fbpx

Retention and Engagement lessons learned building Atlassian’s growth team from 0 to 50 people strong 💪

Shaun Clowes | Chief Product Officer at Metromile & Previous Head of Growth at Atlassian

  • | Activation | Engagement | Growth | Metrics | Onboarding | Retention
  • May 2019
  • EP8

Moments like these...

Setup > Aha > Habit

In today’s episode, I chatted to Shaun Clowes, the Chief Product Officer at Metromile and previous Head of Growth at Atlassian.

We talked about how Atlassian measured success in the early days and their relentless focus on delivering value.

We also touched on why enterprise users are more similar to their SMB counterparts than you think, Atlassian’s land and grab strategy that fueled their expansion revenue and the differences in tackling churn between B2B and B2C companies.

This was one of my favorite interviews to date, I hope you enjoy the show.


Highlights

Time
How Atlassian measured success for their growth team in the early days 00:05:25
Why enterprise users are more similar to their SMB counterparts than you think 00:08:00
The land and grab strategy that fueled expansion revenue 00:09:30
How Atlassian discovered their users biggest indicator to buy 00:12:30
Defining the setup, aha, and habit moments 00:22:30
Finding the right time to prompt user invites and why most of them fail 00:28:00
Differences tackling churn in a B2B company vs a B2C company 00:36:00

Don’t miss our weekly episodes

A new episode every Wednesday. Subscribe to receive the latest episodes and exclusive resources

View Churn.fm's Privacy Policy

Leave a Reply

Your email address will not be published. Required fields are marked *

Shaun Clowes

Chief Product Officer at Metromile & Previous Head of Growth at Atlassian

Shaun’s recommended resources on churn

About the podcast

My name is Andrew Michael and I started CHURN.FM, as I was tired of hearing stories about some magical silver bullet that solved churn for company X.

In the real world tackling churn and increasing retention is one of the hardest problems a subscription business faces.

In this podcast, you will hear from founders and subscription economy pros who are taking a systematic approach to increase retention and engagement within their organizations.

Transcription

Andrew Michael:
Hey Shaun, welcome to the show.

Shaun Clowes:
Hey Andrew. Great to be here.

Andrew Michael:
Yeah, thanks very much. It’s exciting to have you today. As you know … I don’t know if the listeners know as well, but you were one of the early people to actually start growth at Atlassian, starting as one of the first people in the team there and [inaudible 00:01:06] and then growing it out over the six years to over 50 people.

Andrew Michael:
I wanted to just maybe start as well … and obviously I think you’ve progressed and you’ve moved onto new roles and you are now at Metromile, but for the beginning of this I wanted to actually just start and get touched on that growth team. What was the motivations at Atlassian to start the growth team? When did you decide you actually needed one? And maybe as well you can give us a brief intro into what Atlassian is and what it does.

Shaun Clowes:
Sure, yeah. So that’s right, over about six years I built up the growth team at Atlassian. For those of you listeners who don’t know about Atlassian, Atlassian is a software company that makes collaboration tools. They’re most famous for a tool called Jira, which is like an issue’s tracker that’s used for all different types of tracking, but primarily software development. They’re also famous for other products like Confluence, Bit Bucket and other tools like that.

Shaun Clowes:
Yeah, in a bout late 2012 actually, I guess the origin story of the growth team at Atlassian is that Mike Cannon-Brookes, one of the two CEOs, had been observing the emergence of a thing which I guess the Valley was calling growth hacking. He was like, “Okay, well here are these guys who are out there using these interesting tweaks to products that appear to be able to massively boost their growth.”

Shaun Clowes:
The bit that was interesting about that was not the hacking bit, finding random channels that you could spam for acquisition purposes, that wasn’t the bit that was interesting to us. What was interesting to us was the realization that such little things can prevent users from successful capturing value from software. Similarly, such little things can prevent software from effectively spreading from person to person where those people could genuinely get value from it.

Shaun Clowes:
Here Atlassian was this already very successful software company back then. We were probably doing about 100 million in revenue would be my guess at that time, so we were already doing pretty well for ourselves, but Mike was like, “Well, I wonder if this could be advantageous for us even though we’re not building a consumer product, even though we’re not building a product that’s B2C. Could there be a way which we could leverage this inside B2B or enterprise?”

Shaun Clowes:
I think that Atlassian was particularly fertile ground for that type of idea because, as some as your listeners may know, but Atlassian also does not have a sales force. So even thought it’s in B2B and enterprise software, it didn’t have a sales force. So very specifically for us it was a clear and present problem to make certain that the software captured as much value as possible and it landed as well as possible with as large an audience as possible.

Shaun Clowes:
Basically, Mike was like, “Well, I wonder if this thing has value to us.” And then he ended up wanting to start something, and so he picked me to lead that thing. We ended up with a team of five of us I think we were at the start. And then, as you said, over a period of five years we grew out to be a significant division, about 50 strong, with design product management, product analytics, voice of the customer and a bunch of other different sub-departments inside the growth team.

Andrew Michael:
Wow. And I love as well that mentality and the thought process at Atlassian. Never looking for the short-term growth hacks, it’s always about delivering value and how do you keep on delivering value to customers?

Andrew Michael:
In the early days when you were setting-up, just five of you, how did you go about actually defining what success looked like for the team? How did you see this evolve over time as you grew into a bigger team?

Shaun Clowes:
Yeah, that’s a good one. When we first started out, our actual approach was just to prove that we could add value because when you’re in a very big, successful software company, there can be this, I guess, perception that these small tweaks to this big complex software that’s been worked on for many years might not have very much value, and why bother with the small stuff when there’s other big stuff to go be tackled?

Shaun Clowes:
So, when we first started, our actual goal was just to prove that we could meaningfully inflect the business. We kind of zeroed straight in on conversion. We looked at AB testing against evaluators of Jira and Confluence where we were attempting to meaningfully change the conversion rate between the groups we experimented on and the groups we did not experiment on, or the control groups.

Shaun Clowes:
From there, we just wanted to prove that those differences would be worth more than we cost to run as a team and obviously actually change Atlassian’s business in some way. That’s a pretty good place to start I think, because basically we were kind of picking some of the low hanging fruit at the start anyway, so we would expect to find some big wins and some ones that had meaningful, direct financial impact.

Shaun Clowes:
Over time, our thinking became a lot more nuanced. So, instead of simply looking at the world in terms of, “Well, okay, how much revenue can we convert over a specific time period?” We started to look much more at, “What other different phases that users go through, where do they drop off, why do they drop off?” And we started to view the world instead of being specifically about revenue, it was actually about how could we touch as many users as possible and make those users as happy as possible? The net result of that on the other side would be revenue, but you have to solve the first principle. You have to solve it from first principles in terms of actually delivering value to real users.

Shaun Clowes:
And so we started to look at activation, onboarding, cross cell, the sales funnel and all the other different kinds of ways in which a product grows, expands and monetizes.

Andrew Michael:
Yeah, [inaudible 00:07:14]. I think as you evolved over time as well and you got smarter, you started to get this insight into really focusing on the value side of things.

Andrew Michael:
You mentioned activation being one of those areas where you started to focus on. One of the questions I had in my mind is, Atlassian do serve quite a wide user base and you have a lot of [inaudible 00:07:32], but you also on the flip side have enterprise. How did you go about activation when you were thinking about these customers [inaudible 00:07:40] and these different personas? In the beginning, did you just treat them as one single persona and work on an activation flow? Over time, how did that evolve as you started learning more about your customers?

Shaun Clowes:
Yeah, I think what we learnt over time was the enterprise users and [inaudible 00:07:59] users were more similar than they were different. What I mean by that is that kind of the journey of a thousand miles starts with a single step, and that single step even for Jira inside a large enterprise is usually somebody somewhere goes, “We have a problem with our tracking. We need a new tracker,” and they form the genesis of a new landing spot for Jira. That human, whether or not they’re in a very large enterprise or a small enterprise, needs to take this software, they need to make it able to do something that others within the organization would find valuable, and then they share it with others inside the organization. Those people need to be convinced too. And then when you have a team that is successful within an organization, then it becomes much easier because then you spread from team to team, and then you go from team to team to department to department, and eventually, you have a very significant footprint within an organization.

Shaun Clowes:
That was actually how Atlassian just always tended to think about the world in terms of land and expand, that success came from individuals and at the team level, and then the rest would then fall from there. And it is true that at times Jira, for example, might have been driven tops down, but what is somewhat obvious is that our product that is very, very good at landing with individuals and empowering teams and making teams better, that type of product can land in [inaudible 00:09:21], but it’s strengths are not useless within the enterprise. In the end, even in the enterprise when it’s been driven tops down, if people hate the software, if they resent it, if they don’t feel like it actually accelerates their process and their ability to get things done, then they will eventually find ways to usurp it and try and get rid of it and bring in other competitive products.

Shaun Clowes:
It’s almost like products that are good bottoms up are also generally very good in a tops down environment, because they just don’t demonstrate necessarily weaknesses. I think what is driving that is effectively what used to be, in the old days, there was a buyer persona and a user persona. The buyer persona was the one that truly mattered because you were using a sales force to drive in software from the top at high ticket value, and so you cared about the buyer persona and box ticking more than you necessarily cared about the user persona, the ones who would eventually use it.

Shaun Clowes:
I think what’s just happened is [inaudible 00:10:20] and the modern Web 2.0 and everything else have just completely turned that around to the point that now the user persona is the most important persona, and the buyer person is actually the second most important.

Andrew Michael:
Yeah, that’s very interesting. Talking about that just a little bit more, the persona side of things. When it came to activation itself, were you treating them differently than the user persona versus the buyer persona itself, Atlassian[inaudible 00:10:49], or did you have these insights from the get-go and said, “Okay, when we look at retention and engagement overall, we really want to be looking at a user level and not at this buyer person level”?

Shaun Clowes:
Yeah, I mean I think from pretty early on at Atlassian it was clear that in order to win we were going to have to build a product that landed very well with the developer, with the developer customer base that were the primary users. To some degree, we had a natural tilt in that direction.

Shaun Clowes:
On the other hand, we also observed it happen. What I mean by that is that in the early days of Atlassian, there were certainly more [inaudible 00:11:31] attempts to get the software into certain organizations, but then we also observed at the same time we would just randomly get these large orders from customers. It would turn out that what had happened was they had started buying one or two copies of Jira just on a corporate credit card somewhere inside the organization, and over time that had expanded and expanded and expanded to the point that it became a significant enterprise order. It was almost like the market was telling us what was important.

Shaun Clowes:
Thinking about that from an analytical perspective what we noticed, and one way that reinforced that, was very early on in the growth team we were basically creating the analytical platform for Atlassian. So remember this is back in 2012, so analytics was not as sophisticated as it is today. We were building an analytics platform, and we were looking at various data and we’re trying to understand, “Okay, where is it that people fail?”

Shaun Clowes:
One of the first reports that I ended up pulling was one that basically looked at the first week of a Jira installation. I was looking at activity in the first week of a Jira installment, and I think that our internal belief in the organization was that even though we were bottoms up, in the end, because it was a complex piece of software, we were expecting that what would happen in an evaluation was that somebody would download it and install it or enable a web version of it, and they would probably have a check list of things they were looking for. Things that were must haves, things that were nice to haves and things that were deal breakers, and they would probably go through some sort of check list or criteria evaluation of the software and then make decisions.

Shaun Clowes:
We expected to see that in the month long evaluation of Jira. We expected to see some usage happening in week one and week two and different types of features in week two or week three or week four as people were learning and deciding whether or not it could do what they wanted it to do. And actually, it may sound funny these days, but the data didn’t show that at all. The data showed completely the opposite which is that even for a complex piece of software that was to a relatively technical audience, there was no evidence that people did that, that they showed up with a check list and that they gave you multiple days and they thought about you and they evaluated you.

Shaun Clowes:
What the data showed was that in general, the decision was made to buy Jira within 30 minutes. What I mean by that is the analytics showed that in general there would be up to … basically, if you did a histogram of activity inside of Jira, you would see that the vast majority of Jira’s that did not purchase spent less than 30 minutes doing anything. They were not in the software for more than 30 minutes. The vast majority of the ones that did convert spent more than 30 minutes. And if you did a histogram, it was actually really stark that there was this perfectly bi-modal distribution.

Shaun Clowes:
So, even if we had believed that there would be an enterprise evaluation, that there would be an enterprise sales cycle and everything else, the data just didn’t bear it up.

Andrew Michael:
That’s super interesting. Talk to us a little bit about how you went and how you setup the histogram itself to figure this out.

Shaun Clowes:
Yeah. So basically all I did was I took all the product analytics from inside the product, and it was pretty well instrumented. Long explanation about how and why. It was relatively well instrumented in terms of surface area, and so basically I took any given event, and then I considered the instance as being used, as being active, for a five minute period after any of them had been triggered. Events were triggered by humans, and then I gave everybody like a five minute trailing usage sum based on every event that they triggered. So if I triggered two events in ten seconds, then that would be five minutes and ten seconds of active usage. So, then I just histogrammed it from there in terms of total active usage by converted and non-converted instances or installations of Jira.

Shaun Clowes:
What happened was it was really obvious, this bi-modal distribution was super obvious the first time I ran this query. I was kind of shocked myself. I showed it to one of the CEOs and his first response was, “This data is kind of unbelievable.” He was like, “I think there might be a bug in the data.”

Shaun Clowes:
I think in hindsight, if someone is asking if there’s a bug in the data, you usually know you’re onto something. Either that or there is a bug in the data. This data was so surprising to us that it was basically, “Is there a bug in the data?”

Shaun Clowes:
So, I went and ran the exact same analysis, but this time I did it from the access logs. So basically used every request into the Jira instances, and I gave them a five minute count up or whatever as I described before. It was exactly the same basically. It was very, very clear that a Jira that was not going to purchase spent less than 30 minutes playing with the software, which is kind of incredible for a piece of software that at that time had already been being developed for like eight years, nine years. It was an incredibly powerful piece of software, and yet people gave it less than 30 minutes before they made their purchase decision.

Andrew Michael:
Yeah. That’s fantastic. And then from there did you sort of try and dive into what that usage looked like of people over 30 minutes and did you find any patterns in that data? Or was it just really key insight that they needed to be using it for more than 30 minutes the first time?

Shaun Clowes:
Yeah, I mean the 30 minutes, it’s intellectually interesting but it’s not that interesting in that whenever I advise people these days in growth, I explain that time doesn’t really mean anything to an individual. What I mean by that is that what actually happens is that I start with the piece of software, and then I have to work out how to set it up to get it to do something that I care about, I have to reach my aha moment. I have to have something that makes me go, “Oh, I see that I can solve a problem that I care about with this software.” Then I have to have my habit forming moment, so I have to form a habit where I go, “Okay, every time I have this problem, which I have somewhat regularly, I immediately go to this tool.”

Shaun Clowes:
When you think about that, the reason time is not relevant is because some people will go through that process very quickly, others will go through that process very slowly. There is natural variation because people have meetings and otherwise go around and do things. The real problem is when people do not achieve those milestones at all. And so I feel like it’s a growth team’s job to truly understand what is driving those moments, like the setup moment, the aha moment and the habit forming moment, and make certain that all of those things happen to as great a degree as possible.

Shaun Clowes:
And so in Jira, for example, it turns out one way that you can measure activation in Jira is you can measure if three issues have been created. In a Jira where a project has been created and three issues have been created, you have a drastically higher chance of that instance converting and becoming a paying instance. However, that three issues doesn’t really tell you anything because the three issues is an output, not actually an outcome in terms of the process by which a user gets enough value to want to create three real issues and being using Jira in that way is what you’re attempting to incept. It’s a valuable measure, but not something you can push or force or insist that users do in a meaningful way.

Andrew Michael:
Yes, absolutely. So these insights as well, the activation and you realized the user needed to do three issues within that time, was this not something that you pulled out when you initially looked at usage across the board and tried to understand what happens within that first week? Was there this insight that they needed to create these three issues within a certain timeframe? How did you go about getting that understanding and insight? And was it a combination of looking at that initial histogram that showed you the amount of time they needed to be spending [inaudible 00:19:36]?

Shaun Clowes:
Kind of. Basically what we knew was that basically it was game over within the first week. A different way to think about that histogram was that if you had not successful gotten an instance activated in the first week, then it’s highly unlikely it will be activated at all. And so that was a useful insight, but then we needed to dig into the exact activation flow and how to quickly get instances activated in the first week.

Shaun Clowes:
And so the three issues thing flowed out of trying to get to a deeper understanding of what it meant to have successfully completed your aha moment. As we dug into more of the analytics, we kept finding these interesting correlations.

Shaun Clowes:
So for example, after you first collaborate on an issue, you share an issue with somebody else and somebody comes to that issue and does work on that issue, that is also an incredible predictor of conversion. When we saw these strong correlations in the data, we then had to fit that to a model of what it means to be activated.

Shaun Clowes:
As I said, what you’re really looking for is you’re looking for proxies that tell you that somebody has made it past setup, they’ve made it past their aha moment, and they’ve formed a habit. And so the three issues and the collaborating on an issue bits of data were found during our search for what it meant to know that an instance was activated, but all they did from there was inform a set of experiments.

Shaun Clowes:
We then setup a set of experiments. We were going, “Okay, well what is the quickest possible setup experience that we could have? What do we need to know in order to setup the instance as quickly as possible? Okay, how would we, after they had completed and setup, drive them to an aha moment? What would that aha moment be? Is it when they collaborate with a user for the first time? If so, how do we get them to do that? Okay, after they’ve done that, how do we make the instance pull them back? How do we repeat the value of that outcome when they have that need again? How do we make certain that they form a habit rather than just trying us once and getting some interesting value, but not thinking of us again and deepening their engagement with us?”

Andrew Michael:
Yeah. I find it fascinating as well, on the other side as well, Jira is quite a complicated product. When you think about as well the different use cases that you can use Jira for, there are a number of different use cases that teams could use the tool.

Andrew Michael:
How did you then go about understanding and optimizing for which actions you wanted to sort of drive when it came to you at the setup, aha and habit moments with the different use cases?

Shaun Clowes:
Yeah, that’s actually super interesting. You’re right, Jira has a large surface area in terms of features, and also a large surface area in terms of the types of users who use it. So for example, there might be a developer who has a need to understand the code changes that they made as well as get requirements clear et cetera. There might also be a product manager who cares about prioritization and ranking and sprint reporting et cetera. There might just be a management stakeholder who cares about release cadence and what’s coming in the next release. There are a bunch of different stakeholders who have different sets of needs.

Shaun Clowes:
I think there are two ways you can approach this. One is the number one thing you have to do is identify what the user might be there for and get them to some moment of feeling like this product can actually give them that as quickly as possible. And so if you’re starting in Jira and you don’t have enough information to necessarily guess who they are, then you can still work. You can still, for example, if you’re setting up Jira, there’s a bunch of painful stuff you might do to setup Jira, but there’s also the very basic stuff that has to be completed. So you would setup the fastest possible flow to get that minimal setup completed so that they can move onto some use case that might show value to them. Then you might setup some beautiful moment in terms of project creation, an issue creation, with the idea that you’re trying to build a very generic experience of what it is to setup Jira and then start tracking things and see some of the collaboration and value from there.

Shaun Clowes:
You can build a, I guess, generic but best practices activation flow. Or you can take it the other way, which is you can attempt to work out who people are, what part of persona people are, and then give them exactly what they want. So for example, you might ask the, “Well, if you’re a product manager, then here’s the minimum setup you need to do. Here is how you might do sprint management or prioritization or something inside the product.”

Shaun Clowes:
We did a lot of that. We experimented with all sorts of things including, like I said, the generic but best practices onboarding as well as very segmented onboarding. I think to the degree that segmented onboarding is usually going to work better, expect that in general we found that trying to segment people by who they said they were was not very successful. If we, for example, had an onboarding flow where we asked you, “Are you a developer? Are you a product manager? Are you an executive stakeholder?” For example, and then tried to have an onboarding flow behind that, what would happen would be that people would get frustrated. If you’re a developer, you might still be very interested in sprint reporting. You may just be a very product-oriented developer, or you might be very oriented towards source control.

Shaun Clowes:
So, they’re not a very great way of defining what you’re there for. Who you are and what you are here for or how you think about the world are not necessarily that perfectly overlapped. So, we ended up with a much simpler approach which was basically offering three available options which was, “Okay, would you like us to handhold you in starting to track some work in Jira? Are you an expert and you would just like to be left alone to explore the important pieces of the product? Or would you like to be shown what is already configured inside this Jira?” Which I guess is a very generic set of options.

Shaun Clowes:
At that time, that turned out to be the best at threading the needle between the very simplified onboarding and very handheld onboarding and the segmented onboarding.

Andrew Michael:
And just thinking about the different levels of sophistication of their users.

Shaun Clowes:
Exactly right.

Andrew Michael:
So, I wanted to touch on the next thing as well because we touched on it a little bit earlier. I think one of definitely the big successes of Jira is that land and grab model as you mentioned where expansion becomes a really critical part within an organization, as you have an individual bring it in and then it slowly starts to spread organically through the org.

Andrew Michael:
Were there new things that you tried though yourselves to facilitate that organic growth within an organization, trying to get teams to try Jira in different ways and getting them to introduce it to other team members?

Shaun Clowes:
Sure, yeah. That’s always the goal. In collaboration software, and in fact a lot of different software, there are actually these cases where the softwares interacts with non-users. Users interact with non-users using the software. Really, that’s just a viral loop as might exist in a B2C company or a social network for example. It’s super important to think about those engagement points and think about how you can convert those non-users into users.

Shaun Clowes:
So, we used to look for, for example, inside the product where could we encourage somebody to invite other users? We actually spent a lot of time looking at where exactly invite moments happen, where there is a natural invite moment. And so it turns out, for example, in Jira, there’s a natural invite moment after you have created a sprint and you’ve put work in the sprint and you’re ready to commence work. You naturally have this moment where you want to invite the people who might care about the work that is in that sprint, or be the people building the work that is in that sprint.

Shaun Clowes:
We actually experimented with, for example, invite dialogues and invite flows in many, many different parts of the [inaudible 00:28:11]. Most of them failed, except when they happened to be at that natural collaboration point. That one really, really worked.

Shaun Clowes:
We had another one where we experimented with setting it up so people would share the issues, for example, via email. Share the work that was being conducted over email, and then with the point being that some of those people who will read the email will then click on the links and that will take them into the product. And then we set it up so that if you weren’t a user, you could request an invite into the software.

Shaun Clowes:
Effectively the non-user gets exposed to some of the content in the software, then they click through to see some of that content. They can only access some of it in that way. Then they are invited or they request an invite, and then they complete the loop by coming back into the software.

Andrew Michael:
I love that. I think so often the user invites are something that’s done during sign-up and then it’s typically forgotten about afterwards, but just hearing you say … it makes total sense as well … really triggering those moments when you need to encourage the collaboration because they’ve started to get that value, they’ve set things up and now it’s time to work with people.

Shaun Clowes:
Yeah, just quickly on that one, I think it’s really interesting because when you think about the emotional state that a user is in when they’re first signing up for a piece of software, they’re usually in a state of disbelief in terms of they would like your software to be able to help with the problem that they think you solve, but they’re not really sure you’re going to be able to solve their problem, or even if you could, whether or not they would be able to successfully be able to make your software successful inside the organization they’re part of.

Shaun Clowes:
What’s ironic to me about invite flows being very, very early on is that one of the highest political risks or emotional risks you can subject yourself to is inviting your colleagues into a piece of software. When you do that, you’re actually effectively staking your political reputation, or some of your political capital on this piece of software. When the software’s unknown, when you’ve not yet demonstrated to the person that you can actually successfully solve their problems, it’s a very [inaudible 00:30:21] thing. It’s exactly backwards, right?

Shaun Clowes:
Whereas, when you think about having successfully shown them some value, that’s the moment they’re like, “Oh, this thing’s really cool. I want to share it with somebody.” So how do you get in front of them at that moment when they know that you can solve some of their problems, and that their colleagues will be impressed?

Andrew Michael:
Absolutely. This is what I think goes a little bit then against, and a subject I want to touch as well is about word of mouth. I think Jira as well, or Atlassian as whole, is one of those things where 80% of incoming sales come from word of mouth.

Andrew Michael:
I want to see, did you ever do any sort of research and diving into trying to understand the type of quality of traffic that was coming? Did you see any variances when it came to word of mouth traffic?

Shaun Clowes:
Yeah. About 80% of all of our incoming traffic was unattributable in terms of it was basically word of mouth. Yeah, I mean I guess what we observed … and perhaps this is the opposite of some other companies out there … which is that the unattributed traffic had a tendency to drastically outperform the attributed traffic, certainly anything to do with branded search, unbranded search, search engine marketing. It just outperformed everything. It was easily the best performing traffic.

Shaun Clowes:
I guess the way I intuit that is that the whole land of expand strategy was always about getting to some user somewhere in an organization, giving them a bunch of value, like way more value than you cost, drastically more value than you cost, to the point that it made perfect sense for them to spread it within their team and then from team to team. And then eventually, you just get this incredible word of mouth machine because you have a lot of users who love what you are doing and you are so inexpensive and easy to spread that you naturally spread.

Shaun Clowes:
When you think about that from Atlassian’s perspective, the way that manifests itself is this unattributed traffic because what has actually happened is that somebody else who we did know in some company has seeded the software inside that company, but now another team inside that company has seen value here, and they have just [inaudible 00:32:50] a new one entirely independent of the first Jira. That’s why so much of our business ended up coming in this expansion [inaudible 00:32:58].

Shaun Clowes:
We used to talk a lot about, inside Atlassian, we talk about new to existing and new to new. So basically, we would look at expansion revenue, inside customers and then we’d look at new to new. New to new was basically clients who we’d never seen before, or we wouldn’t identify that we’d ever seen before, and we were really, really focused on making certain that both were amazing. Certainly we did not want to just expand revenue from existing clients, so we cared very much about negative [inaudible 00:33:32] et cetera in those groups. We never mixed the two types of revenue together, because we needed to take the new to new revenue and keep it and to make those people successful, and then they would become the new to existing revenue pool.

Shaun Clowes:
We tended to never talk about those two metrics together. We always kept them separate so we didn’t convince ourselves that our success was being driven by revenue expansion and [inaudible 00:33:56] successful customers. We still had to be landing inside new teams and new companies in order to continue to build a massive word of mouth business.

Andrew Michael:
Very interesting. What was the main motivation to go about it and report it in that manner? You mentioned obviously not kidding yourselves as well, but what led to that insight to say, “Okay, let’s break it down this way”?

Shaun Clowes:
Yeah, I mean I think the main reason was that we could see that we were starting to [inaudible 00:34:27] bigger and bigger ticket deals. That was happening because we were seeing massive expansion inside logos we were already successful inside. When you consider that Atlassian’s pricing structure is setup so that the opening price is so, so low that it deliberately makes it easier for new clients to use the software, or to commence with it.

Shaun Clowes:
We recognized that the new to new revenue, no matter how successful we were, would always be dwarfed by the new to existing. The thing is, if we merge those two metrics and we talked about them together, then the whole business would just naturally become oriented towards the bigger number, and the whole business would naturally become motivated to drive up the new to existing number. But that would just be sealing the seats of our own demise because our model is predicated on top of landing easily and landing inexpensively with new customers.

Andrew Michael:
Makes a lot of sense. I want to fast forward a little now to your current role as Chief Product Officer at Metromile. I think this is quite a big change in terms of the environment itself. For those that aren’t familiar, Metromile is a B2C company who’s disrupting the insurance space. Shaun, maybe you can as well give a little bit of a pitch on that as well.

Andrew Michael:
My question when it comes to this now is, you’ve traditionally been working a lot in the B2B space and you’ve been working on retention and engagement strategies in a B2B company. What are some of the biggest changes you’ve noticed now moving over to B2C? Specifically because Metromile is trying to disrupt insurance in a way that insurance is typically a set and forget type of thing, whereas now Metromile is trying to do an on demand … excuse me if I’m butchering this a bit Shaun, but it’s an on demand service for car insurance. How are you seeing this change now and how are you tackling retention and engagement in this business?

Shaun Clowes:
Sure. Yeah, Metromile, we make pay by the mile car insurance. Basically, if you drive less than other people, which it turns out that 65% of Americans do, if you drive less, you pay less. I guess the not so dirty secret of the insurance industry is that in general, the people that drive more have much larger losses, and so the people who drive less are fundamentally subsidizing those people.

Shaun Clowes:
On a deeper level, what we’re trying to offer is a high-tech insurance product. An insurance product that you can control your costs based on the amount of driving you do as well as a friendly app and everything else. It’s fundamentally a technology driven insurance company.

Shaun Clowes:
I guess you’re totally right, it’s a bit of a weird jump from B2B to B2C. The one thing I love about it and one of the things I was most keen to experience was that the different challenge of selling a product in a market where it’s a compulsory good for anybody who drives, people kind of purely just shop on price and they kind of resent their insurance companies. I was like, “Well, can we bring a better digital experience to this product? Can we bring an experience that offers continuing value from your insurance company, rather than a set and forget and resent, maybe, product?” Which is how most people engage with their insurance company here in the United States.

Shaun Clowes:
That’s what we’ve set out to do. One thing that’s been really interesting about that is that when you think about landing a Jira or a Confluence, it’s not easy. The reason it’s not easy is because you’ve got all these different personas who access the software and you need to make them successful and it’s quite complex and everything else.

Shaun Clowes:
However, ironically, my experience at Metromile has been that the same tools used in an environment like Jira actually work quite well in Metromile, a consumer product, because in the end, the same goal is to simplify the product so much that it is perfect. It’s perfect because it fits exactly to a consumer’s need and the way they think about the world.

Shaun Clowes:
The thing is that in the enterprise context, the B2B context, people will cut you more slack because their colleagues were really successful on this piece of software, or because they just know that you’re the winner in the industry or just because they’re developers and they’re more willing to experiment with a piece of software. I thought it was really hard to convince them and to make it simple enough for them, but it’s nothing compared to the consumer world.

Shaun Clowes:
In the consumer world, people fundamentally don’t believe you. They don’t believe anything you say, and they’re right not to because people are just always broadcasting these messages of, “Better savings. Better customer experience. Better quality.” Everyone just says the same thing. In the insurance industry, you have to take that and take it to the whole next level. How do you make your product so perfectly simple and yet good and valuable that … ? In our case for example, a retiree in Florida can understand it just as well as can an urban millennial here in San Francisco. That’s really hard when you’ve got people with such diverse starting points, truly diverse personas.

Shaun Clowes:
To double-down on that, as you mentioned, it’s like a set and forget product, so once you have sold them this insurance, how do you meaningfully help them in their life other than when they have to make a claim? Which unfortunately about 10% of insurance policy holders will probably make a claim in any year, which is a very traumatic experience. For the other 90%, how do you deliver them value so you’re not just this product that they have to pay money for but they don’t get any value from?

Shaun Clowes:
We spend a lot of time thinking about that, and we have built out a series of features that are enabled by the product we sell. As I mentioned, we sell pay per mile care insurance. What we do is you install a device in your vehicle and that device tracks your miles, and that’s how you pay us, basically pennies per mile that you drive.

Shaun Clowes:
The upside of having this device in every vehicle is that, for example, we know the locations of the vehicles. For example, if you can’t remember where you parked your car, we have a Find My Car feature in the app. If you want to budget miles or your fuel costs, you can do that inside our app. We can show you the engine codes in your car if your car has engine codes showing. In San Francisco for example, we give people street sweeping warnings, so if their car is parked on a street where there will be street sweeping, we warn them that there’s going to be street sweeping the next morning so that they don’t a get a ticket.

Andrew Michael:
Nice.

Shaun Clowes:
So, those are kind of just these consumer experience things we can do to make what is basically, normally, an infrequent use case piece of software, or infrequent use case product, and we can make it more frequent in a way that benefits the user and allows us to deliver a much better experience. A much more engaging experience.

Andrew Michael:
And keeps you in that habit zone as well of that monthly cycle and continually using [crosstalk 00:41:47] and bring that sort of mindset that you had from Jira itself there, like really focusing on delivering the best product possible to an industry that’s typically sort of just, “Let me take your money and then hopefully you won’t have to come and claim it back at some point when you need it.”

Andrew Michael:
So, I think for myself today, as for Shaun, this has been a super interesting call. Thank you so much for joining. It was really, really good to have you on the show. For the listeners out there, how can they keep in touch and how can they follow some of your content online?

Shaun Clowes:
Sure, if they’d like to follow, my Twitter handle is shaunmclowes. That’s S-H-A-U-N-M Clowes, C-L-O-W-E-S @twitter. Or you can follow me at shaunclowes.com which is my blog as well.

Andrew Michael:
Awesome. Well, thanks very much Shaun. It’s been a pleasure having you on the show today, and I wish you the best of luck now in the new role going forward.

Shaun Clowes:
It’s been great chatting Andrew, thank you.

Andrew Michael:
Thanks.