Why Trello doesn’t work for bug tracking
Via flickr user goingslo
In my post on Trello I briefly mentioned how the software doesn’t work well for bug tracking. I wanted to dig into that issue a little deeper.
Bug tracking and bug fixing require much more process automation compared to feature development. This is primarily because of the large number of cases that will be open at any moment in the development cycle. For our company, we churn through between 40-60 bugs in a given week when we are focused on bug fixing. By comparison, when we are really cranking on features, we might do somewhere around 4-8 in a week. That is a fairly significant difference and the core reason that Trello doesn’t fit well for bug tracking. The overhead associated with managing the workflow of those thirty extra cases becomes a nightmare without a more rigid process.
Our process is fairly simple, we have a triage list where all new bugs go. We have a working list which contains the bugs we intend to fix in the upcoming iteration (an iteration is usually a week long). We also have a backlog list for bugs we intend to fix “someday” but not right now. We will periodically we go through the triage list and move bugs into either the working list or the backlog list. Bugs are fixed, developers resolve the bugs and they go to QA. QA validates the fix and closes the bug. Test automation will then decide whether to add this bug to their list of automated tests.
We use Fogbugz and find it works great for this workflow. It has enough process automation to make sure that cases flow smoothly through the system, but it isn’t so heavy handed that we can’t tweak the process when necessary.
Why Trello doesn’t work for bug fixing:
First is it’s lack of bulk editing features. If I want to take all my critical/show stopper bugs and move them into the working release, I will need to move all of them over by hand. Since moving by hand in Trello is literally moving by hand (via the mouse) this is actually a large amount of work.
Trello is at it’s best when you only have 6-8 items in each “active” list. When lists grow larger, they are very quickly pushed off the screen, making scanning and locating cards difficult. Some weeks we will ignore the bug list while development is working on a new feature and QA is testing something that was just handed off. If we were using Trello to manage these tasks, the “bug” list would grow quite large and very quickly become unmanageable.
Trello doesn’t have great categorization and search features. We categorize our bugs pretty heavily by priority and area. So for a given product and milestone we may have 100+ bugs we intend to fix across a variety of functional areas and at various priorities. Also, when we start working on features in certain areas of the product we will also look for any outstanding bugs or feature requests in that area to see if we can lump them into the upcoming enhancements.
Trello doesn’t have good screen capture integration. While it does support attachments, these aren’t enough. Fogbugz integration with snagit and their own desktop screen capture tool are great accellerators for entering bugs. Sometimes a picture is worth a 1000 word bug description.
How Trello could be made to work with fogbugz:
Don’t get me wrong, Trello and fogbugz integration would be awesome. I would love to be able to turn features into cards and then mark the cards as “finished”, automatically resolving the corresponding features in fogbugz. This would likely necessitate a stricter board setup, but being able to visualize feature requests this way would be a nice addition. We are essentially doing this process manually today anyway.
The ability to grab all of the bugs in a given milestone or assigned to a certain person and create a card that has those bugs as checklist items would be an nice way to help developers prioritize and manage the list of bugs they are working on. As a manager I could highlighter 7 or 8 critical bugs that need to be addressed and make sure the visibility is raised through the creation of a card. This would let me do away a little bit with our “working” list in fogbugz and instead replace it with cards in the current iteration. Not a blindingly brilliant enhancement, but again, we are doing something similar now. We make the card in trello “Fix the bugs on the list” but I have to check 2 places daily to know how the developers are progressing on their tasks.
I can only presume that Fogcreek is working on ways to integrate Trello with Fogbugz in the future and I look forward to seeing what they come up with.
Not enough bits in this byte 
I’ve been using Trello for web dev bugs for about 2 weeks now and found it pretty good, but not perfect. Like you say, multiple selections at a time would be great and SnagIt would be great for quick bug uploads.
For free though, I’m pretty happy and can work around that for Fogbugz $25/user fee considering the number of users that we have varies from 2 to 22 depending on the project.
I would agree Ben. We have an install base north of 600, so we tend to push a lot of bugs around, not the best use for Trello. On smaller projects I could easily see Trello being adequate for bug tracking and fixing. Using a separate list for high profile bugs like the trello on trello board does is very feasible for projects until they reach a certain size.
So start with Trello, until the list becomes difficult to manage, and then start looking for another tool to integrate with…
I think Trello is better suited for task lists rather then bug lists. Instead of listing all the issues list all the improvements or maintenance work. I would use specific tools for tracking bugs. such as Mantis or bugzilla. These have automated features that can hook into re-visioning or wiki systems.
I’ve used Trello for over a month now and I find that it works very well for new projects. I have a card for each project and a series of checklists for each group of tasks required to complete the project.
I’ve written a blog post on the six reasons why I like Trello. Feel free to take a look and pass comment. It would be interesting to hear how other peolpe use Trello for Web Development.
I do not really see the difference between a bug and a task. Fundamentally they are “work to be done”. I think life is complicated enough as it is. I have only two comments about Trello.
1. Milestones. Is there a way to make them visible and visually display that we are approaching one?
2. The cards are too cluttered. It is trying to be design tool and all sorts. Simplicity in the new beauty!
Apart from that, good work and good luck 🙂
Good points. One feature I like about Trello is the fact that you can covert individual bullets in a task list to a new card.
I create one card for each iteration of my software called “NAME 0.8 bugs” and make a checklist of bugs found. My developer moves that card to “Doing” if he is working on these issues and ticks off each thing he fixes. If one bullet requires attention he creates a card from it so we can discuss it more.
The only feature I miss is the ability to bake QA into this to verify if bugs are in fact fixed. One way to do that is to create a board with bugs only. You can have one list for reported, one for “being fixed”, one for “QA” and one for “completed”.
But yeah, keeping track of versions and so forth can become tricky quickly with Trello.
We keep separate list of bugs. One lane that becomes useful is “For Review”. This lane is before done so nobody can report things as done without somebody else checking it. This increases the responsibility in the team.
A bus from the buglist that gets the attention becomes at task. the task then goes all the way to the “For Review” lane and that is how it is verified.
Simple 🙂
We use Mantis for bug-tracking. But I’m wondering if Trello can take over our manual test sheets. Basically, have a test suite, and a list of test cases within that suite. Entries for our platforms that we test on… then for each case, a list of steps to perform, with the expected result… and a pass/fail/did not run tag for each case… Do you think it would work for this?
Hi Drew,
You might be able to use the card copying feature to create templates that have all your test cases as checklists on the card. For a small project this might work, but I suspect with a large number of tests you would quickly cross the threshold where trello’s organizational tools would work well. Trello naturally wants to limit how many cards you have in a particular area and test cases often get very very large over time.
The thing I like about trello is that there is very little harm in trying, but I suspect that for test cases you will simply end up with too much information to manage and trello will end up being frustrating.
I suspect you’re right. I had a quick bash at creating some templates last night, but just a bit too convoluted for my needs. Do you have any thoughts on something to meet basic smoke and functional test execution? We currently just use spreadsheets in Openoffice Calc. Have tried TestLink, but to be honest, it’s a ballache to work with and has some very strange limitations.
We have also been using testlink and I agree that it’s not the easiest to use. Unfortunately I don’t have any other good recommendations for test case management tools at the moment. I seem to only be helpful in giving advice on tools to avoid.
We’ve fixed the email integration piece of the pizzle with Trello via emelloapp.com. It lets you send your email straight to trello via a dedicated custom email address.