The Human Interfaces of Data - Arynn
Arynn Martin-Post of Shipyard
This week's guest on the "The Human Interfaces of Data" is Arynn Martin-Post. Many of you will already know her or follow her on LinkedIn, etc.
Arynn used to work at ThoughtSpot, who I mentally still associate with her over any other current ThoughtSpot person, even though she has moved on. This shows the long-lasting nature of how Developer Relations activity can affect B2B tech businesses - we remember people more than product marketing. I was surprised when someone recently told me that, before I had worked for Metaplane, they hadn't heard of it and felt I had brought it to a European audience.
Arynn strives and succeeds in being very open-hearted with what she shares. I really appreciate that she shares the rough with the smooth - working in data is rarely plain sailing.
This week, the post was made asynchronously, and drill-down questions are left in line, between square brackets.
Give us a bit of an introduction about yourself:
Hi, my name is Arynn. Like many people that I’ve met in data, my path to where I am now was anything but linear. In college, I studied psychology with a minor in biochemistry. I was fascinated by genetics and neuropsychology. In graduate school, I did a 180 and studied Industrial-Organizational psychology. I became “the stats person” in my cohort. So while the psychology aspect of business and the interrelatedness of people and process still fascinated me, I got sucked into the numbers side of things for kind of a dumb reason. I was just good at it.
My first job out of graduate school was as a compensation consultant. I did market research and built salary structures using a combination of qualitative and quantitative analytics. Well, turns out I hated that. I tried to find myself again and wound up working as a host at a restaurant in Mall of America. Weirdly, I would also consider that a data role. Seating charts, reservations, balancing seating times for servers - there was always a lot of live metrics at play.
Then I landed in what feels like the actual start of my career. I worked as a Marketing analyst, Analytics consultant and then Data Viz Developer at companies anywhere from 100 to 10,000+ people. I learned how to work in different settings, team dynamics, and corporate structures ranging from flat to heavily-tiered.
And yet, I moved on from this practitioner life. At the end of 2020, I had a stroke. It make me rethink A LOT of things in my life. Life is too short to keep doing something I didn’t love. I unintentionally stumbled so far into the weeds of data analytics just because I was good at it.
So, what did I really love? What reasonable changes could I make to bring me closer to happiness? The answer was: helping people. I found a sweet spot working as a solutions engineer. I still got to work with data, but more importantly, I got to help lots of people and navigate the relationships between them. These relationships were always different at every company, too. Each relationship structure was as unique as the people in it. It turns out my psychology degree came in handy way more than I bargained for.
Now, I work as a Data Advocate at Shipyard, creating content to help people learn about all things data orchestration.
Can you describe the human interfaces you have had as a data person:
Ones that existed
That worked well
What - The hands-off manager who secretly cared. I was handed a bunch of tools and processes that I didn’t know a thing about and told to “figure it out”. If I had a question, he wouldn’t answer it directly. He always made sure that I could answer it myself, by guiding me along in the thought process.
[David Jayatillake: Being a good coach is such a great skill as a line manager, and knowing when to switch out of it because someone is struggling too much to help themselves.
Arynn Martin-Post: While I was only on the receiving end of this style of management, I hope to take this approach to any future management positions I have.]
When - This was my first true analyst job. I was very green to all data terms and technology.
Why - Growth and trust. I can’t say I’ve ever had quicker skill growth at any other time in my life. I was so unconfident because I had never had a job like this before. However, his trust in me to figure things out allowed me to trust myself more. And, because I knew he’d never let me fail, it fostered my trust in him as well. There will always be people who can do the work better and faster that you - it’s the ones that take the time to help a newbie out, instead of just doing it themselves, that are special.
[David Jayatillake: Part of why I started this Substack in the first place was to help people who are going to end up in situations I’ve been in before. This was one of the original motivations, long before I had the opportunity to found or do anything else related.
I’ve always felt that, just as people helped me get to where I am, I have a responsibility to help newer people in the industry to succeed. Your manager here has done exactly this - do you think we all have this responsibility universally, or is it just for specific people?
I’m trying to help some of the people I’ve worked with before to become leaders in data themselves and then hopefully to raise up another generation of data leaders after them… I feel like if we all do this, then we won’t be short of great data folks in the future.
Arynn Martin-Post: It is interesting that, despite the plethora of resources online that exist now to teach yourself, we still are constantly hearing how there are skill gaps in the industry. Why is that?
There are some intangible parts that you can really only learn on the job. Those parts are best learned from someone that is a manager or on-the-job mentor.
That said, I do agree that it is the responsibility of more experienced people in the field to give newcomers those intangible skills. It will fill in the gaps and, as you said, create the next generation of data leadership.]
That didn’t work well
What - Unclear team structure. One of my jobs was a contract, and the person I reported to was not the same person who was my manager.
When - The person I reported to thought (as did I) that I was their resource alone. We talked nearly every day and had weekly 1:1 sessions. I almost never spoke to my actual manager. There was, on a totally separate team, an entire process and analysis that nobody had built out. When it came time for it to be presented, it turns out I was supposed to have done it. Nobody told me.
Why - At massive companies, it’s extremely hard to build relationships and navigate all of the structural reporting channels. If nobody tells you something is your responsibility, how are you supposed to know? The standard set by my actual manager was one where we simply never communicated, so it never occurred to me to ask him if I was accomplishing my tasks. Communication matters. If a manager sets the standard of low communication frequency, don’t expect your employees to foster a high-communication environment themselves. Lead by example.
[David Jayatillake: This situation is incredibly common, especially for contractors. There is an issue where some people in a team don’t want to treat them as employees (which is right), but then actually don’t know how to treat them.
They forget that, even though the relationship is temporary by design, and that you’re effectively using a service, you still need to communicate with the human. They may be better at requirement gathering, etc than your average employee, as they need to be, but they still need to be communicated with well.
Do any of these themes factor into why this interface went wrong?
Arynn Martin-Post: That feels very close to the truth, based on my experience. There’s an impermanence baked into contracting. While some contracting roles promise the possibility of extension or hiring on, it still creates a pseudo employee for a period of time.
However, no human can be expected to be a mind-reader. Even contractors need that communication and sense of belonging to do the job they were intended to do.]
If you compare these two examples, the difference is glaringly obvious. Foster a relationship of trust and communication. While not all managers have the time to work upskilling their direct reports, you can certainly do better than never speaking at all.
Ones that didn’t exist, but should have:
The situation: Group A that didn’t talk to Group B, but we all talked to the same customers. We had our own separate learning sessions and hand-offs almost entirely in writing alone.
The reason: I left this fairly vague on purpose. It occurred more than once for me in different iterations. Oftentimes in analytics, you’ll have the movement of data from engineering all the way up through business intelligence. Sure, we talk a lot about “data pipelines”, but how much do we talk about the people pipelines? How much do these groups actually talk to one another? What is the communication like? What are the relationships like? Are they all separate islands of work, or one interconnected community, working towards the same ultimate goal? Much like a data pipeline, you don’t want things to get cut off or lost in translation. The relationships between people involved in a process work the same way. If you don’t make good connections between the people points, you’re more prone to confusion, time loss, and error.
[David Jayatillake: I often think of this as a relay race.
Except sometimes in data, the person on the previous leg to you is running on a random vector (not towards you) and chucks the baton at the end of their leg.
Do you feel this is because the groups were different in nature or had very different reporting lines (like one in Eng and one in Marketing)?
In hindsight, how do you find out the answers to your introspection questions in “the reason” section? I feel that just knowing that a team in the people pipeline is acting as an island, is the starting point to resolving the incidental pipeline problems.
Arynn Martin-Post: The relay race analogy you described is perfect.
There are so many unique situations of this issue happening that it’s difficult to generalize enough to cover them all. That’s why I gave some prompting questions instead.
Before you can give all of the solutions, you have to ask the right prompting questions to find out what the exact issues are.
In my situation, the reporting lines were entirely different, but the people that they interfaced with externally were the exact same people. One group was amazing at creating learning initiatives, while the other was (mostly) left on their own to upskill. This became painfully apparent. The external people both groups interfaced were left unhappy when something like “well group A said this, but group B said that’s not right/possible” happened.
It was an alignment issue. While in terms of organization structure everything made sense, there should have (could have) been something connecting the islands in training and knowledge.]
What are you doing now in your current role as Data Advocate, that helps make human interfaces in data better?
Tooling
Meaningfully-curated slack channels. Trust me, you can definitely overdo it on the slack channels. But sometimes, less is more.
Project management .Asana, Notion are both great options for alignment - keeping them up to date is important to instil into your data team culture.
[David Jayatillake: I think the boundary between something like Slack and something like Notion is something we all struggle with. How do you decide what to put where?
How do you decide how much time to invest?
Arynn Martin-Post: This just reminded me of something helpful I’ve had in Slack. There’s a setting where you can have employees list their typical response time. Depending on the role, the expected response times can vary drastically.
Slack is meant to be asynchronous, in my option, but being able to list how asynchronous you can be is important.
The options we had:
I reply almost instantly
I reply within the hour
I reply within a couple hours
I will reply within a day
Adding onto that - this is very much an opinion, but trying to do project management in Slack is a nightmare.
It’s too messy to have the structure needed to keep all your ducks in a row, herd all the cats, etc. I don’t care what project management tool you use, but don’t let it only be Slack.]
Process
Asynchronous daily statuses. An anonymous poll for the data team simply asking them how they’re doing that day. Are they at 100%, have too much on their plate? Not feeling well? When evaluating team output, you can compare it to the results of team function and then be much less surprised if you’re not meeting the goals you set out to accomplish.
[David Jayatillake: We had something like this weekly at another company, but engagement with it dropped off over time, as it became a bit onerous to do. I think it was called Officevibe.
How do you counteract this drop-off in engagement? Do you think there could be a way for AI to do it automatically and perhaps with a more independent perspective? Did you do this as part of stand up somehow, or separately?
Arynn Martin-Post: I also used Officevibe and completely forgot about that! I was an individual contributor. During bi-weekly team meetings when my manager took the time to address the anonymous feedback and scores, I found that really helpful.
It’s best to address issues at the source, but having an anonymous way to submit feedback or issues fosters more voices to be heard.
To answer your questions - It was a Slack bot that reminded you to fill out the question each day. If, by a certain time of day, there aren’t enough responses, the manager can ping the team to ask them to fill it out. It’s behaviorist reinforcement. Don’t. Stop. Reminding. Them. Until it becomes a habit and a set expectation of the job. Eventually, the manager will have to do this less and less. Then, when new people join the team, they will see the cultural standard that “we always answer this questions each day”. It’s less onerous to get response rates from new people when a process is baked into the culture.]
Weekly or every other week, have a learning session. Don’t talk about work to be done. The ONLY focus should be on learning. Never schedule over it. Never move it. Foster learning and prioritize it. Growth doesn’t happen unless you make it a priority.
This seems like a silly one, but it’s weirdly been awkward for me more than once. Know your “pass off” process. If you aren’t feeling well, who do you tell? Who covers for you? It’s the worst when you wake up with the flu and then have the stress of not knowing what should happen, and who to tell.
People
I normally would have a lot to say here, but just read this:
[David Jayatillake: I’ve thought about this topic a lot: I’ve heard people use the term sports team as opposed to family, but if you look at the internal rivalry, elitism, and mass dysfunction in many sports teams it’s not a great example, either. My favourite analogy is actually a pack of wolves, which is somewhat familial, but also very effective, where position is understood and able to be contested.]
The only thing I’ll add that I practice as much as possible - proactive kindness. Get to know someone, anyone, well enough to send them a friendly hello. Compliment them on something they did recently. Ask how their weekend was. Check in on them after they were out sick. There’s so many options here, you can find what feels comfortable for you. But remember - empathy is like a muscle. It may feel difficult at first if it doesn’t come naturally to you, but the more you do it, the better you get.
 | A guest post by
|