I like to call Confluence a knowledge cemetery - in the huge datasets you have in mind, finding the page you are looking for is more a matter of luck than anything else. And once you found it, you can never be sure if it's still up to date.
> Anyway, they have 8000 employees... they can always drastically downsize and refocus, the company will be around for a while.
I'm actually wondering how they can not make a profit with the huge corporate user base they have?
Spot on from my experience using it in big orgs.
The issue IMHO is that organization of knowledge is not following any system and/or no system that could be followed was ever set in place by the organization when they bought into Confluence.
It's kinda like ordering an Airbus because they're a reliable plane. Then you sell tickets and everyone is board of course because ... Airbus. But you never hire a pilot.
If organizations using Confluence had at least one full time "Wiki Master" position or the like (with resp. superpowers), this wouldn't have to be like that.
I.e. I'm not sure the argument can be made that this is Atlasassian's fault.
As someone who spends a lot of his work life administering Jira and Confluence, I disagree. The problem is that these products try to be everything to everyone and as a result users have a massive degree of freedom with what to do and how to order things.
With that amount of freedom you just have a bunch of end users, many of them non-technical, who have no idea how they are SUPPOSED to use it. Especially because they don't want to spend a lot of time learning the right way to use the tools. They have a bunch of other things to do and just need to interact with this tool to complete their current task and then move on.
You are right, a sort of Wiki-Master would help. But, at least in my organization, we don't even have enough capacity to do all the basic administrative tasks required like setting up projects, workflows, custom fields, screens, etc. We can't also manage the content inside. The users are on their own in that regard.
If you are not willing to hire stuff with the sole purpose of managing the content of your Jira and Confluence instances, a more opiniated, rigid product is probably a lot better for your organization.
I think part of the challenge is finding and retaining this person. It takes a fairly unique kind of person to want/enjoy a (let's be honest) thankless job like managing and maintaining a huge corporate wiki.
Finding good tech writers or information science professionals has always been a difficult task in any company I've been at.
Documentation and wiki writing ends up being side-of-desk work for someone else, and so of course it never gets maintained.
Jira - Everyone has the performance issue, to avoid it they tell them to carefully choose extensions and not randomly litter it, especially with extensions that conflict each other.
Confluence - They recommend to establish a proper structure in advance that everyone follows with templates that everyone can use(not the default ones), so that the cemetery issue doesn't happen.
That said, I haven't seen any company who's confluence wasn't a knowledge cemetery. But regardless of how much I dislike Jira, I'd still pick confluence over google docs.
But what about storing documentation as flat files in version control?
What about even storing tickets as flat files in version control?
It eventually got replaced by some JIRA-like monstrosity, I don't remember which. All the text-based goodness was lost. The excuse for the move to a different system was "performance", which was half true. I'm guessing that nobody wanted to go fix it.
The point behind all this rambling is that yes - flat files can be made to work in the way you mention. It just isn't an approach that is cool enough.
I agree that any approach should not have many hurdles!
I've been using https://github.com/just-the-docs/just-the-docs and find it much easier, nicer looking, and people are excited to not use Confluence.
Of course it doesn’t apply just to confluence, but to the whole paradigm of uncurated, informal knowledge dumps.
Unless there is a real process around documentation, it just rots, and the key problem is you can tell what’s fresh and what’s rotten.
Exactly. Other products comes with opinionated workflow or are much easier (and simpler) to use. Once you setup Jira way you wanted (good luck with that) and able to maintain this setup for all of your projects you're stuck with Jira forever - you won't be able to export all of your content to another software. The whole process takes ages and its a nightmare of going thought totally different UX pages, unintuitive settings, slow UI, setting up extremely expensive plugins (why till today you need to buy tempo to get basic functionality like time tracking) etc.
It's called vendor lock. Nobody ever got fired for buying Jira
Other products are not for every business, have limited settings - but are quicker, saves you time and are able to be useful since day one.
All of the companies are using Jira differently, even if you're familiar with this, and you have been using Jira in all you provious jobs for years you still need to have onboarding on how does Jira works in current company - as for instance I've worked with Asana in 3 totally different companies and workflow was the same.
But not everyone needs 'comparable' functionality - an awful lot of people use just some features and not others - because of the awful UI and terrible user experience, these people are ripe to be picked off by a competitor (existing or future). I agree, if a mega-corp is completely in bed with JIRA, its harder to switch, but for a company not making any money (Atlassian), having their small and medium sized customers peeled off bit-by-bit for better products is going to hurt.
Most companies don't need "comparable functionality" and most did not built entire workflows around it, and those companies will leave for competitors. The few that locked themselves into Jira will face sharply rising prices in next years, that will make them reconsider their choice.
They have bad products and they are losing 700 million a year, future for Atlassian is bleak.
So it ain't nonsense, users do switch.
I believe that TFS product evolved to Azure DevOps now.
My main problem with confluence is that it doesn't know what kind of text editor it wants to be. Sometimes its markdown (or rather its own undocumented flavour of markdown), sometimes its a rich-text editor, sometimes its just plain text, and you kind of figure out on the fly what to do. Also the codes boxes are absolutely massive and a bit of a pain to work with (which, like all of these problems, could be solved with just normal markdown).
In the grand scheme of things though, its a minor annoyance.
And while Jira is clunky, but other enterprise grade solution I've used are worse
Don't get me wrong I'd pick lean alternatives any day of the week, but for what it does there's few replacements
And mercurial too. I do miss mercurial.
Caveats: new job is small so it's a small instance. JetBrain's own YT instance is pretty fast though, so I guess it can scale well. Also, we don't use the new KB/wiki feature so can't comment on that. It's probably got less features than Confluence. We don't use the scrum features.
Biggest caveat - I speak as a developer. One of the biggest problems with Jira was the culture that surrounds it. Clueless PMs who had never written software in their life kept creating fixed workflows in which you couldn't transition tickets to arbitrary states (for no apparent reason, there seemed to be no benefit to these rules). So we were constantly being slowed down by the need to get the handful of people with Jira access to edit the workflows and add more transitions, which was pure makework, especially as even figuring out who the admins were was remarkably complicated. PMs also had a weird Jira fetish, they were like, if it's not in Jira it doesn't exist. So they forced the project to change from a more normal SW dev project early on to mandating that every commit had a ticket associated, and that ticket had to be scheduled into a sprint etc. It was just rank stupidity. Want to do a quick refactoring, as part of your current task? Roll it all into one giant commit because otherwise you'll need to file a ticket. New place doesn't work like that. YouTrack is used more simply, like a todo list. Git logs are the source of truth.
Space wasn't there yet. IIRC the main issue was lack of a way to set custom fields via API, or some weird restriction around that. My general impression was that it was promising but too immature at that time.
For any Jira competitors reading this - start with providing 2-way sync between your product and Jira instance. It would make moving away from Jira soooo much easier! The current state is that even importing Jira data is below what I'd expect - e.g. Linear (last time I checked) didn't import comments at all and all issue IDs were changed to new ones. Really, if there's FOO-123 in Jira it's table stakes for it to be also called FOO-123 in the new tool.
I wrote a tool to migrate an issue tracker to Jira which imported everything as expected and updated links in issue descriptions/comments to keep their integrity. It took few days for a one-time migration so I'd expect someone who builds a business in this area to just have that as a baseline. It's sad how bad the "import from Jira" journeys are.
But I didn’t use Jira for years. I hope I never will. To the degree, I’ll need 20% more pay for comparable job if they use Jira.
Confluence is the single, worst Atlassian product, in my opinion. I describe it as a wiki written by someone who has never used a wiki. It seems like everything about it is wrong in some way, and it's painful to use.
Footnotes? Ha! https://jira.atlassian.com/browse/CONFCLOUD-68553
I think you misunderstand the way some companies work... Some companies, a VP will send an email saying "As of next week, we will use Asana instead of Jira. Jira will be deleted, so copy anything important off it now. NO EXCEPTIONS!"
Sure, there will be mayhem for a couple of weeks, but the migration will be successful. The VP will have one more thing to put on his 'done' checklist.
They bought Trello for that
One problem is, that not everyone is willing to educate themselves and people expect to be able to use things immediately. This causes the push to dumb down things into a limiting experience. If I were surrounded by only technical people, then a plaintext file like an org-mode file with predefined todo-labels, commited to a git repo would beat all of the Jiras, Trellos or whatever out there and it would cost not a cent and be safer to work with, as everything is stored in git.
Every time there is an update I wonder what feature they removed or is broken. I still cannot properly add images in lists or continue the numbering in a new list to work around this "feature".
They also broke markup/down import which is another kind of hell.
But there is no competing product (suite) that offers comparable functionality and provides the same long-term flexibilty.
This is maybe true, but the future belongs to more flexible software. Eventually Notion [1] or Fibery [2] will replace Jira. They may do it now for smaller team already (10-500 ppl), it just takes time to enter large companies.
[1] https://notion.so [2] https://fibery.io
If Atlassian wants to be profitable, they need to replace their founding CEO's. Mike and Scott have done well to grow the business, but the required skillset is different from this point on. Mike in particular seems to be very distracted these days.
They want to hire 17000 more people: https://www.bloomberg.com/news/articles/2022-04-07/atlassian...
The problem is you can’t much go ala carte - you have to go all-in.
It still doesn't have native Markdown editing, does it? Major issue IMO.
It has been years, since I wrote in their little feedback box, that it sucks and does not properly support Markdown. They do not give a damn and they are not going to fix their issues. Best to look elsewhere for an acceptable wiki experience.
Simply put, Atlassian sucks at building products because their culture is not good. They got lucky they had the right idea when they were young and didn't have competition.
Nowadays there are just better products out there.
That, kind of like Mozilla started fresh with Firefox, would indeed be a killer move (edit: provided they _can_ structurally do so).
That's just plain nonsense.
I dislike using Jira as much as anyone, mainly because it's so excruciatingly slow.
But there is no competing product (suite) that offers comparable functionality and provides the same long-term flexibilty. Many companies have built entire workflows (not just development related) around Jira, and have huge knowledge bases in Confluence. Switching would come with a lot of effort.
I will say that most of the Atlassian products are ... not great.
Bitbucket has completely stagnated and is way worse than Github and Gitlab. They actually had an opportunity there. Once upon a time Bitbucket was the only site to offer free private repos, and it attracted a decent number of users to also publish open source there. A good amount of companies adopted it. But they let it linger and didn't keep up with Github or even Gitlab. Just like Jira it's also slow and awkward to use. (never do large PRs ...).
I see a lot of companies switching away from Bitbucket now because developers are not happy with it.
Bamboo ... meh
Opsgenie ... meh
Confluence is actually fine. If only it had better search and wasn't so slow...
And Jira is so complex, improving it is probably extremely cumbersome because so many features have to be considered and everything has to keep working.
It might be advisable for Atlassian to start fresh with a new ticketing system.
Anyway, they have 8000 employees... they can always drastically downsize and refocus, the company will be around for a while.