You know those conference talks when everything seems too perfect to be true? When people sell a version of their success story that is so glossy you think: ‘oh, come off it – it can’t have been that easy!’ This is not one of those. Scott Farquhar shares four crises that almost killed Atlassian.
Video & Transcript below
Atlassian today is a huge success story and its growth has been remarkable. But it hasn’t been without hiccups. Scott Farquhar, BoS’s favourite Aussie, tells the story of a few significant roadblocks, catastrophes and, err, cock-ups along the way. How can management respond? What do these crises tell you about your culture? A hugely worthwhile talk for anyone who wants a realistic view of what makes a company resilient.
Don't Miss a Thing - Get BoS Updates
Want us to let you know about new talk videos, speaker AMAs, Business of Software Conference and other event updates? Join the smart people who get BoS updates. Unsubscribe anytime. We will never sell your email address.
Scott Farquhar: Can I ask you one more round of applause for the lighting talks and it was absolutely amazing. [Applause]
I don’t know how I got roped into this talk, Mark is very convincing and as you all know this is a pretty special conference compared to I guess all the other conferences that I have been to. And this is the only conference I speak at. And Mark convinced me I said, “Oh, I wanna speak about culture.” He says, “No, no sorry Dharmesh does that every year.” [Laughter] So, you can’t speak about that. And I went through all the list of all things to speak about and we had a few drinks and I woke up the next morning and suddenly I’m speaking about all the dirty laundry that we did at Atlassian. So, today I wanna go through some of the things that I wouldn’t know we talk about in a conference and some of the mistakes we have made, I know one of the best parts of conference is when they hand up and say, “Here’re the things that I did and they went wrong and that’s why we make a lot of the learning from.” So, we go through some of the mistakes that we have made. Why am I qualified, I guess?
I started the Atlassian about 10 years ago, I’m the Co-Founder and about 10 years ago two of us started Atlassian on a credit card about $10,000 with the credit card debt and we are since growing that from the two of us to between 700 and 800 people today. And we sell to about 30,000 companies in 130 countries around the world. And I guess we have seen every stage software company growth, we have seen the first sale you know, first one the code that’s get written, you are hiring your first support person, realizing they are not just a developer and these guys are actually different type of persons, international offices and we have seen the whole gamete of things. So, we spent some time at the end for QA. So, any questions you have across sort of anything to do with software over the time and the thing I wanna point out I guess that if we have got to this sort of scale we added the 30,000, we added the 6,800 customers last year. And that’s adding a customer with 77 minutes for the entire year with a new customer. And we are growing about 30% a year. So, every week, this year we are selling about a million dollars more than we sold the same week a year ago. So, we sort of got to a reasonable size and scale now and that brings a whole bunch of issues.
And when I ask of the audience who knows who we are Atlassian? Well, keep your hand up if you are a customer. So, we have 30% of the room. So, just what we do, we believe that software development is a collaborative problem. So, it used to be that Compliers and Ruby, and Java, and that was where you go productivity increases but these days it’s actually a problem of getting people to work together is getting UX design as to work with the developers and with the 3D support. And we will often focus on what’s gonna change in 10 years? You will ask the questions? So, what’s gonna be different and how do we get hand in that way? And Jeff Bezos says is another question that people don’t ask which is, “What’s gonna be the same in 10 year? What’s not gonna change?” And at Atlassian we believe that what’s not gonna change in 10 years is still people gonna have to get together and they solve software problems whether it’s in Ruby or Scala, or Node, or anything else, people still have to come together to build software.
So, I wanna talk through four crises that we had at Atlassian and some of these crises may seems small in retrospect but they were huge at the time. Some of them we managed to knit the bug when they were small. So, there are small files that could go out of control and we manage to solve it and some of them are still painful memories. And there are few that I can’t go through. So, I might be catch up with you later, there’re few things we can’t put on camera but I wanna to go through today just some of the lessons we have leant and overall show people that all crises are solvable and you can still grow to large company even if you fuck up a few things along the way.
So, the first crises we had at Atlassian were ‘Innovation’.
So, when you start a company you know, you almost innovative by default because you are a Founder and you say, “I’m gonna solve some big problem that hasn’t been solved before. I’m gonna invent something, I’m gonna solve customer problems in a way that’s never been done before. And I’ll start this business.” So, if you are successful you go from that you know, sort of a found a new innovation and then you have this memory of kiosk where you have everything sort of goes and you hire people and know what they are doing and overtime you get better of the kiosk, you have to manage it, you have to control employees, and you have structures, and you have roadmaps, and you have all the things that lead to the chaotic to more organized company. And the problem with that is you stamp-out all the things that lead to innovation in the first place.
Before innovation happens getting groups of people together in random ways and we at Atlassian found that between 2004 and 2011, we had released zero new products that we had built at Atlassian. We acquired a whole bunch of small products that we grew internally but in that time we hadn’t actually gone to customers and found another problem that we could solve for them and brought a new product to market.
I felt this is a huge problem because as a company you wanna be innovating, you wanna build new things and if you are building new things you eventually die.
I guess we set down and I credit a little across internally. So, how we are gonna get better at handling innovation?
And the three areas that we focused on that we used to get pass this innovation crises were as follows.
The first one is ‘Vision’.
We found that a lot of employees came to us and say, “How we are innovating and you know, do you like this, and that’s been weeks working you know, they come to present to me.” And I think, “No, we are not in that business of iPhone Games that doesn’t make any sense to me, why would you build that?” And then you know, they hopefully aren’t completely dispiriting because that’s been a lot of time and then they never want to innovate again because they got chastised. And after a few times doing this I started thinking, “Why do they not get it? Why do they not get exactly where this company is going?” I mean look to that purpose and this is the purpose we have built earlier as a company because you are really excited one day and we say, “This is our purpose.” We thought, “Wow! You know, useful software people lust after.”
You got the German usefulness and you got the Italian lust-worthy. We thought if we could do those two things you are gonna be successful. And what we realize and this wasn’t specific enough you know, you can build anything to do with software almost everything in one of your companies could fit into this particular vision of value statement.
So, what we did is we changed it, we set down as a management team and said, “What really get us excited about coming to work everyday? Was it about software as I mentioned it is for all of you? But the thing that really lights us up is solving problems with teams of people.”
So, when we put this vision statement together every person that spend days & nights outside of the office in their own time coming up with you know, great ideas that will solve things for Atlassian a new playground that we are operating in.
The second aspect was getting customer feedback and again when you are a small company, when you are 10-12 people, it’s really to get customer feedback because you are answering the phone, you are doing sales calls, and every person is doing everything. But as you grown and put a structure and you support people to do a great job, you have all specialization, you will find that feedback doesn’t get through the organization.
At Atlassian we always had a public bug tracker and that’s one of these servers… Anyone else in this room has a public bug tracker one person, two, three, four, okay. So, it’s very uncommon, it’s depend on a lot of ways because customers always like, “These bugs have been open for 10 years.” [Laughter]
Yep. So, all these other you know, and they never see the ones that have solved of all of the ones you haven’t solved, this is the number one. And you think, “Well, okay as soon as solve that there’ll be a different one, it’s one number that hasn’t been solved.” But anyway, but it’s good you know, a lot of ways you get this feedback from customers.
About a year ago I said in a management meeting I was sort of concerned that we were getting customer feedback. And I asked my management team who and almost 10 of us, “Who has spoken with a customer in the last week?” And a person sort of sheepishly to the person next to them and not a single person in our management team had talked to anyone, any customers in the last week. So, I said, “Well, there’s nothing really for us to discuss because we don’t know what the hell our customers are asking for?” So, it’s all break and let’s spend that this time talking to customers. We codified that into a program because when you talk to customers and ask them what they want? You know, you get a feedback and get from a bug track and what features do a bug associated with, what you really need to do is understand your customers.
We implemented a process called, ‘Contextual Enquiry’ where we actually send three or four our staff onsite with the customer to spend a half a day or a day with the customer. Contextual Enquiries: they take a lot of photos, they attend the daily stand-ups which were actually well except those one awkward that we stand up you know two people started yelling at each other about the burn down shot that was kind of awkward. But generally people welcome you into their workplaces and you can find problems you have not solving for customers because I won’t tell you that they wanna cover it you know, they want a faster horse but it’s actually when you go and see the problems they try to solve everyday and what are the tools that they use?
We take that contextual information and we use it sort of really high touch and then we can get that to over several hundred people that were there. Is there anyone who does that sort of onsite customer interviews? So, that would be about 15 to 20%. So, we always just recently did that. One twist on it, when customers are often little bit cautious to tell you what they really think and we ask them to describe at Atlassian as an animal. [Chuckle] And it’s fascinating because you get a whole gamete of things that people come back and describing the same product here. So, it’s obviously everyone has different opinions. I like the guys so, there was like a tiger but it was more friendly so, as a tiger and the dolphin. [Laughter] And I don’t know who they hell manage to Photoshop that in or found a photo with a tiger and a dolphin.
I suggest if you wanna really understand what people think of your product ask them to describe it is an animal.
Lastly we implemented a thing called ‘Persona’
Does anyone used Persona in their software development practice? Okay, it’s probably 20% as well. I thought Persona is a completely shit personally when we started because Persona is the idea as you trying to get your customer base and segment him down and say, “Okay, this person you know, we have here this is Warner the CEO at the top and the there is Jack the Project Manager and the person that get stuff done.” And then we have got William, the Wise from the top right and he the beaded guy obsessed with technology who’s kind of spends all his days & nights in coding. We have got Ahmed Aiya who’s the young person really improved himself and kind of get to the next level so, he’s super ego person. And then we got Harvey, the hesitant who’s the guy that sort of thinks he knows technology but he’s got too much to do and just wants to make it really simple and doesn’t wanna know anything.
I thought it’s just dramatic simplifications of the real world but it turns out when you got 700 people and they all need to build products. So, they may not interface with all the different customers and understand the new answers of the 17 different types of runs out there in the world. Simplifying it down to and Persona made a big difference to the way our development team talks.
So, we are building features, too many points of you know, building features that people want, we can now say actually the Warner is gonna use that feature or it’s gonna be William the Wise who’s gonna use it and explain it to Ahmed the Eager. And we have this sort of language internally. And having the language helped us build and innovate in the right direction.
Thirdly it was creating space to innovate.
What we found is that moving to the companies like Google and said, “Okay, we’ll have to innovation.” And then realize that, “Well, big companies maybe have time in their innovation budgets to have select time and take it day out a week.” And they end up building these types of things. And we felt there’s a company we didn’t need self driving cars, we just needed the next product that was gonna help existing customers that we had. So, I call I guess small companies, you think how do I take something that the big company does and do it a way more efficiently and try to get the best process.
We came over the idea of we called them FedEx days at the time and the ideas as you can build and ship something in 24 hours. So, instead of doing 20% time once a week we say, “Okay, let’s compress that down and take two days and we do them every quarter now. They take two days and get as much innovation out of people as we can.” Does anyone done Ship It or FedEx day? Good. Keep your hand up if you thought it’s worthwhile, okay. So, it’s all of them.
So, this is the number one thing I think if you haven’t done with this your work before, it’s a number one why do you get innovation happening? It’s a number one why do you re-invigorate an engineering team that maybe feels a little bit why they are phoning. We have so that the process is as you start by saying, “Here’s the FedEx day, and it’s on this date.”
If you wanna participate you have to write a shipment order. So, you have to describe what you are going to build in advance. We are just new to lot of developers and to getting Product Managers hand them specs that you have to write spec themselves. And that process because your whole bunch of energy because you have got two days in which to build something and impress your colleagues. So, that entire three month period leading up to and this people trading ideas in the whole way, they trying to form teams and it’s really energy boost to the entire company.
We then started midday Thursday and we go through the 5:00 PM on the Friday, there’re people, there’s a high correlation of people who sleep at the office, and waiting for FedEx day. Some people do with 36 hour a code stint, and the Friday at 5:00 PM we have a presentation where every person in the company that build some code or even sort of signed up to do and had to get up and present.
We find that’s really useful for developers because often they have never presented anyone in their life and having to get up and explain why the feature is useful as opposed to it’s just technically right it makes a big difference to the why the view customer features and why they build things. And we have as a result I guess this is growing from a very first FedEx day where we could all really crowd around one computer and then we could crowd into one room.
And then this is one of five heats at Atlassian for one of the most recent FedEx days. And as a result we have run 550 experiments over the period of time will know FedEx day for about six years and 47 of the shift as features in their product. Three of them have actually ended up as actually four of them now have ended up and shipping products with their own items.
So, we now ship the service desk product that you know, you can use to helpdesk and not actually started as a FedEx day when we decided to expect something else. Scott Farquhar: We are so successful that unfortunately FedEx said, “Please stop using your name.” [Laughter]
So, if you do refer in the future or on the video please its Ship it days, please nothing to do with FedEx at all.
Those were the three things for us we had a long term vision. So, your staff knows where to innovate customer feedback so that they have the loop and they can sort of understand the problems they are solving and give people time and space with FedEx days or some other way of doing it, those are the three things.
The second problem we had was ‘Scaling’.
And I say you can imagine going from two people to 700 every year, we get actually 100 people last quarter so more than one person a day. So, again it was Scaling Problems and the reason I find Scaling is important, firstly why do you need to Scale? You could just take small there’s a whole book here that took… There’s a whole book that talks about staying small you know, being right but staying small. But I think that’s a word of crap in the tech industry. I think if you even gonna get big or you become absolute because of the nature of the technology industries pretty much of winner takes all and the largest company can invest the most in R&D, and most in sales and therefore have the best product in the best sales force and that bridge on itself. So, at Atlassian we felt that we had to get become the biggest and most successful company over the period of time rather than just sort of staying as a small family, I guess a cottage industry business.
We got to the stage where a company needs to create a set of values to define the types of people they’re gonna be part of an organization as it grows. If you gonna build a company for a long time you need to setup the values. And when you are small, the company values really reflect the founders. And that’s good or bad as the kids and then the kids have these weird habits that you realize are actually your own habits that you do or don’t like. If you have a company and its small company it really reflects you and you know, if you have a culture that’s good or bad really reflects the founders. But something happens when you no longer hire everyone is suddenly you are not sitting at the tone of the people that get brought into the organization.
What we found in 2006 we doubled from 28 people to 57 and we found that some of those people that we have hired didn’t really fit out coming values. People will turning out and that was sort of phoning in and what passion about what we did? And it’s sort of didn’t really feel like they’re part of Atlassian. And then we sort of reach that tipping point where we hadn’t hired them all personally.
We went about a process of establishing core values. And hands up who in this audience has the set of core values today? So, it’s 35 to 40%.
Firstly, I’m gonna take a little test and I have shown before so, if you haven’t seen it don’t answer but scream out if you know which company values these are?
- Attendees: IBM.
- Scott Farquhar: Well done, anyone work for IBM here?
- Attendees: [Inaudible]
- Scott Farquhar: [Chuckle] Next one.
- Attendees: [Inaudible] GE, Microsoft.
- Scott Farquhar: GE, Microsoft, Apple, Enron, good.
- Attendees: [Inaudible]
- Scott Farquhar: It’s Apple.
- Attendees: Apple.
- Scott Farquhar: Generic Company. [Laughter]
So, we thought the values, there are so many products you can put in values and you can see the difference you know, it’s not of words on the page, it’s really how you move them that makes the difference.
We use a thing called, ‘The Miles Explanations’, Jim Collins’s Harvard Business Paper, you can pay $8 and they will tell you exactly how to do this but the process is you sit down with your leadership team or groups employees and you say, “If we had to recreate our company on Mars.” So, if you know, I guess probably not building software but if we had to rebuild the company what we like what we have today on Mars, who do we sent? It’s not necessarily the Senior Leadership team that you need to send, it’s gonna be that cultural champion that always stands up for you, often to customers when they complaint or it’s gonna be the person that innovates or that’s something that you value. So, we did this exercise and we came up with initial list and then we call it down to another list and then we ended up with these being our company values.
We feel pretty proud about these and it’s actually when you chat with people that come to work at Atlassian and say, “Why did you join?” The number one reason that they joined Atlassian is that they’re associated with our values.
We made a video about our values and that’s actually our number one recruiting tool.
Number two is actually that we have a foundation and we give back.
So, the first two are actually more about sort of who we are as a company and then the work that they get to do. And we found that once we have these values articulated everyone can hire for them and that what I’m looking for. And more interestingly, that these values I think are used internally.
On our intranet we actually have thousand instances of people using our values, the acronym BTCYS is ‘Beat the Change you Seek.’ DFTC is, ‘Don’t Fuck the Customer’.
So, we have a whole bunch of our values and that get used internally. So, once we have the values we can hire the right people and we ended up now with 700 people, we feel like we got the right team.
Second part of having the right values is letting people go.
So, get your pen and I have got all have it do to item, when you go back to the office because at least 50% of you should go back to the office and fire someone. [Laughter]
You all know this that one person that you have been meeting to fire for a long time now that haven’t really being doing the job well but you haven’t get around to firing them. And I have one of those, you know, one of the first peoples I fired, the easy ones to fire lack cultural fit. So, I think it sounds hard in theory but without culture fit it’s really easy to say, “Hey, look you have been better off with someone else.”
The other one is the performance issue, if they’re terrible they job, it’s really easy to say, “Hey, look you’re doing a shit job, here’s the metrics you haven’t made them go.”
The really difficult one I found was when someone did do a great job and the company is growing and passes them. So, they will also that running a team of 50 people but now we have 300 engineers in engineering and you have never seen that before and that happened in the year and a half. So, suddenly you don’t have, you know the person is not fit for the role as it grew. And that was the really difficult one. And it feel like this, it feels…
You know, when I, the first person I had to let go, he was literally the Coach of the Australian Women’s Deaf Soccer Team. And he was the Solo bread winner of the family with three kids and you just couldn’t feel terribly both set there and cried for about half an hour when I fired him and it was horrific. But what I realize is that you can actually let people go in a different way that people feel you know, that they are heading on to something bigger and better. I showed this to Peldi who said, “Is that the challenge of team?” [Laughter]
And that wasn’t the point I was trying to make. [Laughter]
But you can’t let people go and will be thinking there into bigger and better things. No, no you can’t actually let people go because people wanna do a great job and if they’re not doing a great job, where they are? You can actually sit them and then say, “You wanna be the A player of somewhere, it’s not in this role, it’s in another role, inside the company or outside the company.”
I’m amazed once you have that conversation, how much better it is about letting people go? So, you all have home assignment that half of you should give and fire someone at the office.
The three things that we learn about Scaling and crises of scaling were you have to grow and die, you have to create and articulate and live your values and you have to keep people out when they are not doing that.
So, the big crises and probably the biggest moment we had at Atlassian was the security incident.
This happened about three years ago when our business is going fantastic, we have been around for eight years, we decided that we wanna take venture capital. So, we lined-up a whole bunch of meetings with Silicon Valley Venture Capital who really came to invest in Atlassian.
I was on my honeymoon in Botswana, and Botswana is a fantastic place but not a great honeymoon destination front of you who will be thinking about going on a honeymoon. The certain things about the malaria tablets and getting up 4:00 in the morning and so forth that are not conducive to romantic vacations and the ants and mosquitoes that meant… So, mosquitoes were 20 deep on your arm. So, it’s not a great honeymoon destination you know, we made the most of it.
I’m in the middle of the desert and I got this message from my Co-Founder and he said, that’s all he said, “Call me urgently, Mike.”
Firstly, the first someone who died in my family because why on Earth would you be contacting me on my honeymoon for any other reason. And I knew it was important because the message had gone from Australia to South Africa to the Botswana and head office to then radio to vocal camp and the guy driven an hour to give me a slip of paper.
It took me 24 hours before I could get out in the plane to call Mike. And what I found now was that first of all my honeymoon was over, which is an interesting chat with my wife – but secondly was that we found that a hacker had broken into Atlassian and what he done was in retrospect we found all this out that he had download…
But actually he read every single security advisory we issued over a three year period, he had downloaded our product and then spend three or four months trying to crack into our product and once he found the vulnerability he then attacked our public websites and got into our system, installed effectively not a rootkit but he could then keep interacting with our system. And then went about attacking our source projects and other people are running our software publicly.
We found out for various reasons but we actually had a bizarre thing when he then started his riddles and communicate with us by asking his weird questions in their own basically internal collaboration system.
So, you can imagine I’m sitting on a runway in Botswana about…
Well, getting on a different plane to what I thought I was getting on to continue my honeymoon and heading back to Sydney and realizing that we probably were not gonna get funding now. Then we shut the entire company down to deal with this crisis and I thought a bit like this.
So, what we did was we shut the entire company down and we told every person at Atlassian that unless it was customer facing, it was cancelled and postponed, the number one thing we had to do across the entire company, “We will sort this problem out.”
We split the company into four areas we had to workout what it gone wrong?
So, Historical Analysis, One Falls, and try to find out what the problem was? We had to fix it. So, we had to workout what the security issue was and then patch it. And then we had a separate team who’s dealing with communications for our customers. As you can imagine we’re running 10,000 customers many on the internet who run their own system and then that’s an interesting delicate balance.
The thing is then broke up in teams and you know, an old behavior doesn’t exist to say, “Hey, you know, you now run a team of 40 people, come back in one hour with the solution of this problem.” And that’s kind of the way we ran for the next three or four days. We had four meetings a day with entire management team globally and got together. So, no one got more than five hour sleep for that period of time. There’s a guy here who’s doing a great job sleeping. [Laughter]
We rented a hotel next door. So, all our staff stayed at the hotel, we open the time with the coffee shop you know, people were doing 48 hours straight. My Co-Founder was in a meeting where it was, my Co-Founder going to home at 11:00 PM at night, come back at 4:00 AM for the next phone call. And a guy had been working, one of our staff have been working there. And my Co-Founder says, “Oh, hey Mike, we haven’t met yet.” And the guy says, “Yes, it’s my first week, I love it here.” [Laughter]
So, everyone rallied around to solve this and we have this really tough balance between communicating to our customers who maybe at risk but then also not really knowing what situation, you don’t know the full situation at the point you communicate. So, we actually had a situation where we send an email to every single of our customers saying X and then we found out a few hours later that actually X was half the story and then there was a bit more. So, we have to send a second email back to you know, 20,000 email addresses and as you imagine , the second email makes you terrible.
We talked with the law firms and they all recommended us not say anything to anyone that was the standard response that we had no liability to our customers and we had no reason to tell them that the systems you know, the systems may be compromised and we just had to fix it and patch it. And that didn’t sit really that well with us.
We had the FBI, the Australian Federal Police, two security firms in Australia, two security firms in San Francisco, and a law firm on each side. So, it’s kind of the all in situation and that’s day you not really worried about cost it’s one of those ones that will pay for you later and I’m sure we pay a lot for it. But with the whole company was yelling around and I ended up what happens as we found that there was a guy, a 28 years old, we found, we know who the person is, eventually he’s a person in the Eastern Europe who’s a hacker and eventually he just went away, just disappeared, didn’t you know, worries anymore and we fix the patch and fix the software, and we have now employed security teams and you know, we are with the penetrations and detections all these other things that we have now kind of issue in response to that.
I guess through the whole period of time where you know, telling our staff that we’ll through this and it’s not a big deal and it is actually on the far side of something like this, it is actually something you can pass three years later that as a company we have grown you know, to the size and scale that we have even though that we had this incredible security.
I flew back from Botswana on Friday and on Tuesday I still flew out to a Venture Capital meetings that we organized beforehand and off course all of them got the emails from us saying that, “We would had a security bridge.” And it turns out that we don’t worry about it at all. It turns out it happens all the time. [Laughter]
No one talks about it!
A week later we ended up closing 60 million dollar round with Accel Partners which is the largest ever software deal. So, it was just incredible like you started the company with entrepreneurs and see there was incredible high and low and it’s just the most just the position for me was you know, the row it was my honeymoon coming back thinking you know, I kind of shut the company down effectively and on the high it was a week later when we started we raise funding.
The lessons for me on this one – you can be prepared.
This is something we probably should have been more diligent about and now we have team of security professionals and the head of security and we have Risk Officers and all these things but security you know, backups and disaster recovering and all these other things that get overlooked or not focused on when you focused on delivering stuff to your customers. And I also found personally that in your management meetings you wanna spend time on things that you can add value you know, as a management your sales strategy and your company and developing people, and I found that as a leadership team our job was also to ask questions about things that we couldn’t fix but just by asking that question on a regular basis meant that people across the company would solve it because we found too that as we grew there was an air gap between management and staff.
We found that a certain way our staff said, “I cannot believe the management are running the company in this way.” And when we found out about it later as management we think we can’t believe the staff didn’t tell us if that’s the way things will being run. So, be careful of the air gap you get you know, between because people do honestly believe that you know, that the other area around you know, to know something they don’t.
The second one is ‘Living Our Values’
I think in crises mode it’s really easy to throw your values at a window and just do what’s speed into the time. And the weeks afterwards we had our staff come up to us and say, “I love working here more than I did beforehand because the way you dealt with this incident, the way you actually dealt with the publicly even internally and a lot of the companies wouldn’t tell their entire staff about this, and this would be a small group in the corner trying to fix it.”
So, we are living our values and it was a good test for us.
We put it on the bright face in front of everyone say, “Hey, we gonna need to go through this and this is gonna be just a road bump.” But when you go through your crises as I know you will over the coming years just remember that it was something that seems quite large and it will something you can passed.
And the last crisis that I wanna talk through was one that I actually manufactured.
Sometimes it’s good to manufacture little crises when you know of something is gonna be a problem in down the road but your staff take a little bit of time to get around to it. And our pricing for a longtime was a bit like this – we had one version of the price – in fact their first ever sale was $600, and the list price was $800 but we thought you are the first person so we discount it even further for no reason. [Chuckle]
Then eventually we do this and it’s something you decide, “$1,200 that seems really cheap, how about we add a standard version and let’s add few more features?”
For about six years this is the way that we sold our software. And then I read a book called, ‘The Innovator’s Dilemma’ and I know you have had, they are quite speaking, haven’t we? Who has read that book? It will be about half of the audience. So, I think if you are in technology you have to read this book.
This is my number one technology book and what it says is that, “You will get beaten from below. It’s the competitors that are small or more nimble cheap don’t do as much functionality will eventually beat you out.”
And to give an example, this is actually one of the examples from the book and if you are in this industry this is an article industry been around for 40-50 years and this is the total unit sales across the entire industry. And if you are sitting in strategy meetings you think, “Okay, it’s an industry that’s growing you know, reasonably fast you know, so 5 to 10% a year. You know, that’s good to be industry investing and there’re sort of competitors on the side but doing something sort of the same but any real person, any customer, any real customer wouldn’t deal with that, that’s a toy. It’s growing maybe in a couple of years we should invest in that area.” And then it’s amazing how quickly these things can turn around. And taking guess what that chart is of? [Inaudible]
So, digital camera started off as toys, they started off you know, little webcams and 640 x 480 cameras that couldn’t do much and people will thought that you know, a normal film camera is about 8 megapixles in resolution. So, all the film people thought, “Well, we don’t have to worry about digital because the digital gets to 8 megapixles because then people you know, then we need to adopt it.” But it turns out the most people don’t need 8 megapixles, they don’t print kind of well I’d say A2 but I don’t know what called in US, that don’t print big stuff, they print you know, A4 size things. So, 2 megapixles is enough. So, if you get a chart of camera sales, the digital camera sales, it was one that 2 megapixles that they shot up.
So, when you think about your competitors you think about who’s doing something slightly similar that has all the advantages that they can incubate somewhere somewhere else? And to convert that there’s no other reason in the industry why shouldn’t be out of the cheap and sort of not allow that competitor to incubate. You should always be able to price your product so you don’t know why that sort of soft inability.
So, I read this book and I looked at out software pricing and then I looked at pick random competitors software pricing and I thought that’s reasonable you know, like we get stand for special price of their user based you know, the sort of a rough party there and then it did some digging into the data. And it turned out that actually 80% of our customers go with this version. That’s no longer when I’m really comparing standard professional enterprise against their competitors, they’re actually comparing this version against their competitors. And we found, “Oh! This is gonna be a huge problem for us we must be losing plenty of customers to our competitors when we have three to four times the price at the low end.”
As company, the real way to move this way was to move from a sort of feature-based model to a per-user pricing.
I think a pricing model was great when you can extract more value and the customer gets more value. In fact it’s as a random thing on pricing is that when you sell something that is actually less valuable than your customers believe it to be and I found out that overtime get all your money upfront. [Laughter]
But when you sell something that is more valuable over time make sure that it’s a subscription and something that grows over time so you have add-ons and other things that they can keep buying. Our products are collaborative products. So, it’s actually difficult to get them in and started but as I grow they get sort of more value and more network effects. We wanna price but use it such that we got more and more the value as they grew. So, when you had to move to this direction but the problem was as you move from tier-based pricing to a user-based pricing model where you have got to change every bit of coding your entire licensing system.
We had to change the way where the products works, the user tiers, the billing systems, the upgrades are completely different. So, we had to move that but we’d ensure what the first step was?
So, we said “Okay, what if we did five users for free?”
The issue we have is that we had a sales funnel where most of you and you said, “Well, you all those guys, the five users for free, are they people that eventually gonna buy a bigger license or they are just people that are you know, free version and I have different messages between the two different channels.”
Then we said, “What about if you charge $5 for the product?” So, that we looked at and we can differentiate you know, the people who really want the product five users saying like $5 is a reasonable amount of money.
Then my head of support was worried about and he said, “Well, I don’t know how many people we are gonna turn up in that doorstep asking for support?” And at Atlassian one of the values as you know, legendary to the service that we provide and we want every customer to have legendary service and so all these people or it could be little bit like this. [Laughter]
We said, “Okay, but how do we test it? We can test anything.” So, what about if we did instead of for whatever had but we just run this for five days and we say all the customers that come in and then we see how much support they generate and then we can work and how much of cost that we can adjust from there. So, we run a five years of campaign and we thought let’s give the money with the charity of $5 is not much and if it’s charity we can make some payout about it and get some press.
So, we did this and in fact we actually did five years of campaign even when we are still on our pricing model on a sort of a feature base pricing model. And over the five days this is what we thought we will gonna get you know, we had all these bets internal and this is the most optimistic bet that we made internally. And here’s what we ended up dealing is we ended up selling just over a $100,000 worth in software in $5 increments and it’s pretty crazy when you think about it and how much you can raise in a week with $5 increments. And that from the basis of you know, kind of the understanding and how much support that we generated of that and what all those sort of demands from the basis for all the pricing model that we did.
$5 to charity ended up being actually a large amount of money over time over the first few we actually gave away a million dollars to one charity in $5 and $10 increments. I think some of the audience or our customers will appreciate that.
So, I eventually move to now, I play you some model and as they move to 10 years for $10, the support cost was throughout a mountain then you know, if we compare this to the old competitors pricing model and we end up in a much better situation.
This is a new demand curve sort of mix. So, instead of 80% being on the top two we actually have a sort of what you compare as a normal mix where you have lots of people involving and if you include starter licenses that’s what it looks like. So, the starter license was incredibly successful. So, if you are not doing framing or some sort of free offering I’ll suggest you to do that.
So, that’s from pricing were you beat from below and make sure absolutely that your pricing curve goes so there’s no room for your competitors to beat you from below. And for us who that already give about 5% of our revenue you know, in five and 10 years. So, we feel that was a good long term bet to make.
We manufacture the crises, it’s one of those ones that the organization could have taken two years to move across and I said no absolutely this is a crisis that we need to solve today.
Lastly is that if you have got constraints on time and other things you came with creative solutions to that. So, the four crises we had at Atlassian and I’m happy to answer questions about those or the hundreds of other ones that we have had. [Applause]. Any questions?
Don't Miss a Thing - Get BoS Updates
Audience: Hi, I really appreciate your candid about you know, your experience of all those crises. The last point you discuss about pricing, you are a fairly huge company and you have for your pricing and publishing like that and you deal with all those huge customers and your sales is all accumulated by that standard pricing and just add up like that or you have like other deals you know, beyond your pricing model like that.
Scott Farquhar: So, is your question you know, do we make 150 million dollars? So, if 30,000 customers and 150 million in revenue which makes about $5,000 a customer a year on average, biggest customer price are close to million dollars a year and we have ones that pays very small amounts. We do have, we are introducing high price products that are not yet on the list but I think our highest price at the moment is 24 grand for zero if you wanna to buy it. So, we do actually sell that amount of money at that scale.
We take a philosophy that sales people introduce friction? And they say, “Well, you need to a sales person and find the price.” That doesn’t make any sense why would I do that? And sales people in this friction so they have an excuse to talk to someone. I find that you can find more and more reasons to bit friction in front of people until you have red forms and sign up to get a PDF before you can download and eventually it’s a really complicated sign up process and someone is gonna eat your lunch with the simpler model. So, that’s the way we take this but everything is public as much as possible.
Audience: Well, first of all I have been a fan of your company for seven years or eight maybe even, the awesome work but that aside you are talking about when you manufactured your pricing crises and how you came to the conclusion that you had an issue because when you compared to the random competitor. If you haven’t come across that competitor and how would have discovered this problem and where would have you know, if you hadn’t come across the sources of your problem if the confusion for yours customers and why are you missing out on those cells how you have found it?
Scott Farquhar: Yeah. So, we are really data driven at Atlassian and we will get huge amounts of numbers. So, we have a philosophy which we are sure anyone who I think waits at everything that Atlassian in the database. So, I can search staff when they join other things that you know, and all my customers and all the folks… [Laughter] So, found it through the data and interesting part is that really have to find customers that don’t buy your software. And ask them why they didn’t buy it? Because generally you get a buyer in your surveys, you send us surveys we didn’t buy and you actually find the people that have a brand affinity of the ones that reply back to you, the ones that didn’t care and you know, I want a software why’d I answer that survey and so you end up with a mess.
And what we have done is try and call people I actively go and really test the people who don’t buy to find out if there’s a competitive issue there that’s the biggest thing we have got, it’s really hard to find market share in lot of the industries. So, you wanna find the customers that didn’t buy and how I tell you.
Audience: Hey there. So, you have started licenses which imply you wanna move them into higher tier pricing. I guess first how successful are you in moving them that 80% starter licenses users into higher tier and I also I heard to talk that you guys don’t have sales people. So, how are you moving them without sales people?
Scott Farquhar: Yeah, it’s a good question. So, about half of our sales in our products come from someone who envier a starter license. Now, that maybe that is wanna a four month evaluation and then decided to keep it small but and you never know that but about half of our leads now come by that channel. So, it’s been hugely successful for us in terms of the no sales people aspect that’s why we will be known quite publicly and says, “Hey, Atlassian has no sales people.” And that actually the conference come up and say, “I don’t think we should have sales people at my company either, tell me why?” And I was sort of, “Well, it depends, this is the philosophy that’s the important part. And the philosophy for us is that sales people are friction in the process like any person is friction in the process. The customers to call you for support, you should have fix it in the product.” So, whenever you anytime you have to talk to a sales person effectively a bug in the process and what can we do to automate that? For example, what we did is we have people that chase to renewals. So, if you pay us you know, $10 in the first year we collect $5 for a renewal for behind the firewall software a year later. And we originally, we had you know, a sales guy that came in and say, “We just throw people in this problem.”
It turns out that you know, we gave the leads to these people to certain date but a lot of people just renewed later without any human involvement at all. So, we found that even though these people did add value, they did as much as they would have thought having on their own experiment. So, now we actually experiment when we give the leads to the sales people.
So, at the moment we give our sales people the leads 30 days after the due to renew. So, after we have given them emails and opportunities, and invoices, and send all the automate stuff that we can do then it pops into a person’s bucket and we find they get from a 75 to 85% renewal rate. But if we sent all of the leads that would have to call 100 people to get that you know, that 85 points but now they are calling here 25 people to get 10 points. So, it’s actually much more efficient and then we do experiments.
We said, “Okay, what if we give them the lease 30 days earlier? Do we change the conversion rate?” And no, it turns out that actually you know, 30 days and earlier doubles the number of people I have to call but it doesn’t change the conversion rate. So, they know the sales people will approach for us is really we view anytime they speak to a human as a bug and how do we automate that process as much as possible.
The threat we are getting into the moment is that people inside Atlassian get dogmatic about it and say, “We shouldn’t throw sales people anywhere.” So, even in total and having to fight that a little bit and say to people, “No, what I want let’s draw person at something to show them what’s possible, right. Can I get to 100% when you are right with a person then let’s to back and look at how do I the automated and I think that’s a way you should do things as run an experiment to look at what is the high watermark?” Because with the automation you get decreasing returns overtime and we talked about local maxima versus absolute maxima earlier. You can get a Local Maxima by just increasing the automation, throwing a person can do something completely different. So, we do that and then we make all it mandates.
Audience: What kind of stress that the opening of the lower tier put on your support staff?
Scott Farquhar: Yeah, it’s a good question. So, it’s about a full exact numbers but it’s about 10% of our support load comes via those starter license customers. So, that makes a very small fraction of average in revenue and giving to charity. So, in fact that’s a loss for us and we did all the model again with the five day experiment we can then do all the modeling and we roughly estimated out I think it was 10 to 15% and end up in 10 overtime. So, we really aware that cost is a marketing cost. And the last question over there, what was it?
Audience: Hi, Scott. I wanna applaud you for the, you all are going out and letting your customers know when you had the security breach. We had an issue back, this back in the mid 90s when we were shipping demos on floppy disks, we had gone to a trade show and we ship that about 300 demos post that trade show, started getting phone calls after that where people were saying, “Their CD-ROMs had stop working.” And we were like, “Well, we don’t do any IO with the CD-ROM, there must be something else, and we get to three-four calls, okay there must be an issue here. So, we do some digging, we are looking and what happened it was one of the worker’s kid that came in at weekend and one of the machines we used to make the demos had a virus on it.” So, we have to contact all 300 of the people I got demos and tell them destroy your demo, just throw it away, we’re gonna ship out a new ones because we didn’t know who got the good ones and who got the bad ones. So, they ended up and there was roughly about I think about 18 people and companies that had installed it and we actually had to ship them software to get the virus off their computer.
Scott Farquhar: What was your learning from that?
Audience: Well, obviously don’t let the kids coming and play games on that but we were convinced that there was gonna be…this was bad for us. And we actually got a sales bump from that and I’m curious you’re contacting 20,000 customers that you had and letting them know, did you notice an increase in sales after that happened because you were that open about the factor, issues about with the product.
Scott Farquhar: Yeah, I don’t know if we did and if we did I’m not sure I’d say it publicly. [Laughter] But yeah, we didn’t run the numbers to see that. So, I think you need to go through some stuff and then if we get that video online if that’s possible as people walk out.