capability

--------

I write, speak and consult about how to win (and not lose) in digital strategy and transformation.

I am the founder of TechMoat Consulting, a boutique consulting firm that helps retailers, brands, and technology companies exploit digital change to grow faster, innovate better and build digital moats. Get in touch here.

My book series Moats and Marathons is one-of-a-kind framework for building and measuring competitive advantages in digital businesses.

Note: This content (articles, podcasts, website info) is not investment advice. The information and opinions from me and any guests may be incorrect. The numbers and information may be wrong. The views expressed may no longer be relevant or accurate. Investing is risky. Do your own research.

Transcript

00:05

Welcome, welcome everybody.  My name is Jeff Towson and this is the Techmoat Strategy Podcast from Tecmo Consulting.  And the topic for today from learning and adaptation to intelligence and modes.  And this is kind of a follow-up to some pretty, let’s call it wordy theory-ish, articles I sent out this week. Two of them on the idea of rate of learning and adaptation as a concept.

00:35

which tees up the idea of businesses with native intelligence, which is kind of like the huge thing happening right now. And yeah, they were a bit much. I'm still trying to get my thinking together on this, so was kind of a lot of theory put together. I think looking back, it'll probably be about 50%, 40 % on target, and then a lot of 60. It's like, yeah, that was wrong.

00:58

So anyways, that's where I'm at. I thought I'd follow up to that, because it was kind of a lot of theory to send out in two articles and sort of distill it down into something more useful.  A little bit clearer. Anyways, that's going to be the topic for today, sort of rate of learning and adaptation as a concept, which is really on the frontier of an entire new wave of business models, operating activities. It's kind of a huge, huge thing.  Anyways, so that's what I'll talk about today.  Let's see, housekeeping stuff.  have...

01:26

Two China tours coming up, one in August, one in October.  The first one of which I've mentioned before is the Urban Tech Tour, which is going to be looking at sort of urbanization issues, smart cities, connected cities, green cities, financing for infrastructure, things like that. So that's kind of an interesting one, and that's me and Lola Wetzel, my partner in many books and courses.

01:52

And then the other one, will be October, which will be for brands, merchants, really anyone who sells anything online.  Basically, looking at the frontier of sort of China e-commerce, media, getting attention, all the new tricks and business models coming.  Because e-commerce in China really is quite a few years ahead of pretty much everywhere else.  So, there's a lot of good stuff you can copy.  So that'll be the point of that one. That'll be cool.

02:17

If you're interested in those, just go over to techmoatconsulting.com. You'll see the details, including pricing and such, on the tours page there, easy to find, or just send us a note.  The email address is on the website too. Okay, standard disclaimer, nothing in this podcast or my writing or website is investment advice. The numbers and information for me and any guests may be incorrect.  The views and opinions expressed may no longer be relevant or accurate. Overall, investing is risky. This is not investment, legal or tax advice. Do your own research.

02:47

And with that, let's get into the topic.  Now the concept for today is obviously rate of learning and adaptation.  It's in a couple places on the concept library.  You can look under digital operating basics and it pretty much goes under number two and number three, which is constant customer improvements and the digital core and number four as well, interconnected business models.  But it's mostly under...

03:13

the digital marathon, you know, I talk about five digital marathons, areas where if you focus on the activity long term, like a marathon, you can actually create an operating advantage that is relatively stable.  So, you can start to see an advantage, not come sort of out of structural aspects, like modes, but more out of an operating activity. Basically, if you're running a marathon and you pull ahead of the pack and you keep pulling ahead of the pack,

03:42

At a certain point, you're so far ahead of the pack of other runners, it's effectively an advantage based on operating activity that has accumulated over the long term, and I detailed five of them.  The little acronym is SMILE, S-M-I-L-E for the five.  Well, the L stands for rate of learning and adaptation. Plus, the idea of intelligence is sort of coming into that, which I'll talk about.  So that's where most of the thinking is, but it's actually...

04:09

If you've read my Motes and Marathons books, it's almost at every level of strategy. You start talking about rate of learning and adaptation. It's kind of a big deal.  So, it's in there a lot. I'm still trying to get my thinking around it really.  But, you know, I guess let me tee up why I'm thinking about this so much. There's a couple subjects in the world that I'm sort of struggling with that I'm trying to take apart. This is one of three of them, basically.

04:40

And I spent five, six years, really seven years building out sort of what is a good digital strategy. How do you combine Warren Buffett like classical, here’s a company that’s dominant with constantly changing digital tools, technologies and business models? So, I spent a lot of time putting those two questions together, came out with a whole framework, which is what I've written the books about. The short version is.

05:08

You build a moat and you choose your marathon, motes and marathons. That's kind of a strategy, okay. But that's for digital. And when we're talking about digital, I mean, in theory, that's everything.  Really what we're talking about is we're talking about software, we're talking about data, and to some degree, some hardware. Okay, but historically, we're not talking about generative AI. We're not talking about agentic AI.

05:36

We're not talking about robots that think for themselves, sort of embodied intelligence.  So, there's a whole new thing coming.  And I've basically been setting out to build a similar strategy for AI. So instead of digital strategy, AI strategy.  Now, what does that mean, strategy? It's kind of a meaningless, well, not meaningless. It's not a helpful term.  Most CEOs like...

06:02

Okay, you have a strategy, what does that mean? The way I think about it, I'm almost should stop using the word strategy. It's really, I focus on two things. Number one is what CEO senior management level problem are we trying to solve?  Usually that’s something like our growth is slowing. Our rivals are eating our market share. Our margins are shrinking. Business is getting harder. Business is getting stronger. Okay.

06:32

That's all part of a strategy. If you outline a strategy, if you understand what's going on, you can start to solve problems.  And that's really what you should do as an advisor is, okay, let's not get into theory here. Tell me what your biggest problem is this year. Let's try and crack that, which is, you know, the term is consultant.  That's also what you call doctors in a lot of the world. They're consultants. Same thing. Oh, my knee hurts. Okay, let's talk about that.  Okay, that's sort of one.  Tell me what your biggest problem is this year.

07:01

or the one you're worried about the most.  The other way to think about it, which is why I spend a lot of time doing, is let’s talk about what winning looks like in a year to two years.  Usually that's kind of fuzzy in most CEOs. They kind of know sometimes.  Sometimes they don't.  But crystallizing that really helps you sort of chart the path.  So that's a lot of what strategies. Let's talk about what winning looks like. Also, let's talk about what losing looks like.

07:29

You do not want to be looking like this in two years because the world's going to get a lot harder. That's really what I mean by strategy in practice, but I don't have a better word for that. If you have one, let me know.  Okay, when we talk about AI strategy, I wrote 10 articles on this at the end of last year, which is basically me trying to figure out the strategy. And I've got one more pending and then I’ll boil it down to something short and readable.  But, you know,

07:59

I spend a lot of time looking at companies like Lazada, Alibaba, CBG companies, and I basically look at what they're doing with AI tools. Show me your use cases. That's my standard question. What are the use cases that you're scaling up? And from that, you can kind of get some sort of AI strategy. This is what people are doing in practice right now.  Not theory, scaled use cases.  And that's really kind of three things.

08:27

right now, which is number one, everyone's putting AI into their own internal operations and they're trying to increase their productivity and quality. So, let's do what we're already doing. Let's just do it cheaper, faster, and in many cases better. And it turns out generative AI is really good at that. Everybody's doing that.  It's an easy first step and there's not a lot of risk because it doesn't hit your customers directly.  Then step two.

08:56

Let's start to put generative AI into our products and services. Number one, because we have to cover our flank.  We can't have one of these new apps come up with a solution that makes us obsolete.  Number two, maybe we get lucky and we come up with a 10x product that is dramatically better. But everyone's putting it into their products and services gently, because you don't want to wreck the customer experience and make them upset.

09:24

But yeah, everyone, you got to do that just because it's a point of risk. And then number three, which is a question I'm trying to solve is, what does a moat look like based on Gen. AI products, services, tools, technologies? Are we going to see new types of business models? Are we going to see new types of motes? What is that? We know what a digital business model looks like. A platform, standard, direct e-commerce. We kind of know what they are and we kind of know.

09:53

what the motes look like. We don't really know that for generative AI.  It's a free for all.  So, I'm working on that one.  And my little process for doing this has basically been meeting with lots of companies, figuring out their use cases, and then trying to piece it together. Now, the pathway, which I've heard people describe, which I agree with, which is, okay, we had predictive AI.

10:22

Now we're in the phase of generative AI.  What comes next is agentic AI agents. I think that's this year.  And then after that, we get to real world AI, which is embodied intelligence in robots and you know, Tesla's arguably the frontier there with their, you know, their, their robots that can walk around and learn things on their own, like watching three-year-olds. It's pretty spooky actually.  And they're really good at it. Like the latest versions.

10:52

they can learn to do things like dance just by watching television and then they can get up and do the dance apparently. I mean that's the press reports but yeah that's a little spooky. Okay so that's what I'm trying to do.  Now we'll get back to the point.  Within that there's three sort of questions ideas concepts I’m trying to figure out.  Number one which is what we're talking about today which is rate of learning and adaptation.

11:22

Now this has been around for a hundred years as an idea that businesses can learn, teams of people can learn, individuals can learn, and it’s been a big deal forever.  And it's just gotten a lot more important because data is getting richer.  We know things that we didn't know before.  It turns out AI is pretty good at this. And now we're getting this idea of sort of business native intelligence.

11:49

where if you go to a big conference by a cloud provider, let's say Huawei, the word intelligent is every third word.  We're building intelligent operations, we're building intelligent chatbots, we're doing intelligent R &D, it’s everywhere. And one extreme vision for that would be like, okay, we've got a 200-person company, but only 50 of the people are human, the others are all just AI that do stuff, agents.

12:19

Well, that's a form of intelligence. I how do you get intelligence into a business?  The three cases I just gave, a person, a team of people, or a business, the intelligence is all in people's brains. It's all people-based intelligence. Well now we're talking about non-human-based intelligence. That's kind of a big idea.  Also, a little bit spooky. Okay so that's concept number one. I'll talk about that today.  Number two, which is

12:48

this idea of we’re going to start to see very different cost structures in businesses than we've seen before.  Most businesses, we talk about the gross margin, we talk about some technology, we talk about some operations, we look at the gross margin, very, very important. Okay, this is going to be different. We're going to get rid of a lot of people, or the people we have are going to be dramatically more productive. Well, they are usually in the gross margin.

13:19

That's usually a variable cost. And we're going to replace them with basically AI. So, the technology spend is going to go up. In traditional software, we'd say, oh, that's good because our fixed costs will go up, but our variable costs will go down.  That's classic software economics. Well, that's not really the case with AI.  It turns out AI has a lot of variable costs.  It's very different to say,

13:49

You know, digital business means we create a piece of software once it’s done, we print it on disks and we ship it out, and it's just a fixed cost. But it does the same thing over and over. It's like a recipe.  Well, AI is not like that. AI is more like humans. If you ask AI the same question five times, you'll get five different answers, like humans would give you.  Software doesn't do that. It gives you the same answer every time.

14:18

And it turns out it doesn't scale up that easily. In fact, when you try and scale it up, what you tend to get is a lot longer tail use case. And those can be expensive to answer. It turns out it’s relatively cheap to answer the common questions.  It's relatively expensive to answer the rare use cases or questions. And as you scale up, you get more and more of the long tail and the cost can actually deteriorate.

14:46

and the economics get worse and worse and worse the bigger you get. it also turns out it’s not one and done, where we've created the software, we print it, we ship it on disks, whatever, we update it every now and then.  You have to continually retrain. mean, it's an ongoing, know, AI looks, when you look at the economics, as far as I can tell, it looks like 50 % human-like economics and 50 % classical software.  But.

15:14

that doesn't mean you're going to have the same types of motes. when we, know, looking at this, I'm trying to get my brain around the cost structures for what these new business models are going to look like.  How much are we spending on the cloud? How much are we spending on infrastructure? How much are we spending on AI agents?  How much on generative AI? How much on humans? We're looking at a very different cost structure in a business. So, to get my sort of...

15:42

get into that I've been meeting with the big cloud providers literally as often as I can. In the last two months, two and a half months, I've met with Huawei cloud, Alibaba cloud, Baidu cloud, Tencent cloud, and I'm reading everything I can to get out of Silicon Valley, these new apps to try and get a handle on their cost structure basically. So that's bucket number two that I'm trying to get at. And then number three is

16:12

the moats and the competitive advantages, which 50 % of the time the moat follows from the cost structure. So, if you don't know the cost structure of these new business models, you can't really figure out the moats.  Economies of scale, which is a common competitive advantage, you know, that's all about the costs. That certain companies have a cost advantage over others because they're scaled. Okay, if we don't know the cost structure for AI and how this is going to work.

16:42

hard to get at the moats. Anyways, those three buckets are what I've been kind of digging into the last six months really.  I feel like I'm getting closer.  It's a little difficult.  The problem I'm having, honestly, is when you try and do digital strategy, the data is quite available because you just look at public filings for companies.  People have been digitizing their operations for 20 years. So, there's a lot of data in public filings.

17:11

We don't have that for generative AI, not really.  And the economics that you see in public filings related to generative AI, it’s so early stage, who knows if it's going to last. So yeah, you kind of have to go to use cases. And I've been going to cloud providers, because I figure they know what businesses are actually building at scale right now.  That's the theory.  Okay, that's a little bit of teeing up.

17:39

why I'm thinking about this. Today's going to be a little bit lighter than normal, I suppose.  Okay, now let's get into rate of learning.

17:50

If you go onto my website, you can find rate of learning everywhere. It's in the books. I think the case I talk most about is sort of traditional rate of learning, like Henry Ford's, know, cars got cheaper and cheaper the more cumulative volume they produced.  Aircraft assembly got cheaper and cheaper the more aircraft you assembled, which was a human-based activity, the more efficient, they got faster and cheaper.

18:17

And in human labor, those things are the same thing because if a human spends less time doing something, it's cheaper.  BCG writes a lot about this subject. They've been writing about it since the 1970s.  They talk about learning curves. They talk about what they call the experience effect.  Pretty interesting stuff. A lot of it didn't really last, didn't sort of stand the test of time. A lot of those ideas about why learning is an advantage turned out.

18:46

In a lot of the cases, it wasn't.  The other type I talked about was sort of Steve Jobs based learning, which is he wasn't great at making things cheaper over time with cumulative volume, that would be type one.  He was better at innovating and creating the next version of something and jumping to that.  So, the first iPod led to the second iPod, to the third iPod, to the iPhone. So that's a type of learning.

19:16

more based on changing the product as opposed to doing the same product over and over. I would argue Tim Cook is very good at type one learning. He's making things more efficient, definitely more profitable, but he doesn't seem to be able to launch any new products, or at least I haven't seen any in 15 years. And then type three would be algorithmic learning, where an organization, a business can learn.

19:45

instantaneously in seconds, which humans can't do.  And you can feed that into things like, let's have all the prices in our supermarket update every hour based on what people are buying, based on what our competitors are doing, based on what's online.  And it'll just be little LED price tags on everything and it'll automatically update. Okay, humans can't operate at that speed. So, we call that algorithmic learning.  Okay, so I talked about those in the book.

20:14

But you really have to sort of take one step back on this and say, okay, it’s learning plus adaptation. An organization learning something doesn't make any difference unless you adapt it, right? Unless you adapt and respond to it.  So, you need adaptation, but okay, adapting what? And it turns out that can be a whole lot of different stuff and with different implications. So, the simplest level, we could just talk about tactics.

20:43

marketing, promotions, changing the inventory in a 7-Eleven two to three times a week, changing maybe the store out, changing pricing, which I talked about. Okay, those are all tactical moves where we would learn and we would adapt. Fine, I would put that into sort of daily or weekly activities. We could also think about how we view our customers, how we gather, customer segmentation.

21:13

things like that. We could move at one level up. Okay, let's talk about something a little more core. Let's talk about, let's say, customer retention. Okay, if we look in our numbers and we see that we have a lot of churns, we're losing customers, that's very, very bad.  How do we understand about, how do we figure that out?  Well, that's learning and adaptation.  But...

21:38

That type of question is very, very different than let's just change the pricing on the items because there's a promotion at the market across the street. know, customer retention strategy that falls out of this, well, that's going to require a lot of data, a lot of feedback, but it's also going to require a lot of human expertise. And it's probably going to require a lot of experimentation. So, okay, those are both rate of learning and adaptation, but, you know, those are very different things.

22:07

from a management perspective.  We could go one level above that. What about personalizing our products and services? You know, go on Netflix, you see different shows than I go on Netflix. Well, that is rate of learning and adaptation, but it's a lot more core to the business when we start changing the product. Now in this case, we’re not completely changing the product. We're just customizing it and personalizing it.

22:37

But yeah, that's a big deal.  Okay, we can go one level beyond that. What about adding new products and services? Well, that’s also rate of learning and adaptation. We could see variations on current products.  here's, what was it? Coca-Cola Oreos? Coca-Cola and Pepsi, I think it was Pepsi.  They did some co-branding thing a couple months ago where they had...

23:06

I think it was Pepsi Oreos and they had Pepsi soda that was Oreo flavored.  Turns out the cookies were really, really good.  They were shockingly good.  Anyways, we can do variations on existing products or we can launch entirely new products. Well, okay, now we're in product development. That's a big deal. But again, rate of learning, adaptation, the process is similar. lot of experimentation there too.

23:33

We could change internal processes to the operation. We could change how our factory works.  We could change how our call center works.  We could learn and adapt our internal processes. And if you want to go to the biggest level, we could change our business model.  Netflix, they used to send CDs. And then one day, well, they didn't say one day, but they went to streaming.

24:02

And then they did this move where they were going to try and do streaming with advertising and not, and people hated it and they abandoned it almost immediately.  All of that is rate of learning and adaptation. So, when you talk about this subject, it’s very easy to talk about the simplest cases of like, we're just changing the pricing in real time, or we're customizing your Facebook newsfeed. No, no, it’s a process.

24:30

for change across the entire enterprise. Tactics, operations, products and services, business model, the whole thing.  It depends how daring you are. And now keep in mind, we’re turning that into native intelligence. Well, how smart do you want this organization to be? How adaptable do you want it to be?  Are we going to be a business that's kind of like an amoeba that just sort of feels its way?

25:00

through the ecosystem and wherever it finds food it just sort of shifts that direction? Or is this more of a classical hierarchical structure? I use a lot of animal analogies for that like a lion. look, a lion is not going to stop being a lion.  It may hunt differently and do things differently, but it's still a lion. Okay, those are two different types of organizations. One is super adaptable and let’s say like the amoeba.

25:28

And that can be a real strength.  Other could be very hierarchical and those tend to be very, very efficient and specialized. When you go for adaptation, unfortunately, you give up a lot of efficiency.  Hierarchical structures are optimized for efficiency and specialization. Well, the more you become like an amoeba, you lose some of that, which can be a competitive weakness.  So, most organizations that are

25:57

you know, digital, really good digital creatures. They try to get both. They try to be very efficient and hierarchical and specialized in certain activities and then very adaptable and amoeba-like in others. And they try and do that with agile teams and other things. Now think about sort of an agentic company or a real-world AI company. Let's say you have a company with 5,000 robo-taxis.

26:27

and they just cruise the streets on their own.  And they're very Amoeba-like because they have the ability to learn and adapt to whatever people need.  And you're sitting back in the head office staring at the screen, watching these robo-taxis just sort of go around the city and do whatever they think is the best thing to be doing. I mean, it kind of looks like an Amoeba. Now it’s still giving rides to people. It's not changing its business model.

26:56

But yeah, you have sort of unleashed this thing out into the ecosystem that's exploring and trying to find the most profitable things to do. Okay, so I think that's enough of theory. Let's talk about some sort of things you can actually do as a business.  And this pretty much comes from Martin Reeves over at BCG who is head of their Henderson Institute, which is kind of their publishing research wing.  He wrote a book.

27:24

with the co-author, I forget who called it, Adaptive Advantage, Winning Strategies for Uncertain Times, not an awesome title really. He's kind of the leading thinker on rate of learning and things like this. Very, very good. And in there, I probably agree with about half of it. It's very interesting thinking, a lot of client work, so there's a lot of reality in the theory.

27:53

But the part I thought was particularly useful was, okay, rate of learning and adaptation.  This is an operating activity. It can be tactical, it can be a digital operating basic, it might be a digital marathon, it can be in lots of ways. But let's talk about the capability you would need to build that. And he lays out four capabilities. Actually, there's five, but one of them I didn't really buy, so I’m talking about four.  Four capabilities.

28:22

to build this into an organization. Okay, that's useful. And they are basically signal capability, experimentation capability, organizational capability, and systems capability. And they lay out steps you can do for each of those.  And yeah, it's pretty useful. So first one is signal capability.  I think their language on this is pretty good.

28:46

the ability to capture, interpret, and act upon signals from an increasingly data-rich environment. Okay, I mean that's just business intelligence, but the point about, you know, data-rich, yeah, the amount of data in the world is absolutely going through the roof. All types of data, video, images, all of it.  And the shelf life of most data is not very long.  Plus, it's often in

29:16

not particularly usable forms.  One of the things about data is everyone talks about like, AI needs a lot of data. Yeah, the problem with that is quality versus quantity. If you scale up the quantity of data, the quality goes down the drain. I mean, it's really, you know, there's so much bad data out there that the bigger a net you throw, the more garbage you're going to get, almost guaranteed.  So, this whole idea of scaling it up, weeding out the bad stuff, yeah.

29:45

The word they use which I like is, look, it's all about getting the signal from the noise. yeah, it’s not really about possessing the data.  That's nice.  Although proprietary data does have some power and a lot of businesses are focusing on that.  But no, it's about acquiring it, analyzing it, and then using it. Data that isn't used isn't really worth anything. Okay, that’s a good one.

30:13

The steps they list for this, won't go through it because it's pretty obvious. Capture the relevant data, hunt for patterns in the data, make operational changes in real time. one part that's interesting in there is you can start to shape the information ecosystem. One of the things you discover when you start gathering data is you realize, oh, we don't have enough internal data to do this.

30:41

because almost nobody does.  So, you start looking for external sources of data, could be public, could be purchased, but more often than not, you're starting to look at your supply chain partners and you're starting to look at sort of your consumer facing partners, complementors, things like that. And you can start to do data ecosystem agreements. You can do data sharing agreements.  That is a really good idea for most all companies.

31:11

The second capability, experimentation capability, this is actually the one I like.  And this is one where I think when I think, okay, everyone's doing signal, everyone's learning and feedback and fine, but where can you sort of do a marathon where you can build an advantage, an operating one or maybe even a structural one?  Experimentation is interesting in that regard because most companies have a very difficult time running experiments. Maybe they don't have enough customers to do it.

31:40

maybe they can only do a couple here and there. Like Facebook can run experiments every day, all day long on its customer base for every little change they want to make. And they do make tens of thousands of experiments every year. Alibaba can do the same.  Experimentation is an area I think about a lot in terms of building moats and advantages.  Now one, so that part's interesting. Two,

32:09

You know, most of the things are not that predictable. If the world's changing this fast, know, looking at the data and doing analysis is important. It's worth spending a certain percentage of your time just throwing stuff against the wall and seeing what sticks. You know, things are changing too fast and so much of it is not predictable. Certain products just take off and nobody saw them coming. So yeah, anytime you do analysis, you look at the signal,

32:39

You get the smart people together in a room and you make your best bets.  These are what we think are going to work. Then you go test those. Fine, experiment.  But then a certain percentage of time, just do stuff and see what happens. Because a surprising percentage is not predictable.  Now, the process for that, which BCG, this book they talk about, okay, step one, generate ideas. That sounds simple.  Until you realize, wait a minute.

33:09

Generative AI is really good at generating ideas. It's very, very good at this. And you don't just want to generate ideas because we're starting to think about advantage.  Can we generate ideas faster and cheaper than our rivals? That would be interesting. That's an interesting thing to measure.  Okay, step two, we got to test the ideas. Fine.  How can you test your ideas faster and cheaper than your rivals? Interesting.  And then there's another part.

33:39

How do we test our ideas with a lower cost of failure than our rivals? One of the reasons companies don't like to do too many experiments is because it can annoy the customers. You don't want to roll out a hundred mediocre products to even a small subset of customers all the time.  They're going to get annoyed.  There's a cost of failure. If you have an ability to do testing of your ideas faster, cheaper, and with less risk,

34:07

that's an interesting capability to have. And I think that starts to become an advantage. Alibaba can definitely do this.  If they want to test out new products or services, let’s say products, new Nestle types of coffee, and they've worked with Nestle on this in China, they will come up with the ideas with them, and then they will test them on the usually Taobao or Tmall, probably Tmall in this case.

34:33

but they won't call it Nestle, but they'll test and see the reaction and see the consumer behavior. So, they can test things very quickly, very cheaply, and with little to no risk to the brand. That's really interesting. Okay, step three then, we've got some winners, a lot of losers, a couple winners, how do we scale them up? Now, if you have a captive audience, if you have a captive customer base, if you can roll it to them through cross-selling or bundling,

35:02

That's very, very powerful. And again, Alibaba is sort of the king in this. They can start to cross sell any new product they want or bundle any new product they want with their existing products.  And they get huge adoption in pretty much everything they do.  Not 500 million, but they can get 50 million customers for any new service they try, pretty much.  Okay, so that's number two, but experimentation as a capability is really interesting.

35:32

and it also depends on culture. A lot of more traditional companies, you don't get promoted up the ranks as an executive if a lot of your stuff fails.  Even though they might say, hey, we encourage experimentation, you rack up a pretty good list of medium bets, small bets that fail, you tend not to get promoted. So, there's a lot of cultural work here, which is like, look,

35:59

You have to accept that experimentation and failure is the norm for a lot of things we do. Maybe not new products, but new marketing tactics, new promotion types, new operating activities.  A lot of companies, that's a problem.  And even though they will say it's fine, fine, fine, yeah, then you look who gets promoted.  It can be a little bit of a minefield for executives trying to rise up the ranks.  All right, last one. Well, yeah, last one. Organization capability.

36:30

This is what I talked about earlier, this idea that as an organization, are you an amoeba? Are you more hierarchical? I don't really have a great analogy for that one. I guess a lion.  Are you more optimized for adaptability? Are you more optimized for efficiency and specialization? Because there is a tradeoff there.  And, you know, as I said, certain companies try to sort of do both. have agile teams.  They build modular organizations.

36:57

They talk less about strict operating procedures and more about values and they push decision making down.  Yeah, that's a thing. You usually have to have discussions with management about that. And when you start to build that into your organization, you’re really talking about changing internal workflows and routines. That's how a strict hierarchical organization becomes adaptable is you change the workflows.

37:27

Otherwise, it's kind of talk and presentations.  right, systems capability, which I won't talk about much. Systems capability is kind of what I alluded to earlier, which is you go beyond being a single organization doing any of this.  When there are big moments of technological change, like autonomous vehicles, electric vehicles, we’re talking about wholesale change in an entire industry.

37:57

where the future product is not clear and often the technology is not fully developed.  When you have these major change points in an industry, usually what happens is an ecosystem starts to emerge where a large, a handful of large companies start to work together and basically do the amoeba play, which is look, we don't know what's going to work, so we're all going to innovate together, we're going to pool our R &D, we're all going to join tech associations.

38:27

and we're going to work on defining the new 5G standard and so on.  That's usually what happens is you start to see an ecosystem emerge which tends to be better at innovation in the face of big uncertainty.  And again, the cost of doing that is efficiency.  So that tends to happen now. At the company level, what tends to happen is what I just said is you tend to build a bit of a consumption ecosystem.

38:54

or a supply chain ecosystem where you work with parties and you start to share data.  That makes your sort of learning and adaptation better.  And from sharing, coordinating on data, you can start to share and coordinate on things like research.  You can start to share and coordinate on production.  yeah, that tends to be good when you're you know, high uncertainty, rapid change in tech and or customers.

39:24

When your products are particularly complicated, you kind of have to work with the other parties around you.  Anyways, in my frameworks, that's all under DOB4, Digital Operating Basics 4, which is the idea of an interoperable business model where you work with partners.  Okay, and that's most of what I wanted to sort of cover.  It's a big fuzzy subject in my brain right now, but I feel like I'm getting closer.  And as machine learning gets better and better, as agents get better and better, this whole idea of

39:55

you know, this organization can learn faster and adapt faster than another is going to become huge. It's going to be a major capability to compete on. So, we're just sort of in the early days. That's where I am. I hope that's helpful and maybe not too much mind-numbing theory, which I like, but I accept that some people they're more interested in talking about companies. And that is it for this week. I don't really have any news. I've just been working away here.

40:24

Yeah, pretty good.  Netflix. I mentioned this show before called The Devil's Plan, which is a Netflix TV show out of Korea, which is basically like a version of Survivor, except everybody's really smart.  And instead of like building tents and stuff on an island, they all have to compete, like doing mathematics and poker type games.  And it’s really impressive how smart they are and how they all kind of work together in alliances.

40:54

and there's usually one or two people who are markedly smarter than everyone else and the alliances form around the biggest brains in the room.  It's really quite good. Anyways, they did season two, which I think I kind of said wasn't too awesome, because I started watching it. Anyways, I finished it. It was really good.  It was really quite good.  And kind of what I like about this show is when they get to the end and it goes down to just two people competing.

41:21

It tends to be the people that are markedly smarter than everybody else.  And by smarter, I mean like they put them in front of a screen and you have to do sort of rapid mathematics where they just stare at screens and, you know, start doing calculations really quickly.  And you’re watching it like this is really impressive.  Like there's no way around it. These people, yeah, they got some, they got some horsepower there. Okay.

41:50

So anyways, turns out season two is pretty great too. Season one, season two, both great. Devil's Plan, Netflix, it's a Korean joke. Anyways, that's it for me. Take care and I'll talk to you next week. Bye bye.