Pedram Navid is a familiar face in the data world, who you might already know from social media. He shares his thoughts with more than 3,300 readers through his newsletter, Pedram's Data Based, and you can often find him sparking conversations in data Slack groups, LinkedIn and most recently, BlueSky.
At our last Friends of Data event in San Francisco, we chatted with Pedram, who is now Chief Dashboard Officer at Dagster, an orchestration platform. He shared insights about why enacting change matters more than being right, why building stakeholder relationships is one of the most valuable skills in an AI-driven world, and how to think strategically about AI and data.
Listen to the full interview on Spotify.
Note: This interview has been edited for brevity.
Barry: How'd you get into data?
Pedram: A long, long time ago I was a data engineer and data scientist in finance. I used all kinds of tools that were awful – and then saw the light on the modern data stack. It changed the way I worked and think about data, and led me to working in a DevRole and data role at Hightouch.
After Hightouch I was consulting and got into birdwatching, but was getting bored of that. The CEO of Dagster reached out, and now I run marketing, DevRel, and data there.
You said you love data, as a lot of us do – why do you love data? What is it that is so interesting?
It's not that I love data – I actually love solving problems, and data is the path towards solving really complex problems.
I learned that through my work at a bank. I was not a data person by trade, but we had specific problems we wanted to solve. And so for me, it's always been about: what are the actual changes you can impact through data? It was never this Kaggle thing of like “optimize this problem” to like nth degree and get the most precision out of it. For me, it was always about: how do I enact change and how do I do that by showing proof of outcomes I think will happen?
You wrote a blog post titled, “It's not enough to be right.” Why did you want to write that?
This post was a reaction to a pattern I've seen around the mentality of being right. It’s valuable early in your career and can become addicting — some people love being right. But as the business’ needs grow and the expectations of you grow, they no longer expect you to be just right. You need to create change.
The hard problem is doing something about it: convincing those who don't believe you, networking, and getting alignment. It requires figuring out priorities and making sure that your initiative is a priority of other teams who don't really care, because you believe it’s right for the business.
The job isn't to be right. The job is to enact change. That's how you grow your career. That's how you make effects on the business. That's how, I think, you become successful.
What are the things people can do to affect that? What do you think people are missing?
Figure out who your stakeholders are, that's step one. And see if there’s an ally somewhere in that midst. There are usually people who want to make change and think it's good for the company. See if you can find one person and work together to plan your attack. It doesn't have to be perfect, but it has to start from a place where you're both ready to experiment.
Before presenting to stakeholders, it’s good to think about: what are the questions they're going to ask? What are the objections they're going to raise? How are they going to shut this down? And then just work through it. There's no shortcut, it's about interpersonal skills. It's about finding the right person and working your way up.
And, if your culture sucks, none of this applies to you. You probably need therapy more than you need my help. You can't fix culture you're not aligned with. Culture isn't always bad, it just might be the right fit. But if you and the CEO are aligned, I think a lot can be accomplished.
How do you assess culture ex-ante? If you're interviewing for a data role, how do you gauge that?
It's tough. I think the truth is, experience and time. At a startup, talk to the founder to get a sense of what they care about, how they feel about different opinions and how they solicit them, and what's important to the company. It all comes from the top down, from the CEO.
If you're early in your career, you're going to have to go through a lot of suffering and wrong fits to figure out what's right. I didn't know what I liked until I saw what I didn't like, and I just kept saying, “Oh, less of that, please,” and, “More of that.”
Is AI going to take all of the data jobs?
AI might take the fun jobs away, like writing SQL and doing pivot tables, but it will not take the jobs we don't want to do, which is alignment with stakeholders. That's always going to be something we have to figure out.
It’s rare that a stakeholder has ever looked at data and found their own meaningful insights, right? Setting up stakeholders with data is a conversation about what problems you’re having in the business, what’s keeping them up at night, and the processes that are involved in generating this data. That's the hard work of being a good data person. And the AI is not going to figure that one out.
Are there parts of our jobs as data practitioners that we should be questioning? Do you think AI is going to change how we see our roles?
I hope so. I'd love to take less stakeholder requests for basic questions every day. For example, if the Head of Product wants to know how many users we have and DMs me every time, I would love for them to DM the AI instead.
There's things that humans do for fun that AI can do better, like Chess. If people still enjoy doing certain things, will we still do them?
Sometimes the value in doing an analysis is not the number at the end. It's understanding the process, knowing where the data is, how it's generated, and all the warts that come with it. And then you ask, “How come this field that's supposed to be a number has a letter in it?” And you uncover that it’s manually entered and relates to some problem happening from four years ago.
AI is not going to ask those questions, so I think there's still value in digging in and being mindful of the data generation process and all the downstream effects of that.
I think AI benefits you well when it's answering a known problem that you want to solve over and over again. But novel, interesting, problems still occur and those are still great places to invest.
That’s how I think about writing. I find a lot of the utility of writing is the thinking behind it.
Exactly. The hard part of writing isn't finding a word on the keyboard. It's figuring out the coherent story you want to tell, drafting it and recognizing, “Oh, this sucks,” and going back, editing it, and figuring out what the right message is. That's the process of writing that's difficult and also the part that's really useful.
Why’d you join Dagster?
My early career was in Airflow and I hated it so much. The founder of Dagster, Nick Schrock, was on a podcast a few years ago talking about why he was going to go and solve it with Dagster. I tried it out and it sucked, but the vision was so good.
After joining Dagster, I found my Slack message in the Dagster community channel about what I liked, what I didn’t, and how I hoped to contribute to Dagster one day. I love the product and I like where it's going. It's evolved a lot more from where it was a long time ago.
What has it evolved to?
We're moving towards this conception of the framework for your data platform. Everyone has a data platform, whether you know it or not, whether you like it or not, and you can design one by choice. Often though, you end up with a data platform and you're not sure why or where it is or how anything works.
And so our approach has been to help people understand that. Orchestration is really the start of all that. But observability, data quality, data catalog — anything you care about within the data platform space — we want to build towards.
Data warehouses or data lakes?
I really like a data warehouse because it's convenient and not a lot of work. You can just throw the data in there — you can insert, append. You don't even need to think about it, it just works.
What ends up happening at every big company is you have it all: you have a data warehouse, a data lake, and two, three, four cloud vendors. I am bullish on data lakes, but it's not the death of data warehouses. We're going to just have more of everything.
Do you think we're going to have more or less data roles per capita because of AI?
If your entire job is to process a request every single day and you get a thousand requests all the time, that's probably a great job for AI.
The best way to keep your job is to have the people who you report into say, “I love working with Pedram. Don't get rid of him.”
That's how I think about all the stakeholders I have at my company. I don't wonder what's the value of DevRel? What's the point of data teams? I just make sure that the people I deliver things to are so happy with me that no one would ever think to ask those questions.
Is there a way to measure data team ROI? Is that a thing?
No. What is the ROI? The ROI is success of your company. It's not having a marketing person wonder: Where are my leads? Where's my pipeline coming from? How am I doing today? Do I understand my business? Can I forecast?
There's no ROI number I can attach to that. But if you don't have that, your business fails. And you can prove that by having people who love working with you and are very aligned with like your success. So if you're a data team, align yourself with a stakeholder and help them answer the questions they have, understand their business and learn.
