How do you bring together an organizational archipelago of data teams?
Madison Swain-Bowden – Data Wrangler with Openverse at Automattic
How do you bring together an organizational archipelago of data teams? How do you foster unity between data-focused employees spread across a central division, various sub-teams, and even individual team members wholly isolated from other data folks within your organization? For the head of Automattic’s central Data division, Anna Elek, the answer was in line with an historic foundational practice for Automattic: hold a meetup and empower the attendees to self-organize!
In September of this year, Automattic hosted its first company-wide Data+ Meetup. Anna and the event folks at Automattic poured countless hours over several months into the planning and orchestration of this inaugural meetup. We met in Madrid, Spain, and held our sessions within the business center of the Canopy by Hilton at Madrid Castellana. This was a gathering of many “datums” (data workers) across the various teams and divisions within Automattic; there were folks present from WooCommerce, DotCom (WordPress.com), HappOps (support operations), Parse.ly, Tumblr, Jetpack, ExPlat (the experimentation team), pipe (the machine learning team), Structure (the DataOps team), and more.
The week was divided into 3 primary parts:
- Open Space (2 days)
- Presentations & Workshops (2 days)
- Activity Day (1 day)
Open Space
After an initial kick-off session on Tuesday, Kirsten (one of the facilitators), orchestrated the Open Space. These two days intentionally had no agenda, so we spent the first 20 minutes or so after the kick-off coming forward with topics we wanted to discuss. The only rule for proposing a topic was that whoever did so had to be present to lead that discussion. Folks could then choose which sessions they wanted to attend. We were also encouraged to float between running sessions if we didn’t feel that we were receiving or adding value to a conversation.
This was a fantastic opportunity for members of disparate teams and divisions to come together over common questions, pain points, and goals. Attendees self-organized discussions about career paths, skills matrices within the company, communication across Automattic, positive externalities, satellite data teams, causal impact, specific tooling/software, and much more. I had the opportunity to lead two user-group open spaces: one on Elasticsearch and another on Airflow.
Both of these user-group discussions were informative for me. Within Automattic, I am a sponsored maintainer of the Openverse project, a search engine for openly licensed media. Due to the nature of Openverse, everyone on our team interacts with data in one way or another. One of my primary goals for the meetup was to connect with folks within Automattic who use tools similar to those used in Openverse, namely Elasticsearch and Airflow. The Open Space provided a mechanism for me to connect with those peers, as in a more structured way than I could have achieved if the entire meetup schedule was set before we arrived. Instead of going around asking every one of the other 60 attendees “Do you know anything about Elasticsearch?”, folks were able to opt into an ad-hoc discussion about the tool. While many of those who gathered for the session had some experience with Elasticsearch., There were others like me who had minimal experience and who therefore came to absorb knowledge.
We were given almost no structure for the gathering, so I took the time to enumerate which teams were represented by the members present and talk about our Elasticsearch setups. I shared difficulties we were encountering with Openverse’s “related items” search algorithm, and folks from the Tumblr division shared how they had optimized their algorithm using embeddings and vector searches. We discussed other potential improvements, and I came away from the gathering with some ideas for refinement of our own algorithms. Before we separated, we talked about which communication channels within the company we could share Elasticsearch-specific needs and questions. Even though we were in different divisions working on entirely different projects, we had more in common than we initially thought.
I structured the Airflow gathering in a similar manner – enumerating the teams, sharing setups, and formalizing where discussions could happen after the meetup. Both of these gatherings were crucial for an independent team like Openverse. In a team operating outside of a central data division, it can be difficult to know whom to talk to, or where to ask questions. The meetup provided a space to put faces to names.
In other Open Space gatherings, the data folks were organizing this information at a larger scale. One of the major accomplishments of the meetup was a Data+ Index: a landing page for information on the data division, sub-teams, and individual data projects. This living index serves as a central point of reference for where data is happening within our organization. Across this organizational archipelago, we were able to build bridges of common ground; there was more that connected us than we suspected!
Flash talks
Another core feature of Automattic meetups is flash talks. A flash talk is a short talk (usually less than four minutes) which can be about anything the speaker wants. While the flask talk I gave was an intro to Openverse, we heard talks about back health (from Lucila Stancato), Guatemala (from Raul Arevalo), tricks for beating jet lag (from Dave Machado, exceptionally useful since I flew over 12 hours), the Yi Jing (from @firatgelbal) and even a tambourine lesson (from @robfelty). The flash talks gave us a beautiful slice-of-life of the data folks. In addition to knowing who to ask about Elasticsearch quirks, I also now know who to ask for back stretching tips!
The Fishbowl
We ended our formal time together with a fishbowl coordinated, once again, by our facilitator, Kirsten. The format of the session is a bit hard to describe: five chairs were placed in the middle of the room, and four people occupied as many chairs. Any time someone from the room wanted to speak, they would come forward and occupy a chair. This was a signal to the others within the circle that one person had to get up and return to their seat. This meant that at any given time there was always one chair unoccupied, a signifier for the room that anyone was welcome to come in and offer their perspective.
This session was exceptionally cool and fascinating to observe! The initial topic was interconnections of data at Automattic, and several of the team leads were brought up to kick off the discussion. Many, many topics were discussed, and conversation meandered across concepts, divisions, ideas, plans, and datasets. This also spawned some discussions about potential collaborations between Openverse and Tumblr – an exciting prospect on both ends.
Beyond the sessions
The meetup was so much more than the sum of its sessions. Every night we organized small, randomized dinner groups which went out to various local restaurants. This was yet another great chance for cross-pollinating and mingling, on top of being able to enjoy the local flavor! One of my favorite meals was at Azotea del Círculo. We spent the evening on a rooftop at the heart of the city, surrounded by gorgeous views and wonderful conversation.
Our activity day was a blast. We followed a self-guided tour that spanned several miles around Madrid. Among the various sites we saw were El Buen Retiro Park, Mercado de San Miguel, the Royal Palace, and the opera house.
Charting the Archipelago
If there’s one thing data folks love, it’s data! And the data we received by way of feedback from the attendees showed that this meetup was a resounding success. Attendees said they felt a stronger sense of connectedness with their other colleagues and with Automattic in general. A large majority of attendees noted that they had an amazing time. Every attendee surveyed said they came away from the event with a better understanding of various aspects of Data at Automattic, and almost every attendee reported that they had held meaningful conversations or discussions with their colleagues.
This was the first large meetup I attended since joining Automattic, and it’s clear to me why the company values them so much! It was certainly one of the top highlights of my first year at Automattic. I’m thankful to Anna Elek and the other event organizers. This incredible experience would not have been possible without their hard work. Finally, thanks to Automattic for encouraging genuine connections amongst colleagues, supporting our shared learning, and providing memorable and fun experiences to its employees.
If working at Automattic sounds like it might be for you, find out more about it by visiting our company website.
Data for Breakfast 