Facilitating the Spread of Knowledge and Innovation in Professional Software Development

Write for InfoQ


Choose your language

InfoQ Homepage Presentations Variety: The Secret of Scale

Variety: The Secret of Scale



Cat Swetel provides an approach for incurring variety where it makes sense within the coherence of a longer-term vision.


Cat Swetel specializes in using Lean techniques to increase the agility and adaptability of large technology organizations. She has worked with several Fortune 100 companies in highly regulated industries such as financial services and healthcare. She has been invited to speak about her pragmatic approach to enabling technology leaders to make better, more timely, data-informed decisions.

About the conference

Software is changing the world. QCon empowers software development by facilitating the spread of knowledge and innovation in the developer community. A practitioner-driven conference, QCon is designed for technical team leads, architects, engineering directors, and project managers who influence innovation in their teams.


Swetel: Some disclaimers first, the first one, are you familiar with the works of Don Reinertsen? I really like his works, I've read his books. I've been friends with him for a long time. He told me maybe six or seven years ago that people remember hideous things, things that are very ugly. He said that's why his slides look the way that they do so I said, "Cool. I'll do that also."

Then earlier, late last year, he told me that he just made that up in the moment as an excuse for his hideous slides that day specifically. That's what I've based my ascetic philosophy on, a lie, but I'm just rolling with it still, so that's going to happen.

Then the other thing, I think I've been billed as a super agile person. I'm not actually going to talk about agile today, so if that disappoints you, sorry.

What Is Scale?

I'm here to talk about variety and what that has to do with scale. First, what is scale? What do we mean when we say scale?

Participant 1: Lots of people.

Swetel: Lots of people, yes. That's common. Any others?

Participant 2: Lots of servers.

Participant 3: Lots of customers.

Swetel: Yes, so two different sides there.

Participant 4: The ability to cope with the demand of all those things growing.

Swetel: That's a very interesting one.

Participant 5: Bureaucracy.

Swetel: Yes, it often comes with a certain amount of bureaucracy. I like the definitions in this book. How things change with size or how a system responds when its size changes. This is interesting, when size changes so not necessarily always getting larger but when size changes, sometimes things expand and contract and I think maybe we're not always great at the contract bit but what I like about this book is that it talks about relative scale, so like scale of a map or something like that.

How do things change in size and something that is the small version of a thing and something that is the large version of a thing, what do they have to do with each other? What do they have in common? How are they different? Everyone know the Santa Fe Institute? I feel like if you're a nerd right now that's the cool thing to mention, I read a book of one of the authors at the Santa Fe Institute but that's where that book comes from. The author is there and it's pretty great. Why are we trying to scale?

Participant 6: Break even. Swetel: Break even. All right. Yes, that's true though, when you're a startup in your interest line, you have some funding and you're like, "It'd be cool if we could break even and not lose money."

Participant 7: Some things are more efficient at larger scale.

Swetel: Yes. Any others?

Participant 8: Describing a moment we think signifies success.

Swetel: Getting bigger signifies success, yes. Is that a fair way to sum that up? Yes. What's interesting there is a few things came out about what we're trying to optimize, in that there's different optimizations that come with scale. One of the things that's right, when we're small we optimize around an idea. Usually, we have ideas in sequence and not multiple ideas at one time. When we get larger, we can have an ecosystem that supports the testing of multiple ideas or the emergence of many ideas simultaneously. We can create this ecosystem that can meet the demands of multiple customers and different types of customers. Let's really quickly run through what that ecosystem might look like, hopefully, not as hideous as this but it's a mere approximation.


The first thing, unit cost, we talked about this. You are trying to optimize for the unit cost, that incremental cost of one thing. Then we have this engagement, we may have some servers running somewhere or something like that. We try to make that repeatable so that on top of that we can build these platforms. I bet like 90% of the room is involved in something like this right now. Then on top of that, we have the opportunities for differentiation, this is where we introduce a new feature or we make an app or we do an MBP or something like that.

There are interesting things here, this bottom part is that unit costs, if you go to business school you will learn about economies of scale, which is where you make a lot of something and it becomes cheaper to make that thing. That's supply-side economy of scale, there is also demand-side economy of scale, which is like Uber. We're in London so I'm not sure if you're allowed to like Uber here or hate Uber, I'm not sure but anyway. This is where we are trying to create lots of options. The whole thing is better the more people join, the more drivers join, the more riders have an opportunity to get a driver. That is the demand side economy of scale and that's what we see in two-sided markets. Scale could be either of those things or potentially both of those things.

What's interesting is that if you really want to enable scale, you want to enable growth, you want to get bigger, then we need some things to be low variety so that we can have high variety in other things. If you name each one of your servers and you care for them like a sweet little baby, then that's differentiation that you're incurring at the very bottom of that stack. It makes them more costly for you to incur variation at the top of the stack where it's meaningful to a customer.

We have this high variety, which is where we try to meet the needs of a diverse market and it's enabled by low variety of the things lower in the stack, pets versus cattle. What I have seen in many places now is where someone on high say, "We are going to bring together all the people that it takes to deliver something, not create something but deliver something.” That means we have an infrastructure person and we have this type of person, and all the way through the stack.

What that leaves is fracturing in the things that should be highly repeatable where we're trying to optimize for unit cost. Those things should be highly repeatable. Is anyone in this situation where someone's like, "We've got to bring together everyone that it takes to deliver a few people?" You should watch this video, "How to make your own toaster from scratch." Would you ever want to make your own toaster from scratch? Are you going to go get the iron ore or whatever it is? Are you going to try to make your own plastic, like find a dead dinosaur somewhere? No, this is ridiculous right. We need some things to be commodified so that we can achieve differentiation higher in the stack closer to the vast array of needs that are present in our market.

We have this, that's not the only place that we see that pattern where lots of variation and differentiation is enabled by having low variation. We also see that in strategy, if you work in a company that has a long-term strategic goal then you come up with different things to try against that goal. If you work in a company that is thrashing when it comes to their strategy, where the dev teams are on a two-week cadence, so now the executives are on a two-week cadence and they're going to come up with a new focus every two weeks. We've all worked somewhere like that before. What we're looking for in strategy is to have a period of lower variation where we're only adjusting that goal if we get some really like transformative information but then we have tons of variation out at the ends there, at like the capillaries of our organization, where we're rapidly going through new information. We get lots of new information in and, hopefully, we tell people about that and it goes somewhere.

The second thing, low variety enables high variety. If we have a goal it says, "You should play in this space and if you're playing in this space we'll give you some money to try things out," and then you can try a bunch of things out in that space. We see that pattern multiple places where an area of low variety, will enable high variety.

It doesn't look like this beautiful little thing though, it looks much more like this. Why is that? Because the daily things that we try out or the two-week things, all of that experimentation and the processing of information that happens way out there at the people who are actually doing the work, there are tons and tons of those stories of, "What I'm trying today or what I'm trying for the next two weeks." Then there should be maybe one story of what we're trying for the next 10 years. It's not the beautiful little set of boxes, it's much more something like this but this is desirable.

That's actually what we want and I didn't make that up. I'm not imagining these patterns. This is from a book that was written a long time ago, "Introduction to Cybernetics." Anyone a big cybernetics fan in the room? This talk seems like I have a lot of soft skills and social skills. I don't actually, that's why I like cybernetics because it implies that you could just calculate how people will act but this is Ashby's Law of Requisite Variety. It says basically that if you want to meet the needs of a diverse market you have to be capable of providing a diverse set of solutions. Your organization must contain at least as much variety as your market or you won't be meeting the needs of the market.

Actions Available to A System

There is a keyword in there - the larger the variety of - actions. Where do actions come from? Only you can know where the actions come from because they come from you, you think of things to do and try, so let's build into that. You don't just randomly come up with an action.

The first thing is there is reality. This is all the things and you can't necessarily process all these things, you'd go crazy if you did. Then there's selective reality. How many chairs are in this room? I don't know. There are whatever number of chairs. What would make you take note of the number of chairs? We interpret that reality. There's a certain number of chairs. You might see that, and you might make the assumption there's more chairs in this room than some of the other rooms but they're significantly less than in a keynote room. You might assume that chairs correlate with how appealing the speaker will be and you might conclude then that I will be slightly more interesting than some of the speakers and significantly less interesting than the keynote speakers.

That might inform your beliefs, so you might draw that conclusion and then you might say, "People named after animals like cats are slightly more interesting but significantly less interesting than the keynote speaker." If you're looking for a moderately interesting talk then you would look for people who are named after animals in the agenda. That would be your action. When you're looking for a strongly moderately ok talk, you would attend a talk that is provided by a speaker named after an animal.

I did not come up with this, this is something very old, it's the "Ladder of Inference." If we want to maximize the number of actions that are available to our systems so that we achieve scale, we can serve a greater number of different needs, then what do we need to do? We need to open up that pool, so we start out with everything on the table. We start out with a whole world that's on the table but then it gets filtered down. We only notice some things. How can we make people notice more things? What if I told you that ahead of time, I told you the number of chairs in a room directly correlates to the quality of the speaker. Then you would just go around and see how many chairs are in each room, and you would go to the talks that have the greatest number of chairs.

Here if we introduce metrics and things like that we can get people to notice different things. Earlier, I don't know if anyone's was in Katherine Kirk's talk but she talked about habits that people have that they're not always sensitized to, they don't notice that they're doing something that they're doing but especially when they get stressed they fall back into that habit. If we're trying to create change or if we're trying to open up options we have to sensitize people to the habits that they already fall into. When you're thinking about OKRs or whatever your goal-setting process is, maybe something that you could add to that is a mention of the habits that you know that people have within the organization so that they become sensitized to that and they don't just fall into that muscle memory. You can introduce metrics that are not so much about where you want to go but what you want people to notice.

The other thing here is that those things are valid in a context. If I tell you that the number of chairs directly correlates to how great a speaker is, what happens if you go to a place with no chairs? I don't know where that would be but, or something of this sort. The context could change or we could go where it's a single track conference but there's an open space or something like that. Then how do you choose because all of the chairs are in one spot. In that goal setting, keep in mind that those things are perishable. That sensitivity that you're developing is perishable, as people establish a new habit then you have to refresh that. How frequently are those things refreshed and how frequently should they be?

When we think about conclusions and assumptions, I make assumptions about why people do the things that they do and I make conclusions based on those assumptions. If we want to influence what people assume about their world and the conclusions that they draw, then we have to make their world bigger. That means exposing why you are doing the things that you're doing to your peers.

If you have OKRs or you have some strategic intent or whatever the case may be, if you share that more widely people won't have to assume what your motivations are, they will know because they will have heard it before. We should be sharing those things if we want to open up a space for more actions than we should be sharing those things so that we don't assume about why people are doing a thing but we open up that space and we learn why they're actually doing it.

Strategy Deployment

We'll go through a couple of those tools. Has anyone heard about strategy deployment before or anything like that? It's similar to OKRs, one of those things. I'll just tell you what is in mind so that this is literally the tool that I use to try to open up that space in my work.

The first thing is an intent. What would I like to have right now? What will make my world better if I had it right now? Then how will I know when I've actually gotten that thing? I had a friend who was working on this delivery pipeline and they made the mistake of saying what they needed right now was a delivery pipeline, not the thing that they actually needed, which was to deploy more frequently. They said, "We need a pipeline," and they didn't describe what that would look like at the end. They did end up implementing a pipeline that allowed them to deliver at the exact same rate as when it was all manual. You should probably announce how you know you've arrived. It was a very sad time for him, not for me, I was laughing.

What will happen if you don't get it? What does this reveal? It reveals the assumptions and conclusions that I've made about my world. If I say what will happen, if I don't get that, that reveals what I think about the causality in the environment.

The current state. What's the interpreted and selected reality that I'm making note of? You tell people, "Go sit down. Write down the current state of where you're working right now," and have them compare and they write down radically different things, then you realize that they're working with a different set of assumptions about where they're going, about why people do the things that they do so I want to make my assumptions about the current state explicit.

Define the playing fields. We have these assumptions. Has anyone worked anywhere before where you thought something was a rule and it turns out it was just a routine? I've done this so many times, "I don't know. Everyone was doing it. I never asked, I just thought it was fine." If there are things when I tell my team, "This is our goal. We're going to go do this thing," and if I recognize that there are routines that they've fallen into that aren't actually rules and I don't need them to abide by that fake rule, I will put that in the goal, put that in the intent. I will say, "You have the freedom to not do this."

We might make an assumption about what's ok and what's not ok and it limits the set of possible actions because we've made this assumption. Then leading indicators, this is about your beliefs. I believe - of course, I'm setting this goal out there - it's the right thing to do. How can I get information earlier to know if it's right or wrong or if we are on course, if I should change my beliefs? The example that my friend Jay Bloom uses is you are about to head out on a cross-country car trip and both of your kids just down like half a gallon of water, just chug it right there. Is your original estimate of when you would arrive at your destination still valid? No. Maybe gallons chugged is the leading indicator for when you will arrive, and cause you to rethink your beliefs.

Then I wrap that all up in an episode name, I'm fascinated by this. I don't know what some UK examples would be but in the U.S. I've never seen this episode but people will be like, "Oh, yes. The hairbrush episode of 'Friends.' “I've never even seen "Friends." They know what the hairbrush episode is or the pretzels episode of "Seinfeld." I think you should be able to refer to these things by a quick name, like the Darmok episode in “Star Trek”. That stands in for a bunch of other stuff, so that can help us have those conversations about this option set that we are creating more quickly if we have those episode names.

Right now, I'm working on the transaction engine, the core ticketing engine for Ticketmaster, we say, "Let the court ticketing platform sell tickets." That's our episode name, we're not trying to have it do a bunch of other stuff. We just want to let it sell tickets.

A3, anyone familiar with this? Another great tool for sensitizing people to options that they may not know exist. It's just a quick thing, it contains many of the same elements, it can be done much more frequently. This is like, I'm trying an experiment this week and I'm summing it up this way. It's not a big goal, it's just this is the thing that I'm going to try. It contains many of the same elements, the background, the current situation. What are we trying to do here? Sensitize ourselves and make ourselves open up that space for more options.

If you want more options available to people you have to encourage them to first take an account of the situation. Without any judgment, I'll go through and count how many chairs are in here and then begin to account for the situation because there could be lots of possibilities. First, we take an account of the situation, that's what the A3 is good for, that's what the goal is good for. We just want to, without emotion, without trying to explain things we want to take an account of the situation and then begin to account for the situation.

It sounds really easy, just implement these things and we're good to go. Unfortunately, it's not, there's nothing we could do about reality. Reality just is what it is, it's the circumstance into which you have arrived. There is some great story, I can't remember all of it, about trying to put a smart vehicle and it arrived way too early and people were like, "This is stupid. I don't want any of that." Then 10 years later people are like, "Oh, never mind. I want my Tesla to park itself."

You're born into what you're born into, those circumstances are the circumstances. Everything else is shaped by history. It's your experiences, your socialization within your company, and just in your life, in general. If you have the experience that people get punished when they try something new how likely are you to experiment? If you see that the people that are rewarded are the ones who do like really variety introducing innovation, that's usually what we mean when we say innovation, something that introduces more variety into a system. What if you work on infrastructure? You're probably going to be introducing some really interesting variety at that point.


The other thing that is totally messed up is this, that you can think, you can know something to be true that is not true. Your brain will fill things in based on those patterns that you've discovered by living. I'm looking at the audience right now. I'm not going to memorize each one of your faces and later I will remember this and maybe I'll completely get it wrong, I'll say it was 50% women because that's just what I think is true. Your brain fills in these things, even the things that you think that you know, that's why we're really trying to be present in this moment of stepping through all of these things when we're trying to think about what options are available because even the things that we know are true are often not true.

That's because our beliefs inform our view of reality. They're going to inform the way that you filter through those things. You start out and all things are possible. You just have what actually is in the world and then you only notice a certain number of things. Now you only have what's possible in those certain number of things. Then you only really try to explain and make sense of an even smaller number of things and on and on. You don't even know if those things that you're trying to make sense of are real. We end up with this very limited set of options available to an organization that is scaling getting bigger so that they can provide more options. It's very dangerous.

Ashby’s Law

We are back to Ashby's Law, how do we get so that more actions are available? We talked about the tool of trying to build that shared coherence. We have the goals and we have the A3s but it's probably also about individual people. If we want a variety of actions available to our system so we can serve a large number of customers, so we can continue to be in business, which is the thing that I've always liked, then we need to have a variety of people that come from a variety of circumstances and a variety of experience.

I was just chatting with someone who is a fancy person at a big cloud company, let's just say that. He was talking about this as a risk mitigation strategy, because if you have a really homogenous culture they may not even be able to notice the things that could take you down. They are talking about how they can introduce the variety in the people and their experiences and the skill sets within their team as a way to mitigate risk. This is tied to an actual business objective. How are they going to achieve scale? The interesting thing about this is that this is an okay variety to incur anywhere in that stack, regardless of whether you're providing infrastructure or you're building platforms or you're building applications or some little MVP, this is the variety that's good to introduce all over because it'll just open up that option set.

Conway’s Law

Then there's Conway's Law. Anyone ever heard of a little thing called Conway's Law? Wow. Did some people just groan? That's an interesting response to Conway's Law. Organizations which designed systems are constrained to produce designs which are copies of the communication structures of that organization. Basically, the way that we interact will manifest itself in the systems that we make. It's the communication patterns from different companies and how those manifest in the way that we experience those products. That's basically Conway's Law. What if we're doing the same thing with our thinking patterns?

What if the way that we think about what we're going to implement, the way that we think of what's possible within our ecosystem, even just not the actual things that we're thinking about, but the way that we're stepping through that thinking process, whether or not we're open to different things being possible, what if that is manifesting in the systems that we create?

I would argue it is, it totally is. We have Facebook that had this really - shall we say - distinct culture and it manifested in the way that they implement that platform, in the way that they do business in their business model. It basically goes to that layer that I had in there, that's about platforms and it's about demand-side economy of scale. What they're trying to do there is get you to be really engaged, regardless of whether that's because you are very unhappy or because you're part of a mob or because any of these things. Is that encouraging you to think slowly, be present in the moment, and think of the things that are possible? No.

Of course, we know because they've been pretty transparent about that, that they weren't doing that in the moment as they were implementing those things. Not that they're evil people, it's that they had this objective and didn't realize all the different ways that that could maybe not turn out great. It's not a bunch of evil people got together at Facebook and they're like, "Brexit, Trump." I don't know what else.

I think it is not just not good for business, it obviously is good for business if you can consider that lots of different things could be possible but I think it's just good for us as people, to be open to lots of different things being possible.

With Great Power Comes Great Responsibility

What I think is really interesting about right now is that we are being confronted with our own scale as people. The things that we do, my carbon footprint, that was in the opening address. We talked about like carbon offsetting and things like that, there was a session yesterday about all of those things. One country or one municipality or whatever the case may be can make a decision that affects the rest of the ecosystem, the global ecosystem.

Right now we are being confronted with the questions of scale. How can you be both small and big at the same time? Because we are big, we're tons and tons of people, but how can we also be small? Slack is a great example of that. How can we be small at the same time as we're being big? Twitter is big all the time, but when you get into a shared DM or you get into a private Slack that's also small. We're being confronted with those questions of scale. What does it look like to be both big and small at the same time?

Then we are being confronted with the fact that computing is still in its infancy. Computing will scale hugely in the coming years where we can see that. A.I., all of these things that aren't quite there yet. Each one of you individually has so much more power than the next generation and the generation following that because soon those jobs, technology jobs, will be all of the jobs. We're looking at automation and things like that so it will just be all of the jobs. That means that you as you think through the things, what does that mean for us to get bigger and bigger in this platform? What does it mean to serve more and more customers? You think through the options that are available.

You just buy in the math right, there are so many fewer of us right now than there will be in future generations in computing that each one of you has a responsibility to be present in the moment and to think through these things and to look for all of the options before you commit on behalf of tons of other people. You thought this was going to be a funny talk on the first slide. Uh huh, I see your disappointed looks.

How long was the agricultural revolution? How long has it been since they signed the Agile Manifesto? Was the agricultural revolution shorter or longer than 18 years? Just a smidge longer? Even if things are speeding up and up because that's what they do with scale, we already know this, things speed up. Even if things are speeding up significantly we are still in the infancy. The things that you do today matter, how you scale matters. It's not about getting bigger, it's about being both big and small at the same time and it's about doing the things responsibly.

I like talking about this but mostly I like it when other people are involved in the conversation and it's not just me talking to myself for the whole time. That's probably my worst case scenario. I'm @catswetel everywhere, Skype, Twitter, just life in general. You can find me and try to talk to me about whatever these things. I'm open to trying different things and I'm open to being told that I'm wrong.

Questions and Answers

Moderator: That was wonderful and mind-blowing and it's a bit scary.

Participant 9: How do you term books, your opening up of options and variety, in the face of time pressure?

Swetel: I'm so glad you asked this question. I put a little note down here about "The Requisite Organization," which is the book by Elliott Jacques and that research has been carried on but he did all of this research over decades in all of these large companies and it was about what is people’s concept of the present. He arrived at these different time spans and the first one is you are on the front line of whatever you do, you're thinking a week or a month or something like that. Then the second one is you manage those people and you're thinking about months and on and on.

I looked at that research so that I wouldn't be like trying to fracture that timeline too much. It's really bad when you go in one direction, or the CEO has a 10-year plan and then you just go straight to the development team. You're like, "Great. What's the sprint goal?" That's like when you go into the all-hands meeting and then everyone leaves and just goes back to the exact same thing they were doing before. You need that step-down, it's the same way that our blood circulates. You have the heart pumping a bunch and then it steps down and it steps down and it steps down. Basically, what I took was I just looked at the work of Elliott Jacques and then I looked at how much things have sped up now.

Elliott Jacques, when he first started the research in the '50s like you don't have a computer let alone CICD, I adjusted it based on that. That's not really a great answer but I can e-mail you literally the timespans that I ended up with, but what I would say the best guidance I can give you, I use being a consultant. I go in lots of different places. Always start at the people doing the work and say, "What's the longest time span thing you're working on right now?" They'll give you some answer and then go up one from that. "What's the longest time span thing you're thinking about right now?" If it doesn't end up in this - these are to scale - if it doesn't end up in that shape I realize there's potentially an issue, that there aren't enough layers or there are too many layers but always start at the people doing the work and ask them, "What's the longest-term thing you're thinking about right now?"

Moderator: You're giving me thinking to do. I have a question about trading indicators. You talked about leading indicators and how you're going to measure things. Do you do anything with trailing indicators? Do you have them? Do you care about them or is it mostly about in the moment matrix?

Swetel: No, I certainly care about trailing indicators. Dan knows that I love data and I just love it so much. One of the things that I get asked to speak about a lot is process data and data that we collect about our processes and I talk about leading versus lagging indicators. Lagging indicators can be really helpful in coming up with like a forecast or setting SLAs or any of those types of things so that we're managing expectations and maybe establishing some semblance of trust with people but still in that case with leading indicators.

I'll use a real-life example, the duration that things take. You can create a forecast based on historical duration, how long features took to deliver. If you suddenly go from we consistently work on five features at a time to we're now working on 10 features at a time, is your historical estimate still valid? No, because we know that's just a mathematical thing. The more things that are in your system at one time, the longer each one of them will take on average. I think it's important to have both leading and lagging indicators but I think that people just completely ignore leading indicators and that's very sad for me.

Participant 10: I love the idea of diversity as a risk mitigation strategy. Particularly I think you were suggesting that it was an antidote to the fact that my perceptions are flawed. In addition that's a great suggestion, other suggestions about how I can unflaw my perceptions?

Swetel: The question is how can I unflaw my perceptions? Confront yourself with reality as much as possible. Doing that I really do try to not just take the account of and then try to account for but do all of the account fors. Typically, we're not even conscious of that. We just arrive at some way that we're explaining the things, the options that we have, but if I take an account of the situation and then I go through all of the account fors, instead of just whatever one I think is most likely, then that can be very helpful for me.

Also, sharing those things, I think I might be an over-sharer because I'm just constantly like, "This is what I thought about that. What do you think?" That's literally what I'm doing on stage and it's the only reason that I speak because I don't have people skills, this isn't my idea of a good time, but I'm saying this is what I think about the world. Do any of you think the same? Do you think differently? I've got to place myself in this context somehow.

Participant 11: You made a really good case that high variability is necessary if you're in an environment where the context is really variable to just cope with it. Is there any way of knowing or any tools for thought you could give us about how you know whether it's the right variation, whether it's actually a positive adaptation, because it's certainly necessary that you be able to be flexible in a strange environment but I guess you could also be chaotic and have a hard environment.

Swetel: This I think is a great way of deciding where it makes sense to incur that variety. If anyone's read the Scrum Guide before because I hate it when people tell me, "This is what Scrum is." It's like ok, I can easily go look that up so that's what I do. Like if you've read the Scrum guide it doesn't say, "A cross-functional team contains all of the skills necessary to deliver an increment." It says to create an increment, so we don't need a marketing person and all of those things, because that's part of the delivery, we may pull them in as necessary. When I try to think about where to incur that variety, I use Wardley maps.

It goes from the invisible to visible, that's the Y axis and then Genesis, meaning this is something that's highly unique, we just made it up to commodity. This is a real one that I just like anonymized. What they had going on here was there were customers who needed reports so that they could understand what was going on with their customers. They had this and they needed one person from each of these different I like systems that are transactional in case you're not picking up on that. Financial services, ticketing, they had all of these different transaction engines, instances of the transaction engine, and they needed one person from each instance to come in and give them some specialized knowledge.

Instead of asking "How can we do that better," they started to ask, "How can we not do that at all?" By mapping this and realizing it's not visible to those customers at all, it's a product, it's not anywhere near Genesis. That's been around for a long time. They thought, "What would it look like if we consolidated this and tried to shift it more towards commodity," which means not incurring any variation there but those moments of transition are the hardest ones and that's where you need people, going back to your question, that can think about along right now where that's already happened. That's something I take for granted, that we need to do that, so how are we going to get there.

Moderator: Fantastic. I have a theory that Wardley mapping is the new Conway's Law, that within five years every talk will have to have a de facto Wardley mapping reference in it somewhere. Thank you so much, Cat. That was brilliant.


See more presentations with transcripts


Recorded at:

Jul 24, 2019