2 ways to deal with big and complicated features
From feature vision to stories
One of the challenges of managing software projects is delivering big features. We already told you that we prefer to work with small, prioritized and unassigned stories that anyone can take and finish. But how do any up there? Especially, if what the customer comes to you with, is a big document, describing the features that you must implement.
Create tickets upfront
You can create all the tickets upfront based on the document you received or vision of the customer.
This is sometimes possible. When the change or feature request by the customer is in the scope of week work and unlikely to change, it might be worthy to spend a little time, think about how to reach the goal with small steps, and extract all the tickets. The benefit is that from now on, other people can just take the first task and start working. Perhaps some of the tasks are independent so you can prioritize them with your customer and have a very clear path to your goal. When all the tasks are visible, it is easy for everyone to see the progress.
But this is often impossible and impractical for a really big tasks that are going to take longer than a week. But before we dive into another strategy, let’s talk for a moment what we want to actually achieve.
Expectations and assumptions.
We don’t want to assign one programmer for 3 weeks to work on a separate branch in a complete isolation from the rest of the team. Instead we believe that having multiple people working on the feature is beneficial because they provide fresh view and feedback about the code and feature. It also improves Collective Ownership which we care deeply about.
In the spirit of Agile we don’t want to deploy this feature after long time of working on it. We want to build it iteratively and deploy often. We will seek the feedback from the customer and from the users of our software. We want to continuously deliver value. And if the customer decides to change the priorities and focus after 10 days of working on the feature to something new, that can possibly bring greater value, then it shall remain her/his right. In such case we would like everything that has been done and deployed so far to be usable and beneficial to the users.
When the programmers implement small parts of the feature we want them to have a good overview of it. Although when they work on a small ticket, their responsibility is to implement the small part enough to mark the story as done, it is also their responsibility to make the solution friendly to next programmers implementing further stories related to the feature. In other words, to implement something small, but have in mind the big picture.
And that brings us to the second strategy. We call it Documentation as floating ticket.
Create the tickets as you go
You keep the big feature (specification) as a ticket in your backlog. But whenever you reach it as part of take the first task rule, instead of starting to work on it as a developer, you put your project manager hat on.
You look into the specification and compare it to the current state of project. The specification is also a document that can be changed by everyone so you can see what parts of it have already been extracted into tickets in the past and what still needs to be done. Based on the priorities mentioned in the ticket or based on your conversations with the client it is now time for you to extract tickets. How many of them? That’s up to many factors. Maybe you can see clear path that can lead to having nice feature from the documentation so want to extract few small tasks. Maybe you know that the client want to keep working slowly on these features so it is ok to only extract one story from it. Whatever the reasons are, make sure the strategy is discussed with the customer and your team and the rules are clear for everyone.
So you decided to extract two task. You create them on top of your backlog so that you or your team mates can start working on them. Now based on similar factors described in previous paragraph you need to decide how to reschedule extracting next tasks from the spec. If the spec is very important and everyone should be working on it, you can leave it in place so that whenever currently extracted tasks are finished, you will reach the document again and mine new tasks from it, until the spec is fully implemented. If however you want to work on multiple parts of the system at the same time and customer expects progress also in different parts of the app, you should move the spec down in the backlog. The least important spec, the lower you can move it down. You can have strict rule how much it should be moved or you can do it based on your judgment and knowledge about project priorities. If everything from spec is done, mark it as done as you do with the rest of the tickets. If you get the knowledge in the meantime that there are more important issues and your team should stop working on features listed in the spec, then it might be a good idea to put it out of backlog, until the customer decides to bring it back in the game.
You end up with extracted tickets, spec knowledge, updated document about which parts are already moved into tickets and which parts are still only covered in the spec. You can go back to using Take the first task strategy until you are out of tickets, at which point you need to find your project manager hat and wear it for at least a moment again.
Get to know more
I hope reading this article was beneficial for you. If you want to find out more about techniques that will
- help you improve Collective Ownership,
- deliver value to customers,
- have a nice and steady workflow,
- allow you to refactor more easily while working on the feature you were asked for.
please sign up now to the newsletter below. We will email you from time to time about other techniques that work for us.