What I Learned from Domain Modeling in a Team

What I Learned from Domain Modeling in a Team

Introduction

In my division at Guild, we've started a working group to develop and iterate on a new domain model, aiming to create the ideal topography for the next evolution of our space. The process incorporates zero-start thinking (starting from a blank slate) and considers inputs such as what product features we want to support now and in the future, how to design the domain to easily adapt to new requirements, and how to align the domain to maximize ownership and the flow of work. The spirit of the exercise is rooted in Domain-Driven Design. Using this model, we aim to provide a North Star of sorts to guide design decisions, trade-off discussions, investments, and what directionally correct choices look like.

The group consists of leaders in the technology and product organizations. I'm excited to work with them on this large project and learn from them. I work closely alongside the Software Architect in the group, helping to author the model using feedback from stakeholders and our own knowledge and understanding of the domain.

Communicating with Stakeholders

With a task as complex as designing a domain, communicating with others is crucial. Failure to do so dashes any chances of achieving the project's goals. Nobody truly has the "whole picture" in their heads. Getting other people's perspectives on the domain, listening to their concerns with the model, teasing out information by asking thought-provoking questions, and gathering requirements are just some examples of the process of effective communication. If you can do this with people in various roles (especially in Product and Engineering), you increase your chances of success. Just be sure not to get stuck in analysis paralysis.

In our group, the revisions of the model have happened in various ways. We've done some small iterations in synchronous meetings, I've facilitated one-on-one sessions with stakeholders, and we've also collaborated asynchronously using Miro. Getting feedback early and often helps reduce the chances of creating something that doesn't fit our needs.

Making Brave Decisions

Essentially, all models are wrong, but some are useful.

- George E. P. Box

As my career has progressed, I've taken on more responsibility and decision-making, which can be uncomfortable. I often wonder, "Is this decision the right one?" During this domain modeling exercise, I create concepts based on my understanding of different contexts, and that question frequently arises.

In a conceptual model like the one we're building, where the time horizon is at least a year out if the model had perfect clarity, the answer may not be known for a long time. Even then, by the time we get there, the definition of correct may be entirely different from what it was before. My VP of Engineering gave me great advice though: "Make a decision and get feedback on it. If it doesn't work, we've eliminated one incorrect option." This made me realize that though I may be drawing some boxes in the model, the overall decision and direction belong to the team, not a single person. I shouldn't be hung up on being right. The point of the model is to help inform and provide context, not to be 100% accurate in all regards. Modeling is cheap, and we should be comfortable throwing away ideas now, when it's easy.

Balancing Ideal vs. Current State

As I stated earlier, the modeling exercise uses some of the current functionality we want to support as an input. Another is future functionality we want to support. It's hard to keep the former from influencing your thinking too much. These are things we've built; it's easy to reason about and understand them since they're real and these concepts already exist. Striking the balance and allowing ourselves to question everything is key to envisioning what could be. Yes, sometimes we keep some domain objects from the current state around because they make sense, but other times, we take the scary step of not including those objects in the new world. Envisioning what could be is the goal, and we should use the past to inform the future, but not couple them.

The Ever-Evolving Model

I've never been part of a project where the final delivery turns out exactly like the initial design. The completeness of the model is not 100%, and the time horizon for any kind of implementation is long-term. Revisiting and evolving the model is necessary to keep it relevant and useful. Contexts will shift over time: new features will be necessary to accomplish the company's goals, domains may become deprecated, and we'll keep gaining understanding of our product's sector.

After the initial model is built, one of the follow-on efforts will be to see how we can get from point A to point B and do that in an iterative fashion. Think of things like product release milestones or an architecture modification. We should use these checkpoints to pause and revisit the model. We will have certainly learned more along the way, and building that understanding back into the model will adjust it to be more "correct" than it was. I've seen many diagrams and models that are made once and never updated again, reducing their usefulness and potentially leading to suboptimal decision-making.

Continuous Communication and Adaptation

Even though the modeling is done by a working group, the outcomes affect a larger set of people and the systems they work on. Shopping around the model, setting up its context, and getting people's thoughts on it are important for gaining buy-in on the effort. In an organizational culture that emphasizes ownership, "throwing things over the wall" is a mistake. I personally am still learning how to do this well. Who do I share information with, and at what time? How do I make sure their input is heard but also not distract them from their current priorities?

I don't think there's a one-size-fits-all answer for this. Generally speaking though, I think it's hard to over-communicate, especially in things that are high in scope. There's probably nuance here with the particular people you work with and organizational structures. That pushes more importance on staying aware of how communication flows within the organization and doing what you can to ensure everyone is on the same page.

Conclusion

In conclusion, the domain modeling project at Guild has underscored the importance of clear communication, decisive action, and the ability to adapt to changing conditions. This work has taught me to balance current realities with our future vision, continuously refine our models, and engage our entire team in the evolution process.

Have you been part of large-scale domain exercises like this? What strategies did you employ to achieve your goals?