« Scott Nonnenberg


Customizing Agile

Ever since the term Agile started to take hold in larger organizations, there’s been a growing backlash among developers. Sadly, this is to be expected. As I pointed out in my last post on Agile, people want a set of rigid best practices, not a true new way of thinking.

So, how might we save it? I’ve already talked through the reasons for the standard practices. Now that we understand it, we can change it. Let’s exercise a new muscle: taking a standard cookie-cutter plan and tuning it for your specific team, project, or organization.

Definitions

First, some definitions. We’ve been using the word Agile a lot, but what does it mean, exactly? Where do terms like standup and sprint come from? Our first clue comes from agilemethodology.org, the first result from a search for ‘agile.’ It is sure to differentiate between Agile and Scrum. But it’s interesting to note that Scrum comes up after only four sentences, and is the same size in the tag cloud on the page. And the term ‘sprint’, mentioned under the Agile header, is actually a Scrum term.

These two concepts are wound together so tightly it’s hard to tell the difference.

What is Agile?

It’s key to understand that none of the practices generally associated with Agile came from it. And that’s because Agile endorses no specific processes whatsoever. The term came to prominence via a 2001 meeting of industry luminaries which resulted in The Agile Manifesto:

"Individuals and interactions over processes and tools
Working software over comprehensive documentation
Customer collaboration over contract negotiation
Responding to change over following a plan

That is, while there is value in the items on the right, we value the items on the left more."

Yes, Agile does value incremental improvement and regular checkins, mentioned in its more detailed twelve principles. But it does not specify formats, interval lengths, or names. Those largely come from Scrum.

What is Scrum?

Surprisingly, Scrum predates Agile. First appearing in a 1986 paper, it was refined and branded for a subsequent 1995 OOPSLA paper: “SCRUM Development Process.” In contrast to the Manifesto, it’s interesting to note the term ‘process’ in the name of the paper. The familiar terms sprint, Product Owner, and backlog all make an appearance. Other familiar terms, like Scrum Master, sprint planning meeting and impediment do not. Those terms are new.

The additional terminology came from businesses built up around the original concept, providing books, trainings, certifications and conferences. I took one of these Scrum trainings in January 2010, thanks to Microsoft. The way it was presented, it was very easy to think of it as the Holy Grail solution to Visual Studio’s long waterfall-style releases. So you can understand why it has taken hold in larger companies.

But it’s key to understand that Scrum is a process, with rigid roles and predefined meeting schedules. Lots of time is spent debating as to what role has what authority, appealing to some imaginary ‘true’ definition.

So let’s really practice Agile, and value:

Individuals and interactions over processes and tools”

Change for the people

Let’s talk about a few potential changes to the standard Scrum-centric way of doing things, and what kind of effect it might have on your project. We’ll start with small changes, and work up from there…

No subtasks

Yes, pretty much everyone solves complicated problems by breaking them down into smaller chunks. But not everyone does it consciously. In my experience, asking developers to put their subtasks into the system provides them absolutely no value, while also making them wonder how you are using that data. Is management trying to make performance decisions based on it? Scheduling decisions? That’s not a good idea, nor is the mere suggestion of it to the team.

Try taking a break from this requirement. I’m sure your developers have been asking for this. If they haven’t, here’s a scary thought: maybe that’s because they’re afraid of what you might say!

Allow multi-sprint efforts

The whole of software development doesn’t fit into two-week chunks. If you try to force everything into your sprints, you’ll miss out on a whole category of potential great work.

If a developer, in tune with business need, says they need a while, come up with a plan to enable that. Your instinct will be to try to break it up into smaller milestones and deliverables, but deep creativity comes from empty time. In this case your periodic checkins should be less focused on material productivity, and more focused on conceptual productivity.

No cost for tasks

Sometimes developers get frustrated with the time it takes to come up with costs for tasks. Or it seems pointless. Or you’re having trouble with consistency.

There are a lot of reasons you might want to experiment here. Try removing all costs from your backlog and sprint tasks. Instead, you can focus entirely on splitting them up to be of similar size. Once they are all comparable, you have the same visibility into velocity you had before!

Note: You might end up with a kind of bi-modal solution. Full feature-size items and small ‘fit-and-finish’ bugs. It’s a good compromise when you have bugs coming from outside the development team.

No velocity

What you measure is what you are focused on, what is most likely to improve over time. What if your only overall metric was a survey of the dev team asking “Did you perform at your potential for this sprint?” Or how about “Were you happy in your job this sprint?”

I’ve seen a very clear correlation between developer performance and overall mood and job satisfaction. Developers very frequently have very high standards for themselves. At the very least you’ll likely get quite revealing justifications for “no” answers.

Temporarily drop retrospectives

Retrospectives form the backbone of a feedback loop that will gradually transform your development team into the best it can be. But it does take work to address the impediments raised during these meetings. Without that, the loop will never close and the benefits of of the meetings will be lost.

Crunch time happens. If you know you won’t be able to work on impediments in the face of a big milestone, temporarily suspend your retrospectives. But know that you will be building up debt. Keep track of the pressure building up on the team.

You might be tempted to think the feedback you still get without retrospectives is good enough, but it’s shadow of what a healthy retrospective system will elicit. Dedicated retrospectives, with proof that raised impediments are addressed, send a message that feedback is important.

No initial sprint plan

Teams sometimes get themselves into a paradox. They want to sign up for a certain amount of work every sprint, but they can never quite do it. They always fall short. Sometimes it really is because of some unpredictable external factor like emergency production bugs. But most of the time it’s plain overoptimism. Deadlines can make people do unexpected things.

Try eliminating your sprint plan. You’ll work directly from the backlog. Assure the team that progress is very important, but you trust that they’ll work at the speed which is right for them. Here it helps to connect the dev effort with the overall organizational mission to provide the urgency now no longer provided by arbitrary deadlines.

This might actually be harder on your Product Owner, because you’ll want to get a little bit further ahead of the team than usual. Hey, it’s possible that the team will come alive without as much pressure!

No sprints

If discarding your sprint plan works for your team, you are already very close to eliminating sprints entirely. In this approach the team works continuously, pulling from the backlog whenever ready for a new task. The Product Owner is always working slightly ahead of the developers, making sure the backlog will be ready for them. You’ll find that this is similar to the stripped-down Kanban Method.

Of course, even with no sprints, with Continuous Delivery in place, you might still find that demos and retrospectives will happen roughly on the same schedule as before.

No meetings

Something about regularly-scheduled meetings breeds apathy. It’s fair, of course, because that standard slot makes attendees feel that they have no power over whether it happens. One easy step might be to say to your team: “if we get X done, meeting Y doesn’t have to happen!” It will force you and the team to understand the purpose of the meeting.

Meetings are especially painful for developers, whose coding flow is broken by any interruption. The frustration they feel starts there, and when they perceive that their value added or received at the meeting is low, they’ll question its very existence. Try waiting for the team to request a meeting. You’ll figure out what the right cadence should be.

When I was a project manager for the Visual Studio Languages team, I built a weekly status system with email templates that worked really well. We had meetings every Friday morning to go over the information, but it was no longer necessary. I asked around in preparation for scaling it back, and was surprised: the team really liked the meeting. They saw it as a method of getting together once a week, getting face time with people they didn’t normally work with. That’s a far better purpose for a meeting!

Maybe make your one standing meeting on Friday afternoons, with cool demos, tasty snacks, and alcohol! :0)

The Theme

You may have noticed that there’s a theme behind all of this: trusting the development team. If given the tools to succeed, it will. Your developers want to do well. Listen to them!

As the Tao Te Ching says:

“The world is ruled by letting things take their course. It cannot be ruled by interfering.”

Your task is to build a system which will naturally push your developers towards success. Give them a chance to get to know your customers themselves - get rid of your layers of analysts. Ensure that they know the business well, especially the causal lines between their actions and business success. Clear the path ahead of them. Eliminate anything holding them back. Encourage experimentation to settle disagreements or to break through analysis paralysis.

Now imagine a system of quick development and immediate feedback: feature flags to allow features under active development to go to production, Continuous Delivery automatically pushing every merged pull request to production, and near-immediate direct customer feedback after every deploy. Your developers are working in a quick feedback loop with customers, understanding better and better what features and behaviors are valuable.

You can start to see how the software would naturally get good, really fast.

What’s best for your team?

All we have are intuitions, built up from experience over the years. But your experience may be limited, and this kind of stuff is frequently unintuitive. Try things out, see what happens! And above all, listen to your teams. If they have a recommendation, try it out! Adapt!

As The Agile Manifesto says, we value:

Responding to change over following a plan”


Does that customer-focused iterative development seem unrealistic to you? I’ll be first to admit that true agility requires some key technical tools and practices. Check out my next Agile post: The technology side of Agile.

You might be wondering why I’m so interested in this. I have a unique perspective. I’ve spent a lot of time in a number of software development roles:

I won't share your email with anyone. See previous emails.

It's me!
Hi, I'm Scott. I've been in software a long time, and during most of that time I have worked to bridge the people/tech divide. Contact me if your company needs training, coding, or just an experienced outside perspective!

NEXT:

ESLint Part 1: Exploration 2016 Jun 14

I recently spent some time to review my ESLint setup. It’s got lots of rules, and a healthy ecosystem of plugins. It’s intimidating! Maybe I can help you make some sense of it… Read more »

PREVIOUS:

Contract: New techniques, old technology 2016 May 31

Last fall, after having worked with React or similar technologies for a couple years, I took a contract working with Marionette.js. A much older technology, based on Backbone.js, it was uncomfortably... Read more »