Arena’s Cultural Values at Work
By Vinicus Gravina Da Rocha
Originally published on Jul 24, 2017
Tags: Arena, values
At Arena, defining our culture is as important as the quality of the software we deliver. One emanates from the other. In this blog post I share some of our cultural values, using stories of real-life situations to illustrate how we interact with each other on a daily basis: how we go about decision-making, how we prioritize our work, and what we do when things slip through the cracks.
Development driven by client value and suffering
There’s an infinite amount of possible work we could do. We don’t have the luxury of spending time on anything but projects that will make us move forward as a company. Deciding what to do tends to be driven by client value and by suffering.
I believe the reader is familiar with the term “client value”, since it’s jargon in our industry, but “suffering” is an interesting one. By working on things that make us suffer, we are able to free up resources — time, cognition, money, etc. — to focus on things that will bring value to our clients. In a certain sense, “suffering” is a metric that indirectly leads us to value.
In order to define our priorities, all the work that needs to be done is analyzed and weighted against its value. It might seem like a lot of effort and a huge overhead, but it allows us to be confident that we’re building the right things.
When thinking about value, we try to gauge whether a new feature or product will add value to our clients. Will it help our clients to be more successful at their jobs? Are their lives going to be easier or simpler?
One feature we designed recently catalyzed the interaction between a large subset of our users and our product. We went onsite to hear from our users to understand how their daily routine was and how they were engaging with our product.
In those interviews, we realized that recruiting was just one of their many responsibilities and that just a fraction of their time was spent in such activity. This meant that learning how to interact with our product — and even their own recruiting systems — was a cost few were willing to pay.
Considering our users’ busy schedule, and their daily use of e-mail, we designed a e-mail report that summarizes our hire recommendations and predictions. The applicants are linked to their profile in the recruitment platform, allowing our users to quickly move forward with a recommended application. After the beta period and success at one client, we then began rolling this feature out to more clients.
Our feature bridged the gap between the users, their recruiting systems and our product. Needless to say, this feature proved to be a great success.
When looking at our processes and tools, we evaluate how “painful” they are, how long they take and how frequently they are used.
Is it worth the time? Source: https://xkcd.com/1205/
If our processes or tools are slowing us down, it might be time to revamp them. If it’s a one-time thing, it might not be worth spending hours automating it.
An example from one of our processes: data from different sources is ingested, cleaned and matched. This involves several mapping files that describe how the data is transformed and how the different operations are executed. For historical reasons, those mappings were described in the Comma Separated Value (CSV) format, similar to a truth table in which all combinations of input had to be explicitly listed. To make things worse, these files were manually updated.
As you can imagine, there were several pain points: describing all inputs was tedious, error prone and hard to review. Also, across files there was duplicated information that had to be kept in sync manually.
To reduce the entropy in this process, we re-designed our approach.
We designed a Doman Specific Language (DSL) capable of expressing such mappings in a succinct way, drastically reducing the complexity of the files. Moreover, we defined a set of files to be the single source of truth from which old format mappings are generated.
Some of the old mapping formats were kept, because changing them would break a lot of the existing code, entailing a lot of refactoring. Since those interfaces weren’t part of the problem, we focused on precisely what was causing us pain.
Mistakes happen. Blame the process, not the people
It’s part of human nature to make mistakes. Acknowledging this allows us to focus on tools and processes that will keep us on the right track, preventing accidental errors or mistakes.
Whenever something goes wrong, we blame the process, not the people, and then we try to improve the process. We have two distinct opportunities to improve ourselves and our processes that help us to become a better team and a better company: the weekly retrospective and the post mortem.
Every Monday, we meet to talk over the previous week, how we could have done things better, how we can improve. The idea is to build processes that don’t allow us to break things, so the process is running us, not vice versa. At the same time, we want to design a process that is transparent and doesn’t stand in our way. (See Suffering above for how we dealt with a process that became too cumbersome.)
When something fails drastically in our production department, we fix it at the time, however we can. Later, when things are calmer, we pause and evaluate the root cause of the problem. We use a process called the “Five Why’s” in a way that invites people to participate, not to blame or find a guilty person. And then we work to improve the process.
In one of those incidents, we had an artifact that was accidentally deployed to the wrong machine in production. This reverted software to an older version, with unfixed bugs. This happened because the deploy was manual and whoever was running it, ended up choosing the wrong build plan.
After resolving the issue, the team got together to investigate the root causes and evaluate how the process could be revamped to prevent such issues from happening in the future. The result was a design we called “Deploy by code”, in which we leverage our code review system to track, review and notify people of deploys.
Promote individual ownership
Software development is a complex feat. There are so many different flows of information and constraints: different requirements and use cases, technical debt, new technologies and libraries. We trust our people to work directly with clients and we value their ideas internally.
Direct client contact
As much as possible we promote the flow of information directly between the interested parties, with as little noise as possible. This empowers us to bring our own perspectives on how to solve problems, when to pivot and improve upon our iterations.
We trust our people to have conversations with clients, we expect them to ask the questions they need to ask, and we give them the agency and freedom to have those conversations. We want the person with the best information to implement the solution. And we want the person doing the implementing to have the best information.
As an example, when integrating our product with our clients’ platform, often times the engineering team needs to interact with the client, their technical staff and other software vendors. In these situations, we encourage our engineers to reach to the different parties when necessary.
Still, we want to be mindful of people’s time and not have every engineer sending emails with similar or related questions. We ended up with a bottom up approach, in which whoever wanted to send an email to a third party would reach out to the team to gather additional questions or thoughts to be included.
Owners of our own time
Our tech people know that if they want to make an improvement and have an idea, the field is theirs. If someone wants to take leadership on something, they can even have fun doing it.
One example is in front of your very eyes. This tech blog originated in conversations among the engineers. We chatted about this and as a company we agreed it would be a valuable platform to express ourselves, give back to the tech community and share a bit of Arena’s backstage with future coworkers — by the way, we are hiring! =)
The engineers that felt strongly about it took action and are the current maintainers and editors of this blog.
We are continually introspective and self-aware. We always ask, is our approach working or do we need to change it? We want people we can trust, who will be accountable and responsible, so they don’t blindly follow a process and end up in the wrong place.
Designing good process is not trivial. An ideal one is invisible and effective. It doesn’t get in one’s way and at the same time ensures a successful outcome.
This ideal might not be attainable, but we strive for it. It’s why we promote continuous adaptation. At Arena, everything is up for questioning and to be re-assessed: team structure, meetings, tools, process, etc. All are maleable, with the potential to be improved to help us achieve success.
Take meetings as an example — a word dreaded by a lot of engineers. We try to be very mindful about them, since there’s a lot of friction to get everyone together at the same time. Also, it simply takes people’s time.
Depending on the phase of a project, it might be valuable to get together more often, tightening the loop of updates and information. But, as soon as the context changes, those are re-evaluated and usually cancelled.
With respect to the meetings, if they start losing focus, it’s the participants’ responsibility and duty to put a meeting back on its tracks. If it has no clear agenda, people are encouraged to bring that up. In those rare cases, meetings are cancelled on the spot and rescheduled when a clearer objective is provided.
We are humans
At Arena, we’re not just cogs in a machine, churning code, but whole humans. We take into account our feelings, limitations, ambitions and fears. Not as inconveniences, but as valuable aspects of who we are.
We ask, “How are you feeling?” Questions like this create space to open up about how you are feeling, to talk about how to move away from a feeling of anxiety and concern into being energized and excited.
Personally, I have learned that being aware of my emotional state and addressing any needs that come out of it enhances both my creativity and my productivity.
The bottom line is that we want people to be happy, empowered and fulfilled in their work.