Lessons from a successful software project

July 14, 2014 ∞

In my years of working with software projects the one thing I have learned is that doing software projects is really hard. Probably zero projects I have been part of or heard of has delivered their value, kept the quality to a standard everyone is proud of, launched on time and had a team of happy people through all of it. Recently, though, I was part of project that did just that. An actual real life professional software project that was deemed successful by both the team and the stakeholders.

I work at Swedish real estate site Hemnet. We are a company of about 50 people with about 15 people working in product development — mostly developers plus a few designers. Being a small company means we are close to management and all fill a bit wider roles than in larger companies. The biggest challenges in our organization has been making priorities and following them through. This heavily influences where we have put most of our effort and where we have made the biggest lessons making, though I think most of it can be pretty broadly generalized.

The team consisted of one UX designer, one graphics designer, one business analyst, one dedicated developer and me as team lead and part time developer. With the team of 5, we switched map providers, created a solution to render 100 000 hits on a map simultaneously and designed and implemented an all-new view for searching on maps. We did this in just 6 months and, from the release date set in the beginning of the project, slipped only one week with being fully rolled out.

Before we get in to what did matter, let me point out that you can Scrum, Kanban or Scrummerfall all you want without getting much of results. It’s easy to try to change the internal methodologies of a team because that’s something you’ve got complete control over. The biggest challenges though, are in getting the right conditions for the team and within the organization. Only after that is settled does methodologies begin to provide real value.

Here are the biggest things that made the project succeed.

A team with power to do its job

This is where everything begins. Way too often, projects are spread out over departments and even split in sub-projects over different departments. The business side is done in one project, handed over to a design project and then handed over to a development project. And even if that is not the case, people are still spread out without good means to focus and communicate on the project. This completely kills the common understanding and collaboration within a project.

Our team was created with exactly one premise: the goal of the project. We were to create the new map search experience on Hemnet. How would it look? Find out. How would it work? Find out. What would it even do exactly? Find out. This was absolutely crucial in enabling us to do our work — we are product developers and excel in solving problems. Based on the goal, the team formed the plan to do it, the concrete requirements and features and did all the design. Except for getting the rest of the company to buy in on our plan, we were never in need of someone else’s decision.

This of course requires the team to have all the competencies needed to get the job done. Having a team with a common understanding from start to end has incredible value in ensuring everyone works on the same thing and towards the same goal to create a cohesive product. In our team, having our business analyst from start to end allowed us to not stray away from business goals, and having everyone with from the beginning ensured we all really understood why the heck we were in this to begin with.

One of the factors we all appreciated the most was being all together in a room dedicated for the team and the project. If you are in an office, the change from being just one room apart is a huge boost to the social aspects of the team and the collaboration made possible by putting sketches up on the wall and just leaning over to talk about something is invaluable.

Making sure of a stable team with the right conditions is where everything begins.

Awareness of what could go wrong and preparation for it

Things always go wrong. The problem is that we more often than not ignore this and just plan for the best. This results in everything going to pieces when an assumption turned out to be wrong or an unforeseen problem arises. As a result, projects usually hit a point where everyone gets uncomfortable over something that went wrong and starts arguing over how to proceed — a situation where exactly nobody has their best ideas.

We spent a lot of time in the beginning of the project trying to figure out where things could go wrong. Being aware of the risks is key to being able to react in an intelligent way when something eventually goes wrong. In our case, we identified where users were most likely to have a hard time with the change and what business related key measurements we were most likely to impact. We completely shaped the project around the risks and made sure everyone understood and were prepared for the things that could go wrong. This helped us adapt continuously and completely avoid disasters, both in the product itself and among people in the company.

The plan we made focused on the key features that would be realized in the project on a higher level. This allowed us to cut the right corners in times when time started getting sparse while still delivering the value promised. We also spent time discovering where scope creep would most likely pop up and discussed these points beforehand. The things we agreed on not doing were then part of the plan. Designing the plan in terms of general features and including the things we would not do was a key tool for us in delivering the project on time.

Plan as much for the risks and the things you won’t do as the things you will do.

Collaboration outside of the team

All developers have been in projects where requirements and conditions change over their heads. A manager changes his or her mind on something too late in the project or another project gets important and starts snatching people. This is one of the most effective ways to demoralize a team trying to make something great and is probably also the most common way I have seen projects go out of control. The problem here is that managers often have no ability to make sense of cryptic scrum points, refactorings and vague descriptions of what is done and what isn’t, which forces them to make decisions with effectively no information on what they are deciding on.

First off, we put a lot of effort in communicating the plans for the project. We made sure everyone with the power to change our faith were fully aware of the plan, including what we would do, what we would not do and what we needed to complete the project. We did this early enough that we still could change everything and made sure that all stakeholders had their say before nailing it.

During the project, we continuously discussed any change that would affect stakeholders with them and held recurring demos open to anyone in the company. This allowed us to work without dropping anyone’s perspective. We had a number of small changes come out of this that would have gotten into big problems if they wouldn’t have been revealed until the end of the project.

Keeping managers and stakeholders in the loop is key to having the project stay it’s course.

There isn’t ever a silver bullet with these things. These were the biggest contributors to this specific project being a success. Even though I have reason to believe they would help significantly in many situations, a project always have to start with the people and environment where it will live.

We have the brilliant Thomas Lindquist as our full time coach working on process and people since two years. This project is the result of those two years of experimentation and much of the reasons why it succeeded are the brainchildren of Thomas’ continued thinking on the matter. This may seem too obvious, but the best way to get better at process is to spend time thinking about it and improving it.

If you are interested in more on this particular project, Thomas has written on the “post-mortem” retro we held with the team, our graphics designer Daniel Feldt has written on his perspective of the project and our UX designer Magnus Burell had a talk on the UX side of things. And for the more technical side, our map-oriented developer Igor Tihonov will talk on this year’s FOSS4G in September.