Kjersten Moody

Hi, everybody. My name is Kjersten Moody.

Sarah Chong

I'm Sarah Chong.

Kjersten Moody

Excellent, and a very warm welcome. We are absolutely honored to be here today to contribute, if at least in a small way, to the collective practice in DevOps.

Unilever is, I would describe it, in the middle of the beginning of our journey. So we're here to talk a little bit about what we have done so far and, quite frankly, to raise our hand to connect into this community, to start to build those relationships as we finish the beginning, and then the middle, and then the end.

Before we start, let me just introduce myself so you can understand who is talking to you today. My name is Kjersten Moody. I joined Unilever in June of 2014. You may be able to tell from my accent that I am an American. So I moved to London as part of joining Unilever two years ago from Chicago. I have spent really my whole career in technology companies, so Unilever is my first job outside of a tech or data company.

I would describe us as, in the CPG industry, a manufacturer or marketing company. My role at Unilever is as the VP for information and analytics. And what that means on the information side is the traditional BI estate. These are the reports that we deliver. My team supports Unilever's functions, markets, and categories, so we support Unilever globally, all of our users, through the BI and the information through to the data science or the advanced analytics.

It is a global team that I have, and we are truly global, with colleagues in every time zone for the planet. And Sarah.

Sarah Chong

I've been around a little bit longer.

Kjersten Moody

I know. She's...

Sarah Chong

A full seven and a half years at Unilever. I joined on the Future Leaders program, and pretty much I challenge you to find anyone else with a DevOps job title better than mine. I am the DevOps Evangelist. I haven't checked on LinkedIn, but I reckon I'm probably the only one.

Essentially what that means is I go around and I find pockets of amazingness, and I tell everybody else about them and say, "You should go and speak to this team. They're doing something fantastic. They're experimenting," or, "They've found a great way of working."

And then I speak to other teams that are saying, "Oh, I'm a little bit stuck. I'm not quite sure what I should be doing." Great. Come and speak to me. I'll try and find what you need to unlock what's blocking you from cracking on with being great at DevOps.

Kjersten Moody

Beautiful.

There's usually at least one person in the audience, usually an American, who has never heard of Unilever. Seriously. So just very quickly, what is Unilever?

We are a consumer product goods company. We have four categories that we work in. We are in home care, personal care, foods, and refreshment. We have over 400 brands globally, 12 of which generate sales in excess of one billion euros per year. You can see some of the brands there. I think everybody probably has a favorite Unilever brand.

My personal favorite is Ben & Jerry's when I'm in the UK, and it's a brand called Talenti in the US, actually, Roman Raspberry. I'm trying really hard to convince them to bring it to the UK because I actually really miss it.

We are the second-largest advertiser in the world, so that gives a sense of scope of Unilever. And I think one of the more impressive statistics about the company is up to two billion people per day use one or more of our products. And it is really that digital footprint of everything it takes to source, make, deliver, market, sell those products that generates that data footprint that ultimately comes into my team for reporting and the output analytics.

Sarah Chong

And you can't create all of those products without having some kind of impact on the environment. So we have the Unilever Sustainable Living Plan, which is really at the core of everything that we do. We need to ensure that everybody involved in the value chain, their livelihoods are improved. We aim to improve all of the health and wellbeing of all of our consumers, and actually sustainably source all of our raw materials as well.

Kjersten Moody

All right. So technology. If that's who we are, if that's what Unilever is, what is technology inside of Unilever?

It is predictably big and complicated. Some key statistics just to help give it some shape: we have over 950 connected sites in 94 countries. Well over 130,000 internal employees who would constitute the user base. That does not include third-party suppliers who would work with us or otherwise engage with our systems. And we process just somewhere under a million batch jobs per day through the various systems and interfaces that we manage.

It is a very complicated estate, I think, is probably the nicest way to describe it. And it's very interesting, actually, if you think about Unilever's history over the last six to seven years. Historically, the model was really more of a holding company, where each country sort of managed somewhat independently as this loose federation called Unilever.

It's been over the last six to seven years that we have made some very good investments in how we start to standardize what should be standardized across the world to take advantage of the scale that we actually do have. So it's coming from more of a holding company model, I would describe it, into something that is more of an operating company. Which has simplified, to a certain degree, the technology footprint, but it's also complicated it as we have started to bring together different business processes from the different countries into single platforms.

You could either think of it as a fantastic opportunity or really hard. It's probably both. But it is very much a complicated and diverse footprint that we have. And it was historically managed very much in a procedural, heavily documented, waterfall approach. So that's our starting point.

If that's what it's like internally, what's actually happening externally? I won't go into this too much because it's probably more of a month-long therapy session for somebody in IT versus just a quick snippet. But we have very strong external forces that are coming into the technology group based on how the business itself needs to evolve.

I will say that it is really being driven by the way in which consumers want to interact more digitally with the brands. And it's causing this sort of chain reaction inside the company, as we need to digitize. We need to be faster. We need to respond much more adaptively to the rate of change that we see in the outside market.

So when you put it together, we have a very complicated legacy estate that has undergone a lot of change as part of this holding company to operating company model. And layered in on top of that, we have a fairly aggressively changing external landscape that has really been kicked off by the changing requirements of consumers in terms of how we interact with consumers.

So it has really driven, with a sense of urgency, in particular over the last 18 months or so, a different way of working and a different way of approaching how technology is being provisioned into the business to support the business in this very complicated landscape.

I'll say it again: we're kind of at the middle of the beginning, and I did warn Gene that we don't have a lot of the sexy statistics yet. We're making them, as we speak even. But we do have two case studies that we wanted to share with you that can give some insight into where we are in the journey, how we're thinking about it, and then also to talk about some of the headwinds that the DevOps transformation itself has faced inside the business.

So for the first one, over to you.

Sarah Chong

Yeah. I get to selfishly talk about my own journey now, which is fantastic.

DevOps has been a bit of a buzzword for a while, but I feel like when my journey started, I was like, "I think we're just making a change, aren't we? We're just probably thinking about things differently." But then I realized it wasn't just me coming along for this journey either, and I wouldn't say it's even just other people in IT.

So we had the ops guys thinking, "Oh, do you know what? Maybe we should come along for the journey pretty early on as well, actually. If you involve us in the story up front, we know what's coming." And it was like, okay, winning, number one.

But then also other parts of the business. So you've got the finance guys thinking about, how do you make the business case? The vendor guys thinking, "Maybe we need to contract differently. If you're going to be building at speed, how do we not calculate in months and years?"

So lots of people coming together, and number one was great because we had a shared goal. That's DevOps 101, starting there, all aligning to the same place. So what we wanted to do was simplify the processes and the day-to-day activities of our marketers so that they can go on to be fantastic marketers and really creative, so that Unilever can be the best marketing company in the world and for the world.

We're not ambitious at all. Just a little goal.

What was fantastic about that is that I got to start part of that journey. So we were looking for, how do we find a technology platform where you can do configured and custom applications at speed? Thought, "Salesforce. That's quite a fun one."

So we started building different applications on there, and the first one that I'm going to tell you about is one that drove product launches for marketing. And we had these claims of them going, "My stuff. Where is my stuff? I can't find it, and it's never on time." How do we change that? How do we deliver our product launches on time and in full?

And we've moved from a paper-based system to digital. So, Kjersten, I know what you're going to ask. You're going to ask me if there are any metrics that I can prove that this was a fantastic app. But unfortunately, we didn't track the paper, but we did move to a fantastic digital. So maybe if we come back next year, we'll have something to say about this fantastic app.

Kjersten Moody

I can't help it. I'm a VP. I need targets.

Sarah Chong

And you need information and data, right?

So what happened here was that we've gone from a tool that was written down on a, well, started off as obviously a blank sheet of paper, and then in 12 weeks we've got a go live, which is fantastic. As I said before, we stopped measuring in years and months, and we went down to weeks, which is kind of unheard of, and something to be very proud of to be part of that journey.

As our confidence grew, we realized that actually having a technology that enables you to dabble in different bits of DevOps was really key to starting this journey, actually. Because you're not always going to want to do it in the purest form. You're going to say, "Okay, maybe I need to find an internal process that's going to help me deliver these apps quicker." Because you've probably got a very long, medieval-style scrolling list of things that are on your service criteria before you can go live. We've all been there.

But actually, how do we challenge that? How do we take what's good and adapt it? And then also tools. So you've got all of these great DevOps tools. They have to fit the technology, and they have to fit your people. If the people don't want to use them, you can try and make them do automated deployments and automated testing, but it's probably not going to work.

And then the last piece is around the culture change. These 12 weeks were certainly around failing fast, but learn even quicker. And so that's a big part of it.

And I guess the middle bit here is adopt and adapt industry practices. We have a great trend of making things fantastic for Unilever. Actually, we want to be able to talk to you guys and you say, "That's industry best practice. That's great. You've not made it weird or blue. You actually have just adopted what we can have a conversation with you about."

And then finally, what I've seen, not only on the Salesforce platform, but actually around Unilever in its entirety across IT and various parts of the business, is that the number of applications that have been deployed has grown exponentially in the last 18 months.

We've got teams that are showcasing Chaos Monkey and experimenting with just testing different things that are out there. So really great success stories, which is part of my job, which is great, and I get to tell everybody else. But Kjersten's got another fantastic learning story as well.

Kjersten Moody

Okay. This is actually one of the applications that sits in my portfolio.

A little bit about the application. It's an app called One View, and it is the global information distribution system for Unilever. So we have about 30,000 end users globally who connect to this, from our board-level CEO down through 30,000 people in that pyramid.

It has a very interesting history, actually. So it started off as this incubated idea inside of one of the countries in Europe. And while it was in this incubation stage, it was actually delivered with agile principles. It was built in AWS. It was the rapid prototyping, everything that we like to think about.

And then the word started to spread, and somebody had the idea about three or so years ago that really we should make this a global application, and we should launch this globally. So it came formally into the technology team, and it got waterfall-ized. It went on-premise.

The requirements were this odd way of generating requirements based on who was screaming the loudest. So there really wasn't much of a priority around it. We didn't really understand what we were doing. We were just making the pain go away based on some executive somewhere screaming at us for this or that need.

We had six-month release cycles, and that's actually a very generous metric because we did release every six months, but it could take up to 18 months to get your requirement actually into production based on how it fit in this cycle of pain.

And it created this culture of a lot of very senior people, because it is very much heavily used by our VIP audience internally. Because it was so hard to move from request into an actual feature function inside the application, we would just be getting laundry lists of stuff, really, that we were asked to provision into the app. So it's kind of, give me everything because I really only have one shot where you will listen to me.

And it was very much a technology-driven design. It was, well, we are using these technologies. It was actually SharePoint on-prem, and if you couldn't do it with SharePoint 2000, whatever it was, 37, then you're pretty much out of luck for how to build that.

And so, working with the evangelist that I have here to my right and saying, "Okay, we actually need to hit the reset button on this and make this, because this is one of the key applications that the senior-most people at Unilever use every day. Our CEO is in this every day." And this is where we were, being on anywhere from a six- to 18-month response time to that audience.

So we said, "All right, well, we really need to hit the reset button here, so let's do it." And the first thing that we did is we fired our vendors. We had a few different consulting partners that were in there from major institutions. I won't name them out of respect for them. And we went, and we hired a very small firm, actually, here in London, whose DNA was app development, web development, agile.

And we said, "Okay, we don't want you to be like us. We want to be like you." So we kind of want this reverse infection of your way of working, your DNA, into the way that we manage this particular product.

We hit the reset button in terms of going back to agile methodology. We were very careful not just to reverse history, to go back to where we had been before it came into the technology group, but really to benefit from the maturity that has happened in the industry with this. So it was improved tooling, whether it's from how we're managing the backlog to the source code repositories, to the deployment tools, to the monitoring tools, and so forth.

So there was a lot of upfront thinking about how are we setting this up as a product with standard tools and platforms underneath it. And it's not just bringing in the tools, but it's actually configuring the tools to work for the product. So you can bring in a tool, but if you don't have the right template inside of it, if you haven't thought about the workflow of how code and people need to interact and flow through the system, then you're sort of not really doing it, I would argue.

We moved from on-premise into the cloud. I think that's fairly self-evident. And this was interesting, because the interfaces that we had in terms of going from on-prem into the cloud kind of got pulled into the product program. So the network team needing to have the right connectivity between all of our various international sites and that cloud deployment, just kind of these things to make it work from the IT group got pulled into the program, kicking and screaming in some cases, but ultimately for the benefit of the company.

And then finally we did something that was, I think, probably the most controversial: that product owner role, we took it inside of the technology group. Because we had 30,000 people. And we said, "Well, we hear you, Mr. Screaming SVP of Whatever Whatever. But we have other SVPs Whatever Whatever who are screaming about different things over here."

So we need to take inspiration from the software industry and have that product owner really be like the product manager for this piece of software or for this product. And we will use a fact-based way of prioritizing, and we will communicate what that queue is.

So it's not that technology is saying, "No, you can't do this, you can't have it." It's just to say, "Look, here's a very balanced view of what this roadmap for this product now is, and here's how it has been prioritized using some attempt at being empirical as well as pragmatic about how we manage that priority."

This was actually one of the first times that we have engaged with the broader Unilever business as a technology product manager, sequencing the backlog and the backlog grooming for a product.

And then that got us into MVP. Now, describing minimum viable products to people who've never heard it before was kind of interesting, because people were thinking it's the minimum product that's not viable, sort of how it got translated into people's heads.

And it's like, well, hang on a second. No. What we really want to do is we want to say, "I want to give you exactly what you need and nothing else, because anything else would be waste. And why would we waste? And I understand that your needs are going to be changing rapidly over time, which means that we're going to have a team that is kind of constantly working on expanding what that MVP is.

"And when we hit a stopping point, we stop. So that way we're not just doing work for the sake of work. We're always partnering that value, if you want to think about it that way, in how we are managing the scope or the backlog that we have."

And that was a big culture shift. That was a really big culture shift. And I think it's worked very well here, and it's starting to spread throughout the business in different ways.

I love it when I hear people in the hallway now talking about MVP. Like, "Is this MVP? I don't think so. I think this is more than M." That's absolutely fabulous, because it means we're starting to see that cultural change, that cultural infection of these kinds of principles.

So we release now on monthly cycles. I will say that this is not yet fully launched into production, so we have a big go live actually in August. As I said, middle of the beginning. But we think that we will be on the one-month cycle very easily, and that is the rate of change that our end users can absorb. So we could certainly go faster, but because we have such a senior audience, probably about a month is the rate of change that they can realistically absorb. So that's really where the one month came from.

But we have automated deployment scripts, so we can take it from scratch inside the cloud to a fully functioning application in just under an hour. We've tested that. All of our QA is now fully deployed and fully automated. So our ability as a technology team to introduce change into that environment is actually very fast.

And I think this is something that I'm very proud of, actually. It's that technology-driven design, or kind of Microsoft-stack-driven design. Nothing against Microsoft, but it was kind of, well, if they don't do it, we can't do it. So no talking. It's just sort of awkward, actually.

And we've taken it really into a human-centric design. So we've said, "Okay, I'm not doing it for me. I'm doing it for you. This isn't for my health, it's for your benefit." And by changing that way of working with the rapid prototyping, the MVP type of discussions, the backlog grooming, we're actually able to bring the business user much more into the center of what it is that we are doing in that application.

And we've gone from the reputation of being kind of horrible, a team that you have to put up with, into a team that is actually empathetic to the business needs, to the end-user needs, and is able now to respond in a much more reasonable way when those needs arise.

So here we go, right? Middle of the beginning. And for us, it took what at the time I didn't quite realize was considered to be kind of a gutsy call, to totally hit the reset button, to re-architect with the technology, to reset the ways of working, to be kind of tough on the business stakeholders to say, "Okay, here's how we're going to work with you, and it's a handshake. Here's how you need to work with us. If we are going to deliver differently, you need to receive differently. And it's for your benefit, and you have to trust me on this."

And we're seeing the results. It's absolutely brilliant. Total cost of ownership is coming down. So I don't have the final metrics in this yet, but the run costs for the technology have come down by 90%. So I have reduced the platform technology, let's call it data center-type run costs, by 90%. So I can say that now. And then we see the metrics, obviously, with the release cycles and this sort of MVP type of concept.

All right, so that's that. So what have we learned so far?

Sarah Chong

Yeah. Well, you touched on it a little bit there around how the conversations have changed, that IT are now, we've been invited to the party with the business, right? We're not that person you come and knock on the door afterwards and say, "We'd like you to deliver something." We're at the table, we're having the strategic conversations.

And not just that, we're talking to each other as well. So we tell the story to our colleagues, and they tell it, and it goes down, and it's not Chinese whispers. It doesn't get lost. It's the same story as it goes down. And actually, having that great communication is really what's spreading the message.

In terms of understanding, I'd say there's probably a fifth of the organization that's really nailing DevOps. But that means there's all of those, that 20% of people that I've got, that I can go, "Can you go and tell the rest, please?" And it's fantastic because they're really behind it. They're really feeling the benefits. Their business partners are starting to feel the benefits of it as well.

So there's this trickle, this story, like you used the word infection. It's kind of like it's not the end of the world, but it's the start of a new world.

And about celebrating, sorry, I keep going back to this point, but I think it's so important. You're not gonna drive a change unless people really believe there's something positive to do, and positive to come out of it as well. So not just delivering at speed. Things might cost more at the beginning. We've all learnt that. We read the books, that you need to invest in DevOps, but there are cost savings to be made further down the line.

And then also the relationships. That's why we're here. We're here to make some new relationships outside. But I wanna make sure that the relationships are happening internally as well. So lots of different pockets. We're very globally spread. There's lots of great things happening in different parts of the world, and we need to bring them all together.

Kjersten Moody

Absolutely. So where are we going from here? This is sort of a very optimistic slide, isn't it, now that I look at it more carefully.

Sarah Chong

Sunshine and rain.

Kjersten Moody

So as I look across the portfolio, One View is one of the key products in my portfolio, but ultimately I have a portfolio of many, many different products. And...

Are you taking a picture?

Sarah Chong

Yeah. They keep taking pictures this way. I wanna take pictures this way.

Kjersten Moody

All right.

Sarah Chong

This is a crowd.

Kjersten Moody

All righty. Bless you.

So, it's really, I think, our next wave of, let's call it barriers, challenges, opportunities, I use these words synonymously actually, that we're working through are in the interfaces. So the applications that have had the most success are ones that are somewhat ring-fenced in a way from this very complicated landscape that we exist in. And as it then starts to spread out, you hit things like our SAP systems.

All right? Nothing wrong with SAP, but if that goes down...

Sarah Chong

You're absolutely welcome.

Kjersten Moody

I said that because you were giving me the look.

No, in all honesty, it's the lifeblood of Unilever as a manufacturer. So if SAP goes down, we don't ship product. Right? We don't ship product, we're not selling. Right? That's huge. It is sort of at the heart of the business from a technology standpoint.

And it is a very, very, very well-protected environment that has a lot of very robust checks and balances with procedures and paperwork and so forth. And it's all for good historical reasons. But all of a sudden now, you have kind of some sparks flying as you get closer and closer to that particular interface.

So I think one of the challenges that we will have is, what do you do? Where do we draw the line in terms of in toto, how far do we want to push these principles? Because the islands are doing very well.

Sarah Chong

Mm-hmm.

Kjersten Moody

But ultimately they need to connect into the bigger picture. And we certainly have that ambition, we just haven't necessarily figured it out. So I think interfaces between the agile DevOps world and a more traditional waterfall world is something that we're looking for an answer for and, quite frankly, haven't found it. And we haven't found external case studies that might help give us some guidance there. So if anybody has any ideas, let me know.

And then I think the second big thing that we're working through now is how do we buy, right? Our legal group loves milestone-based contracts, because you can hold people accountable, right? There's clear accountability inside of the legal language for, we give you money, you give us this, versus we give you money and then we'll figure it out as we go. That's a little hard to translate. It's a great way of working, but it's a bit harder to translate into the formal relationships that we need to have externally with suppliers.

So, it's not to say that it's an unsolvable problem, but as I sit atop a lot of contracts, and I have a new way of working that I would like to be introducing into the team for delivery, there certainly are these other axes that we need to be looking at.

Anyway, but the best thing about Unilever is that we see these barriers. We put our institutional might behind it, we will figure out a way to cross it. So I have no doubt that if Gene invites us back next year, that we'll probably be somewhere in the middle of the middle and have solved some of these problems.

So anyway, thank you. Thank you very much.

Sarah Chong

Thank you.