Technical stories are a controversial topic in the agile world. If you google technical stories antipattern, you’ll get plenty of people advocating never to do them. On the other hand, some teams religiously outsource any refactoring onto technical stories.
Should we eschew technical stories altogether? Or should we embrace them and put all of our (technological) dreams and hopes into them?
I believe this kind of story belongs in a healthy backlog, as long as we don’t use it to hide our corpses under the rug.
What is a technical story, anyways?
I’ll quote myself for this since there isn’t a universal definition.
A technical story is one where the main stakeholders are the developers in the team.
The end-user benefits only indirectly from such a story. As our system becomes better, the theory goes, we’ll have an easier time delivering actual user stories.
Are technical stories an antipattern?
How hard could it be?
Back to the original question. To answer it, let’s picture a user story as an iceberg.
The visible part is above water. Your stakeholders probably won’t care much about what happens underwater. Yet, the complexity hides below. Most of the effort will be spent there.
The problem with technical stories comes when they are used to draw an artificial line. We do the visible part, then extract the rest into a technical story. What an improvement in our velocity!
Except, this is a crucial mistake. We’re avoiding essential technical work in a misguided effort to move faster. Striping a user story of this bottom part is the antipattern. It leads to an endless list of tickets in the backlog that will never happen. It encourages cutting corners and short-term thinking.
If the story is too big, let’s agree on a smaller scope that still delivers meaningful value.
Why do we extract this work when we know it’s so important?
Does this happen in your team? You’re not alone. Many teams struggle with this. There are many reasons, among them:
PO and Dev align expectations
Avoidance of conflict. If the product owner and the developers push in opposite directions, there’s tension that needs to be solved. Avoiding the conflict solves nothing, but it’s the path of least resistance.
Delivery pressure. Yes, features almost always take priority. Technical work is a catalyst for features. There might be real urgency, in which case it’s worth talking about managing the created tech debt.
Lack of empowerment. Developers sometimes feel that they can’t stand for what they believe is necessary, or that their voice won’t be heard. Maybe they’re genuinely not being listened to, or perhaps their voice is too quiet.
Let’s be honest; these are symptoms of deeper problems than writing some stories (or not). However, keeping the platform healthy is in everyone’s best interest.
Then when do you create technical stories?
I thought you’d never ask. The most natural place to create technical stories is the dev huddle. As a result of a successful dev huddle, the development team will agree on certain experiments, refactorings, or changes to be made. Any action big enough is a prime candidate for a technical story. To name some examples:
- Refactor the layout of our product cards to use flexboxes.
- Let’s try out strikt, a new assertions library.
- Refactor our API calls to use React hooks.
Once there is an agreement in the team, reflecting these points in stories helps build the team’s technical vision. These stories will (hopefully) happen soon, and increase the quality of the system. That, in turn, means that the iceberg’s visible part will have an easier time floating, which will lead to more fancy features for our happy stakeholders.
Monitor the amount of technical stories in your backlog!
The number of technical stories is a useful fitness function. If that number only grows and grows, this whole process is not working very well. If anything, it can even discourage further attempts to improve the situation.
The tenets of a good technical story
Let’s say you grudgingly agree to start writing technical stories. How do you do it? This is the kind of story that will likely be written by a developer. And, to be fair, many lack the experience.
Like everything else, good writing takes practice
Poorly written technical stories are sadly all-too-common. They are dense. They don’t get to the point. They are vague on the provided value. Because of this, they languish in the backlog. That leads to important topics not being addressed. This cycle is hard to escape from.
A technical story has to be held to the same standards as a user story. A story with a one-line description like Upgrade Rails to a new version doesn’t cut it. It’s unfair to demand that user stories are descriptive, complete, and clear if that won’t apply to the technical ones.
There are many resources on how to write user stories in general. There is even an acronym, INVEST. Follow these practices while using your best judgment. Here are five suggested sections based on my last five years of story poetry.
- Context
- What's the value
- What to do (Acceptance Criteria)
- Out of scope
- Tech hint
I’ll explain what I think should be part of each, plus some bad and good examples.
Context
Where does this story come from? Understanding the background prevents misunderstandings. It allows us to understand if some parts are more critical than others.
❌ We want to use flexbox now.
✅ When we started this application, we had to support older versions of Internet Explorer. Support for flexbox is limited for them, so we decided to build our product cards using a more traditional inline-block layout to avoid incompatibilities. This requirement has been dropped, so we’re free to switch to a more modern technique.
What’s the value
Technical stories are in a perpetual fight for survival. At the end of the day, it’s the product owner who controls the backlog. Stories without clear value get ignored in favor of user stories that bring a tangible benefit. Getting hard numbers is ideal, although not always possible. A qualitative judgment of what we want to improve can be helpful enough.
❌ I like flexbox a lot.
✅ Using flexbox will lead to a simpler CSS structure, resulting in less maintenance effort. We expect less visual bugs and less difficulty with typical tasks like centering elements.
What to do (Acceptance Criteria)
Yes, this seems pretty self-evident. Still, you’ll find stories that are neither actionable nor verifiable. Is it clear what we want to do? How will we know when we’re finished? Can somebody in the team that didn’t write the story pick it up?
❌ Fix the product cards.
✅ Replace layout based on inline display with flexbox for product cards.
Out of scope
Technical stories can be very broad. It’s useful to be explicit if we won’t touch some parts that folks would expect to be part of this story. In the end, you want small, incremental improvements. Not a complete rewrite.
✅ The customer preview uses a similar layout, but we won’t touch that one until we’ve finished changing the product cards.
Tech hint
Help your teammates by giving them some guidance. If it’s about upgrading React, what did we check already? Are there some pitfalls to take into account? Usually, there was a prior investigation, which is invaluable information that we don’t want to rediscover.
Be mindful, however, of not writing an instruction list. Nobody wants to be ordered around. Moreover, writing down a very detailed list of steps can be as much effort as doing the story in the first place.
An exact blueprint will prevent so many needless discussions
Technical stories are another tool in the belt
Technical stories have helped the teams I’ve been on managing evolution and improving. The two fundamental points that I want to highlight again are:
- Don’t strip the underlying complexity of a story and put it in technical stories that you “will just do later.”
- Do yourself a favor and treat technical stories with the same respect and care as you do for user stories.
Notice that I haven’t talked about the tension between including user stories and technical stories in an iteration. That relates to building a technical backlog and assessing the health of a system. I plan to dig into that on its own post.
Thanks to Anna, Rachael and Felix for the feedback.
Top comments (3)
Great article!
Would going into detail here be worth another blog post for someone who doesn't know where to start?
Really looking forward to this. The team I am currently on does tech debt days once every two weeks, which is better than completely ignoring it (this was the case before), but I find it severely lacking.
I track the technical stories in the backlog and see if the number is growing over time, nothing very scientific.
Getting tech debt stories into the sprint is a hard one. I think my biggest success has been when my PO trusts me and I can pull stories into the sprint without too much discussion. I've tried variations of reserving a budget for tech that don't always work that well. It's a topic for a whole article, though!
Ah, that easy! Thanks :)
That makes sense. Looking forward to the article!