Atlassian Planning U.S. IPO
wsj.com
wsj.com
Most enterprises would do well to look at how enterprise products have been shaped to work and to conform to it, because there's a reason those tools work that way...
Everything installs and runs easily (usually "untar and run the startup script") and it all integrates together easily (i.e., give product X the URL to product Y) and deeply. That's what gets people hooked.
There are lots of agile tools but most of them deliberately ignore the enterprise space.
Disclaimer: MS engineer, these are my own statements etc etc.
The bug/issue tracker we're currently using is not terrible, but it's a little painful. However, we've just barely gotten people trained to use the current system. I can either accept the current, somewhat clunky system, or go through another year of retraining and hope the slicker tool and the additional features are a net win.
At least JetBrains actively works on their products. Our current tool hasn't changed in three years, and that was just a reskin of the UI to a more bootstrappy css.
Huge downsides:
- Doing centralized authentication in a large-organization environment is basically impossible. Its LDAP implementation is broken for anything else than a couple very standard schemas that they've tested against & you can't make it authenticate using a Windows domain. They've recently split out the user management, etc., components into a separate product called Hub, but it's the same code and suffers from the same problems. [NB: Hilariously enough, other JetBrains products work just fine in this regard.]
- The thing uses some custom proprietary database. Integrating backups into your normal workflow is painful. Upgrades are VERY painful (oops, the old version had a database that didn't enforce some constraints that the new one does, please go fix all your issues by hand), although there's some auxiliary API and tools for exporting individual issues as -- if I recall correctly -- JSON objects that was an 80% solution.
Upsides:
- The UI does the expected thing by default in all cases we could find. It is also fast & responsive, relatively uncluttered, and doesn't have a bunch of extra things to click in order to create/comment on/edit issues.
- Searching for issues is very fast and powerful.
- Batch updating issues (which we tend to do a lot) is extremely easy. Make a custom search describing the issues to update, then update all of them with a couple clicks.
The admin downsides were (for us at least) bad enough that I'm not sure I'd recommend deploying it if you already have a working, integrated system. If you're starting from scratch, though, or committed to replacing another system for whatever reason, it's a pretty good choice.
Upsource also looks very interesting, is that also something you have used? We don't really have a good tool for doing code review at the moment. Also, the VCS browsing capabilities look to be miles ahead of what our current setup offers (almost completely unusable, hijacks the browser back button, no searching, definitely no chance of static analysis, links sent out in email notifications constantly pointed at the wrong thing, etc.).
Currently we're looking very hard at Critic (https://critic-review.org/ is a demo site, code's on github) because it looks to offer the closest workflow to what we use.
We're specifically looking for something sorta similar to Gerrit as far as how reviews work, but while actually supporting a feature-branch-heavy workflow where we tend to review branches rather than individual commits, and frequently rebuild/rebase feature branches.
Obviously, if your workflow is different then your needs for a review tool may differ wildly.
Other than that, I took a quick look at your product and it's feature-wise close to something I would want to use. Only comments I would make would be that non GitHub projects generally want the review tool to handle the branch management (a.k.a storing pull requests, flagging their status, etc.): of course only focusing on GitHub lets you not worry about that and get a product running much sooner... You should spend some serious time looking at your page load times as well - I'm 10msec away from your server and the demo code review at https://reviewable.io/reviews/Reviewable/demo/1 took nearly three full seconds to load, even with most static assets already cached since I didn't timeline it on the first attempt.
Developing against the GitHub API does carry some non-trivial advantages: among other things, I never actually clone any repos. This wouldn't be possible with a raw git repo. I'm considering adapting the tool to AWS CodeCommit, though, which might be a decent middle ground. Making my own equivalent of pull requests is actually the least of the changes I'd need to make, though.
And yeah, performance is a never-ending battle... But considering that all the work is done in the browser (fetching sources, diffing, formatting, etc.) I think it's actually not too bad. Again, like running in the cloud, it's a trade-off of in favor of flexibility and development speed.
The gotcha? Runs on Windows Server which probably doesn't compute for startups but works well for typical enterprise shops that have stacks of Windows Server.
Self-hosting is still a huge market but one that is more and more under-served.
Hipchat is an abortion/abomination/anthema, but was an acquisition.
And the Atlassian things don't require launching a virtual Windows machine on my mac every time I want to use them.
Maybe there's better solutions than Atlassian out there, but Atlassian is still a helluva lot better than what a lot of companies are actually using.
I run a managed hosting company that hosts a fair number of Atlassian instances. We have our complaints about Atlassian and our customers often experience sticker shock when upgrading past the 10-user licenses, but we've had trouble finding other issue trackers with similar functionality, flexibility and market acceptance.
The same goes for Confluence as a wiki. While not perfect, most other products seem unable to do things such as access controls easily.
If there are other mature products out there that people are using we'd love to take a look. We do support Bugzilla and MediaWiki as well, but this is usually a different market segment from what Atlassian serves.
After some organization changes -- long time coming -- we switched to hosted Confluence, JIRA, BitBucket/git. The transition has taken some time but communication is going so much more smoothly since.
Confluence especially has been great. Designs/how-tos/best-practices are so easy to write. Permanent records of discussions and decisions are so easy to track. I made sure to collaborate with a few others to establish the skeleton of our Confluence -- without a good skeleton it would devolve to chaos. But Confluence is so much better than Word docs floating around ... and I haven't seen any other Wiki tool that is so rich in authoring ... Confluence has:
* great search capabilities
* easy to organize
* easy to create links to other relevant content
* great text styling
* great tables
* plugins/macros for displaying PDF / Word / Excel inline
* linking with JIRA
* great diagramming tool Gliffy
* easy cut-and-paste-of-images -- capturing screenshots is fast
* (unfortunately to get GraphViz diagrams you need your own server, hosted version doesn't allow for that)
* good export
* and plugins for almost anything you'd need
We compete routinely with Jira and win deals by providing fully loaded 50mb zipped solution that runs on Windows servers, integrates with AD/Outlook/Exchange, etc., custom Apps Framework and obligatory RESTful API. Just the things that "demanding enterprises" look for.
No idea if this helps you.
I'm not really sure that there are better alternatives out there.
Some quotes:
Q: "What was the scariest, most panic-inducing thing to happen to/at Atlassian?"
A: "The most topical fear we have at the moment (and one of the reasons you see us spending so much time on it right now) is that Atlassian is growing faster than the Sydney (and Australian) technology industry. That cannot be good for our long term domicile here - we must continue to diversify product + engineering overseas, or risk running aground."
Q: "...before Atlassian for saving bookmarks which you sold pretty early. What did you learn from that business that helped the most with Atlassian?"
A: "Overall? It was a fantastic lesson. Failure (or forced sale type success in a smart decision before we hit the wall) taught us many things. How to get customers. How to build software. How to prioritise our day. How to get a business name. A domain name. How to run books. Get investors on board. How hard it is to make $1. Harder to make $10 still etc."
And the other co-founder (Scott Farquhar) did an interview on ABC's lateline show earlier in the week: http://www.abc.net.au/lateline/content/2015/s4317805.htm
I remember reading a brief comment by Joel Spolsky saying he wasn't too worried about "those blokes from Australia". At lot more people I work with know more about Jira than Fogbugz.
IMO Atlassian do OK tools. But nothing to justify the hype they are always receiving. What am I missing - Price? Features? Implementation pain (or lack thereof)?
It is encouraging to see a bootstrapped company reach IPO, even if it took them a decade to reach their $3B valuation compared to Github's attainment of a $2B valuation in a much shorter time. Atlassian's forward-thinking private investors provided liquidity to employees, which supported their long road to the public market. From a 2014 interview, http://fortune.com/2014/04/10/atlassian-the-3-3-billion-soft...
"The $150 million it recently raised isn’t going into its bank account — instead, its new investors are buying shares from past and present employees in an effort to offer workers some liquidity and retain talent ... 'the Accel Partners round was mostly about founders, and this [T. Rowe Price] one is about employees. Now neither of the founders are selling anything. So this is a little different. We’ve been profitable for 10 years, so we don’t need the cash, we have a lot of cash in the bank. It’s more a case of wanting to get a partner [that knows about public markets] on board..'"
From a 2011 review of Atlassian culture and self-service sales, http://www.managementexchange.com/story/its-culture-stupid-h..., with a talk about internal systems: https://youtube.com/watch?v=mALn4nyi7Qs
"Atlassian CEOs have instilled an information culture throughout the company with the core value of "Open Company, No Bullshit". The result is a bottom-up democracy of information where information sharing is the norm and information hoarding is a foreign concept ... Product pricing is public, including discount policies. Prospective customers can download software and try it free for 30 days. All documentation is public and available for anyone to access. Bug reports and feature requests are public, allowing people to 'vote' on their preferred features."
Good price structure and business model.
As you said, the software is so-so.
But remember, Bitbucket is free for private repos. As many as you want. You can even share them with a few users before you have to pay. With Github, you have to start forking over cash for private repos.
Bitbucket gets you hooked, but then you have a raft of tools that are tightly integrated that aren't even available at Github without going to 3rd parties. Nobody wants to sign up for a dozen different services and try to integrate that.
You get a ton of tools out of the bag that you need and that work. Nobody else offers that right now. Gitlab is trying, but they have a long way to go.
[1] n.b. I have never broken either my finger or my femur, but I have used Jira[2] quite a bit over a period of 8 years.
[2] I am also amused by the origin of the name: "Although normally styled JIRA, the product name is not an acronym, but a truncation of Gojira, the Japanese name for Godzilla,[6] itself a reference to JIRA's main competitor, Bugzilla." https://en.wikipedia.org/wiki/JIRA
Something to point to when pressuring the Ausi gov't?
The Atlassian founders are known to be good at putting their money where their mouth is. Perhaps they feel they can make a greater impact with the extra funds than with a single statement.
I believe they are quoted somewhere as saying the ASX doesn't understand tech stories like Atlassian though can't dig up the quote right now.
JIRA does feel clunky, but after spending the time configuring it, I find it helps me more than getting in the way. I've tried a number of other tools, especially for Agile development, and I ended up feeling like our team would need multiple told to get things done. By trying JIRA out on work projects and personal ones, I'm now a bit more productive, and my team is getting there too.
The price felt a bit 'steep' after the 10 users, but I've convinced my company that the price per user is nothing as it's how much we bill our clients for a few minutes if our time.
The listing makes sense, they've been profitable for a long time, and should have a possibly sustainable share considering EPS and other requirements of shareholders.
You'd say an entire company sucks because one feature of one of their products is underwhelming to you?
So we went from the Crowd based setup to LDAP / postgresql backed versions of the products. The big surprise for me is how divergent the products are. User setup and the cross application links are so different. The underlying postgres schemas are different. So the migrations were fundimentally different. (In hipchat's case it is understandable as the self hosted version of the softwre was new and somewhat beta)
It reminded me of how back in the 90s I heard that Microsoft kept the developers of different products in the Office suite apart.
Thus when you go to link the applications together you get weird redundancies and it all doesnt work as it should. Partly that is because the licensing is so separate. Atlassian support was pretty good in that they worked with us on it and also they provided unofficial scripts to help withthe process. They also recommended using Crowd to manage users which to my mind only added a layer of complexity - you should be able to add users directly from LDAP.
I ended up enjoying the migration project because I got to work with postgres which is great software.
Atlassian is bloated and in my opinion they havent worked on cross application connectivity enough. It should be seamless.
My company (at the time) ended up hiring a fulltime person to manage the complexity (ie an Atlassian admin) plus a few consultants to get the workflow stuff down.
It is good when you have jira and confluence running well - but it takes a lot of time and money to get to that point.
Offtopic - I have recently been trying to break into full-time consulting / contracting. May I know how you went about hiring the consultants for the migration ?
As time went on and I moved on to do my own gig. I wanted to go cheap so I explored many alternative products to replace Jira/Agile, Bitbucket, Confluence, Bamboo etc.. what I ended up with is different tools with different interfaces sticky taped together working just OK.
This got frustrating and I decided to dump them all and move to Atlassian OnDemand. The overhead of managing & educating staff on all these tools was just not worth the pennies saved. This is the power in Atlassian products, their tight integration and consistency. Yeah there are quirks here and there but those always get resolved and features added over time.
Anyways I love Atlassian products and I hope their IPO is a success! Long journey to get this far but well worth it. Solid company!
* well customized to fit the business (i.e don't let those 'non-practicing-entities' with nothing else to do go crazy creating and reworking convoluted workflows)
* small number of users (or performance sucks)
* well organized (you tend to get 'dumping grounds' of bugs and tech debt that just accumulates). I like trello's concept of card 'decay'.
* deep pockets (you gotta 'buy in' to the ecosystem and not get stingy about who gets access)
* ignore features - don't try to shoehorn, use what you need and forget the rest
What is strange to me is how on-premise focused their business is. This is a market at or beyond its peak. Perhaps that is the reason for the IPO at this point?
http://gfycat.com/FarflungUnderstatedIcelandgull
Sorry about url, gfycat so weird with their url generation