Sense & Respond | Josh Seiden | BoS USA 2017

Josh Seiden, Author, Sense & Respond

Software has fundamentally changed the way business works. It’s not simply that software has given us new ways to reach our customers, or new capabilities to achieve our goals. Software has enabled—in fact it demands—a new way of working. Organisations that seek to thrive in the digital age must examine their core management practices. So much of our management practice was created in the industrial age and not all of it remains useful in the face of software. Managers and leaders inside and outside of the software world need a new set of principles for the next century of work.

Video, Hangout, Slides & Transcript below

Video

https://businessofsoftware.wistia.com/medias/dbzmwz96td?embedType=async&videoFoam=true&videoWidth=640

 

Hangout

https://businessofsoftware.wistia.com/medias/0gvuzqxxtt?embedType=async&videoFoam=true&videoWidth=640

 

Slides

 

Transcript

Josh Seiden: All right! Thank you, Mark! Was that just completely random or what?

Just hold in your mind while I’m talking and trust me when I tell you that the set of circumstances that led Joey to being here in this auditorium today were let’s call them unpredictable. So nice weather and end of summer, hurricane season!

You know, I’ve been thinking about buying a new car recently and something got me thinking about a brand new Cadillac. So I went to the Cadillac website and this is what they had featured today. I don’t know about you but I thought to myself it’s the end of September, I don’t want one of these crappy 2017 models. Let’s move those out of here, cause I’m waiting for the 2018 model. For me the end of the season brings new cars, it’s a natural phenomenon. Every brand, every model, we expect a new model year to start showing up around this time of year. It’s like a natural phenomenon! But it’s not a phenomenon, it’s a deliberate strategy on the part of auto manufacturers to drive seasonal demand. This goes back to what Scott was saying, all these old ideas that we can look at. In the car industry it’s 100 years old idea and goes even further back than that, to the early bicycle manufacturers. We think of this as the natural rhythm of business, every year there’s a new model and so much of our management science is based on this annual rhythm and we take it for granted, that’s the speed which business moves. It doesn’t have to be that way.

For example, if you take Tesla. Tesla is a car company that have to move a certain pace because cars take a certain amount of time to design, plan, test, build. But Tesla is interesting because they overlay a second rhythm on top of the rhythm of the automobile industry and that is the rhythm of software. Tesla has this amazing capability built into their cars and in to their operations which is that many of the features of their vehicles are implemented through software. So if you’re an owner, one of the benfits is that Tesla can do ‘over the air’ updates of your car. You can go out to your garage one morning, wake up and there’s a new feature in your car because they beamed it out over the air and your car updates like your iphone.

There’s a great story that happened at the beginning of the year, a Tesla owner in California drove up to the super charger station to charge his car and there were no open charger slots. And he was frustrated and he Tweeted to Elon Musk and said it’s full of idiots who leave their Tesla for hours even if it’s charged. I don’t know if you can picture that? It’s right by Starbucks and people leave their cars and go and do their stuff. And so Elon Musk tweets back you’re right, this is becoming an issue. We will take action. And 6 days later, Tesla releases an update to the software in their cars and now when your car is done charging, you get an alert on your smartphone and it says your charge is over. If you don’t move your car, we will start charging you a per minute parking fee. That’s pretty amazing, that they can do this. 6 days from Tweet to feature. So this is an example of – this is the obligatory – here it is, I love the egghead and cone head picture. This is an example of what Mark said in 2011, right? Software is eating the world and every company is becoming a software company.

No industry is immune

I do a lot of consulting and training and I travel and work with a lot of large companies and I frequently get this pushback, our company is not a software company, it’s a bank. I want to read you a quote from the CIO at ING, one of the largest banks in the world who is talking about their digital transformation efforts. He said in our case there was no particular financial imperative to begin digital transformation. The company was performing well, the market was favourable. But he said that the customer behaviour was rapidly changing in response to new digital distribution channels and customer expectations were set by digital leaders in other industries, not just banking. We came to the realisation that ultimately, we’re a technology company operating in the financial services business. Now, all of this expectation setting from the best performers in the industry, it’s that when I tweet at Elon Musk, I’m likely to get some response, not just over Twitter, but in terms of some responsive new behaviour from that company. And that expectation is built on the nature of software. Now it’s that software is in continuous production, it’s not in the discreet production that we inherited from the auto industry and that enables us to work in a new way, to build things, to put them out in the world and sense how people are using them, sense changes in demand, things that we couldn’t predict, like a dog at a conference and then respond appropriately.

World Views in Conflict

Now, in 2013, I along with my colleague Jeff, we wrote a book called Lean UX and it was a book about how digital product teams could work more effectively in this new way, how designers and product managers could collaborate inside companies to be more effective. And what we discovered after the book was released and we were traveling around and teaching this method, we got an unexpected set of feedback from the teams that we worked with. They told us we really want to work this way, we like working this way, we think this is the most effective way for us to collaborate, to move the needle for our businesses, but we can’t work this way because our boss won’t let us or because our organisation isn’t set up this way. That was a fascinating piece of feedback for us and we started to dig into this and thought surely this is because your company is big or you’re a bank not a software company and we discovered that this dynamic exists at every company in one degree or another whether you’re a software company or a Ford or a Chevy. And the problem comes from a clash in world views. This new world view of this agile, iterative process that we think is normal in the technology world and the linear planning models that we’ve inherited from our industrial forefathers which made sense, to make a car you had to go through this linear planning model from design to test to build to ship or sell or sunset for next year’s model.

Lean + Agile = Continuous Learning

Now, there are a lot of process models for this new way of working and I think one of the most important things they share, whether it’s agile or lean start up or our lean UX methods, is this idea of continuous learning and I will talk about why that’s important in the 2nd half of the talk. The world we live in now, we’re in that 2nd half of the chessboard, to go back to the previous talk, we’re in a world of great complexity and difficulty predicting what will happen next. So the best way forward in the face of that uncertainty is to use process models that are built around this idea of continuous learning. That implies a great deal of organisational change though and that landscape kind of sucks. This is what it can look like in the world of organisational change. This by the way, is the Agile landscape version 3. So these are the methods that you might choose from if you’re trying to engage your organisation and digital transformation but it’s not about the methods so I won’t talk to you about specific methods, but I want to talk about principles that I hope you will embrace and think about how to drive them forward in your organisation.

5 Principles to Build a Sense & Respond Culture

1. Embrace Continuous Change

And the first one I will talk to you about is this idea of continuous change. Embracing it! I like this picture. It says one day’s output at Chevrolet factory. I’m not sure but this is in the east bay of California and may be the same plot of land on which Tesla builds their cars now. But this is an interesting world. If you go to work in the morning at this Chevy factory, you know what done looks like. Maybe you’re someone who works at wheel assemblies and you know what a finished wheel assembly is and you know when it’s done, somebody is going to attach it to the axle, then the chassis that car will be done and will roll off the line and when you have that kind of clear definition of done, you can plan this way. This is all the work that goes into making a car and you do all these jobs, car rolls off the line.

Software is not like that, how many have an Instagram account? I have one, I also have teenage daughters and one of the things I can tell you is there’s a huge amount of social pressure on social media to look great and if you’re a girl, there are some things don’t make it into your pictures. Your fat dad and whiny brother, those things don’t make the picture because you have to look good. One of the interesting things and I just learned about this a couple weeks ago from my kids, in response to social pressure they have come up with FINSTA’s. How many of you have heard of it? It’s fake Instagram accounts and they’re a second account that kids create that are private and invite only their closest and most trusted friends and it’s a place where they can be real and goofy. That same social pressure was the opportunity for Snapchat –self-destructing messages, we can post goofy silly stuff that goes away. The success of Snapchat led to Instagram implementing their own version of stories. So my question for you is, when is Instagram done? It’s never done, so it’s not like a Chevy or a Tesla. And this reality is what’s led us to the sort of process models that we see in today’s world, whether it’s Agile or design thinking or lean start-up or the one in the middle is called beyond budgeting and it’s a way to do continuous financial planning. These continuous process models are about continuous learning and they acknowledge this idea that software is never done.

2. Manage via Outcomes

If it’s never done, how do I tell my teams what to do? If I’m not saying we’ll build a car and let it roll off the assembling line, how do I direct my teams?

Anybody knows what this is? A rotary engine, you know you’re at a conference of geeks. This is a combustion engine and I don’t think there’s any production cars using it now, this is an example of a complicated system. I can give a team of skilled engineers a rotary engine and say figure it out, make me another one. So you need skill to be able to duplicate this, but I could write a set of specs and I could say do this thing and a team of engineers could make my rotary engine forming.

This is an intersection in India. So this is one might think a whole less complicated than the rotary engine, it’s two lanes of black top, a T intersection coming in to the side, there aren’t any traffic lights here. But if I asked a team of skilled engineers to duplicate this, even in Boston, you couldn’t do it because it’s a social problem. This is what system theorists call a complex domain and the problem in them is that it’s very difficult to specify how you were gonna get this resolved. I can’t say to a team of engineers go make me this as much as I would like to. So translating that back to thinking about management, if I want one of these, I can say here’s the outputs I want, go create these outputs, make me this set of features. If I want this though, I need to specify the outcomes. And outcomes are the desired user in our world, and customer behaviours that deliver business results. So let me say that again,

outcomes are the user and customer behaviours that drive business results.

If I’m operating in a world of complex systems, the way I need to direct my teams is by specifying the outcomes that I want them to create.

Let’s bring it back to software. Let’s say I want to go after Instagram. I want to create an Instagram killer. I could tell my team, take a look at it, figure out all the features, copy them and build me those outputs and features. In fact, that’s what Mobli did in 2013, some reviewers even says they out-featured Instagram. By 2016, Mobli was bankrupt and they were bankrupt because even though they were able to duplicate the feature set, they weren’t able to generate the outcomes that Instagram generates. The outcomes are an engaged customer based, looking at photos x times per minute, scrolling pictures, posting stories, whatever those outcomes, those behaviours, that we’re trying to generate. That’s the thing that we need to do if we wanted to go after Instagram. Now this is hard, I admit it, because so much of the way we organise our work relationships inside and outside our company, are features. I tell my team build me these features and they say got it boss! Done! Or I hire a vendor and I write into a contract build me these features. It’s easy and traditional to manage that way, but it doesn’t make sense.

I recently was working with a large client and we were planning a whole year of work to launch a new product and we had some idea of what it would cost to spend a year doing this, we had to go through the company’s funding process and it required us to specify ok we want a year’s worth of money, a year from now tell us what features you will have built. We said we have no idea what features will work, we will experiment our way forward and learn continuously. It can be done in large organisations though.

This is Pearsons product life cycle which attempts to solve this problem in a way that’s very outcome centric, very lean start-up oriented and very much like discovery driven planning where you can see these are different stages in the product development lifecycle and each is organised around a question that relates to an outcome. The first stage is, can you identify a core user problem or what outcomes are the users trying to achieve that they can’t achieve today? The second stage, are there real customers who’ll get value from our solution or will our concept deliver those outcomes? Third stage, is there a business model and will it generate business results we’re trying to achieve? It can be done, but requires a massive effort to reorient our process from outputs to outcomes.

3. Create a 2-Way Conversation

How do we know we’re right? When Jeff and I were researching our second book, a lot of the ideas for this talk are from it, we spoke to a lot of people working in this way. One of the teams we talked to was the Obama for America campaign and it’s interesting and the first digitally native presidential campaign. And one of the things they said was this phrase with us, everything we’re doing, all of our outreach is designed to create a 2-way conversation with voters. And they’re aware that the traditional advertising conversation and dynamic of campaigns was a lot of one way communications. We shout and try to shout louder and more effectively. What they were trying to do was to create a 2-way conversation so everything they were doing was inviting an explicit response from their audience. This is a campaign page inviting people to post to Tumblr in 2012 with specific hashtags and get their phone number so we can send them text updates and once we’d identified people who are responding to us could we use them to activate our friends? Everything they were doing was about an outreach that was designed to generate a conversation back. This 2-way conversation with the market.

In the Tesla story, there was both a literal conversation over Twitter and also this metaphorical conversation. Musk didn’t respond with a new feature because he thought it was cool or he got one tweet, but because he knew it was becoming a problem. They had data through the car’s telemetric and the data the charger was returning and knew what was going on in the business and that Tweet just kind of crystallised it. Right? So there’s that 2-way conversation.

Now the 2-way conversation can go badly. I had this experience recently. I installed – this is an awesome product, a plum Wi-Fi mesh and it’s a home mesh Wi-Fi you install it in your house and get these pods over the house and it gives you great Wi-Fi. And I love this product and bought it for my house and my parents’ house and then I planned to go away on a trip. And so I shut down the internet in my house and I pulled the power to the plums and a couple days later I got this email: Hi my name is Chad. We’ve received an alert that your pods have gone offline and I wanted to reach out to make sure that everything is ok. Have you disconnected your pods or are you expecting to be offline? I thought that was creepy as hell. It’s none of your business, is my toilet seat up? So I wrote them back a nasty-gram and said it’s none of your business and you should tell your product manager that I resent the intrusion into my privacy. Chad was just really doing his job and he sent me a link to the privacy policy. I was like yeah. So Chad, if you’re here – how do you avoid being Chad? That’s my next talk, Mark, next year – avoid being Chad. So I think that what Chad was not able to do was to turn my response – they weren’t prepared for me to write back in the way I wrote back. The only thing they were prepared for was for me to go omg my wi-fi is offline, let me plug it back in!

What that points out is you need to be prepared for collaborations. You can think of collaboration as 3 levels, the level we usually talk about which is the team level of collaboration. And this is hard enough, this base level of collaboration. These are the PayPal offices after they reorganised to move from a building with engineering, management and designers, they re-organised to create these pod like spaces where small cross functional product teams could sit together. That’s the first level is creating inter-disciplinary collaboration on your product teams. Small, cross functional, co-located and self-sufficient meaning they have all the skills on the team they need to push the software live.

There’s another level of collaboration which has to do how you coordinate the activity of multiple teams. This is a project room at Westpac in Sydney, Australia. Westpac is not only the oldest bank in Australia but the first corporation. They’re a massive bank. This project room is a room in which they’re planning a big program and their goal was to change the way they issue credit card. Currently, from first interest to issuing of the card it takes on average 7 days. And what they realised was during this time there was an interested customer waiting to use their product who wasn’t using their product. So there was all this opportunity cost in that time and so they ask could we bring that time down? So classic customer experience planning, on the top there’s a tier of today’s universe, the next stripe down is tomorrow’s universe, this is what we’re trying to do. And then what they did, they just bring teams to this project room and march them through the wall because this is a cross-channel problem, they have to coordinate the efforts of retail and online and legal and risk and technology, a huge number of groups that needs to be coordinated around this. They say ok risk team, here’s where we are today and where we want to be tomorrow. What can you do to help? Ok retail team! What can you do to help? So aligning around the customer journey helps them create these collaborations that coordinate the activity of multiple teams.

Then there’s this third level and I want to share with you a video that I think nicely illustrates this level. This is a weight watcher’s ad from a couple years ago in which they’re introducing a new product that runs on a smartphone. So we will run this video.

I’m a coach, I’m real! What’s this feature on your smartphone? A phone you can use to talk to other people! Amazing, right? But what actually happens on national donut day in that call centre? Is that just a line or real? What happened on that first national donut day? Were they swamped with demand? How do you know and what are people really calling about? Are they calling Chad and saying it’s none of your business if it’s the national donut day. I’m mixing the stories, but the idea is this business has conversations with its customers in real time. And a business that’s organised around this doesn’t outsource them to a 3rd party service provider but has some way of creating that collaboration line from the customer to the coach to the product.

4. Create a Learning Culture

All right, so the last thing I want to talk to you about is creating a learning culture which, in some ways, is the hardest obstacle here. I want to show you this intersection, this is a treat for me to show this video in Massachusetts cause it was filmed here. Anybody want to share with me? Don’t make me say the word what we call their drivers. Thank you! Let’s say one day I said boss, this intersection is out of control. I will put up a stop sign and a no left turn sign and my boss would say great idea, Josh! That will definitely solve the problem of the intersection. But it doesn’t because again we’re dealing with these complex systems that involve the behaviours of these independent actors. Why is this happening? We don’t know but our first guess of trying to get control of this which is a reasonable guess is wrong. And that’s the reality of dealing in a complex system is that we work in the software world. We can have these ideas – they’re rational, they’re reasonable, nobody would disagree with them and there’s a good chance we’re wrong. How do we deal with that? We need to start from this place of humility that says as a good idea, I could be wrong.

We worked with a client in Germany, they’re a sort of a German Netflix, and one thing the CEO does every holiday season, is he sends his managers out to the Christmas markets to try to sell subscriptions to his service to people who are holiday shopping. And the reason he does that is not because he thinks he will sell any, but it forces the managers to pitch the service and understand what people really think of the products they’re selling. It’s an incredible way to generate humility is to go stop a stranger on the street and try to sell them their product. So this willingness to admit that we’re wrong is fundamental to creating a culture of learning. Laszlo Bock said without humility, you’re unable to learn. On that note, I will leave you with a quick sum of the principles for creating this transformation.

• Embrace continuous change,
• manage via outcomes,
• create a 2-way conversation with the market,
• create collaborations,
• and create a learning culture

Thanks very much!

Q&A

Audience Question: You led with the buying a Cadillac. I know car development cycles is 7 years to get a new car out. They put a new one every year but not much changes. When you have a company the size of Cadillac there’s a lot invested in keeping the lights on and many people that don’t build cars do a lot of things. If you’re Cadillac, what are you picking up? How do I make that carry throughout the organisation?

Josh Seiden: I’m not sure I’m qualified to give strategic advice to Cadillac but I think the reality – I think the thing that I’m seeing car manufacturers struggle with now is whether and if and how they can develop their own native software capability. So earlier this year, you saw an announcement between this google spinoff and Chrysler, they announced at the Detroit auto show, a self-driving van. They each claimed it, they have the first self-driving van and Chrysler press release said the same. In that conflict you see the problem. Who will control the auto industry? Is it going to be the hardware manufacturers or is Chrysler the android handset of the future? I don’t know but if you look at Tesla that’s building this capability into the DNA of the company from the ground up, you have a model that is compelling and should be terrifying and in the Cadillac you have a more traditional outsourcing strategy. To me that’s the question you’re wrestling with is the ability to which you can build that native capability and in that question you see large organisations struggle with that across the board.

Audience Question: Thank you for a great talk! A lot of what we all do is we need to make decisions about resource allocation. You say customer service shouldn’t be something you outsourcing to another player cause you get the feedback from the client. What shouldn’t be the core competency of the company?

Josh Seiden: What should we stop doing? That’s a great question! I don’t know. I know I’m supposed to have some clever answer to that, but I don’t. I think there is – what I’ve seen my clients struggle with is what to pull into the core capability of the product team, to what extent customer service should be a shared resource and if it’s that, how do you stay in touch with customer service? We launched a start-up service a few years ago for a client in which in the early stages, customer service, it was a meal planning service so we had customer service nutritionists and food editors on the core product team. In the long run we thought we will spin some of that out into shared services, but initially they needed to be a part of the core team cause we were talking to customers every day and needed to build that feedback back in to our product development process. I see teams struggle with that with what to balance or how to balance core team versus shared services and and I have less exposure to your question about what we should be stopping.

Audience Question: When we’re talking about embracing constant change, how do you know that you’re making changes in the things you’re aware of in the environment versus just getting caught up in the idea and manufacturing change for the sake of it? Or is it ok to manufacture change?

Josh Seiden: I’ve never encountered a situation where people were changing too fast or for the sake of change. In my experience, there’s always more change that you can respond to and more to do than what you can respond to. And so what I’ve seen is that it’s always a question – one of the ways that I often manage projects is through the structure, what I call the risks dashboard, which is what are the things that are coming to destroy us? So this risks dashboard, you’re looking at it every week and saying what’s coming down the pipe that we need to worry about and what’s not on it that we need to worry about? Given our capacity, what changes will we make in response to the risks? There’s always 2-3 that you can worry about and 50 that fall below the line. So I think to me that’s also – if you had all this capacity, the question would be what’s the highest risk here and the signal that I’m getting from the market that says something’s changed or is risky or I don’t understand and need to understand it?

Audience Question: You talked about humility and your idea is sound and great, but may not work. Our company struggles with this. We spend a lot of time putting together focus groups to really make sure this is what our customers want. How do you balance that and try it in the market so you don’t continue to have this analysis paralysis problem?

Josh Seiden: One way is making the cost of being wrong really low. And the cost of being wrong – and both the actual cost, like you make it easy to rollback bad changes. So you have the infrastructure either full production that has the mature dev ops capability that helps you push new code and revert or feature flags that lets you turn off stuff that’s crappy. Or – I was at a company yesterday that does a lot AB testing in an isolated environment and then turn off the stuff that doesn’t work. So you make the cost of being wrong really low culturally as well by framing what you’re doing as a learning opportunity. I tell teams stuck in that analysis paralysis, when we’re sitting in this room having compelling debates if you’re right, it means neither has enough information to convince one another. They way we get out of this, is we go get more information and the way we get that information is by shipping or taking some learning action. If you can frame it around learning and make the cost low and also don’t let yourself get into that learning debt where you’ve been working on something for 6 months and when you ship it and it’s wrong it’s brutal cause you’ve just wasted 6 months. Making it possibly to learn quickly and at a low cost, emotionally and in terms of the green.

Audience Question: A couple times I’ve encountered businesses where they are resistant to change. Your clients can’t absorb the change. For example, I build ICU software and when you get it wrong, people die. But there’s still that need to move your business forward and to incorporate that change. Is there techniques you can use to reconcile those two things?

Josh Seiden: Yeah, I think there’s two things in that question that come to mind. The first is connecting the need to change – for me this is a compelling argument. Change is like the environment we live in, whatever! So thus we change, but I think for business people who aren’t about abstract ideas, but what will happen to my business if I just stay put? We could make a strategic choice to do that, and maybe that’s the right decision for us. And if it’s not, let’s talk about change programs. First to go back and connect it to the strategic imperative to change. The 2nd thing is this idea of a sandbox, it’s a safe place to play where if you make a change to respirator software and that change doesn’t work, no one dies because you have a testing environment that gives you meaningful learning and makes it safe to fail. I’ve seen it done – I’ve done some consulting at GE software and I presented a version of this talk to engineers who write control systems for nuclear power plants and they were like safe to fail haha! But in that case, you have money to create simulated environments and there’s ways to test everything, but it has to be tied to the strategic need to change.

Audience Question: Hi! I was here last year and went to your workshop with you and Jeff and I brought it back to my company and it was embraced. Everyone was excited about it, we had been putting out product roadmaps that looked out a year and I was like no more that and my boss was like ok, that’s fine. On the team level, we started implementing it and everyone was excited about it. Respond to the customer, sense what they need, do it, build it, get feedback, listen more and do it. Essentially what I’m getting at is we hit a wall somewhere because there’s this sense of ok, sure we want you to respond to the customer and build what they want, but when is it gonna be done and can I just get a roadmap in 6 months of what will be built? And use the backlog based on what we listening from customers and having it be our roadmap. There’s these two competing ideas. What’s your advice on, it was embraced but there were things that were contradictory. We need to know when it will be built and how long it will take and we had our product department doing that overhead and estimating times and making it real to them on a high level.

Josh Seiden: I was just talking to a company yesterday that had taken some of the ideas in the book and they had restructured their product group around what they call outcome teams, which is a phrase they made up, but it was their way of expressing working in this way. The first time they implemented it, they got some positives from it, but they also had some trouble. They iterated on their process. One of the things they learned was they had to train stakeholders and feel like that’s an ongoing effort that one of the things the product managers do is they’re constantly training stakeholders to think in a new way and that’s about trust building and saying you have a business problem, we’re the experts and we’re going to solve it. The other piece was they said that there was a huge backlog of work that they tried to maintain as they transitioned from the old world to the new and they said what they needed to do, they realised they had to declare backlog bankruptcy and they just zeroed out their backlogs. Can you imagine that for a minute? They started from scratch and nobody forgot the important work. And the third thing they mentioned was there’s a slice of routine work that didn’t fit into this model, it was features that had to be built or work that needed to get done and they sliced it that off into a separate cue and it was handled by a separate team. So all of this, doing this is a discovery process with sort of figuring out how this fits into your organisation and who needs those roadmaps for what reason. The other case – there was a terrific piece on road mapping that the UK GDS wrote and if you hit me on Twitter, I’ll share the link. They said how they did road mapping because they found themselves in a situation where they had to make some concrete commitments and how do they balance that and give themselves room to operate?

Mark Littlewood: Thank you, Josh!