Improving Data Trust and Context, and Accelerating Data Governance with Atlan
The Active Metadata Pioneers series features Atlan customers who have recently completed a thorough evaluation of the Active Metadata Management market. Paying forward what you’ve learned to the next data leader is the true spirit of the Atlan community! So they’re here to share their hard-earned perspective on an evolving market, what makes up their modern data stack, innovative use cases for metadata, and more.
In this installment of the series, we meet Tony Baker, Data Governance Manager at Purple, a digitally native company, and the designer, manufacturer, and retailer behind a suite of market-leading mattresses, pillows, and bedding products. Tony shares how Atlan will support their nascent Data Governance program, and the success criteria that helped Purple choose the right Active Metadata Management platform.
This interview has been edited for brevity and clarity.
Could you tell us a bit about yourself, your background, and what drew you to Data & Analytics?
Getting into Data & Analytics was sort of an accident, believe it or not. I moved from New Jersey to Utah to ski for six months, 15 years ago. When I realized I wanted to stay here, it dawned on me that I needed a job that did not depend on the weather.
So, I started working for the State of Utah as a Welfare Counselor and bounced around through various roles at the tail-end of the recession. I found myself in some economic development roles where I was trying to boost jobs in Utah’s local economy. With that, I was tracking grant dollars and doing my own analytics on the side to predict what I could do with these dollars, and how far I could get a tax dollar to go, how to convince our legislature to commit more if I needed it, or how to convince them otherwise if a program didn’t work.
If you’ve ever had any exposure to the public sector, if you don’t stay in your lane, you can kind of get yourself into trouble. I didn’t necessarily get in trouble, but my Director at the time “caught” me doing this and said, “You have a future in this. You ought to pursue a degree in it, or whatever you need to do.”
Long story, short, I ended up going to the University of Utah and got a Master’s in Econometrics. I wrote my thesis on some of the grants that I was doing. So, it was really cool to have work at school, and school at work, and it made the program very seamless.
About halfway through my program, they offered me a role as an Economic Analyst for an executive branch agency. I was analyzing refugee and welfare populations, and how they impacted the labor economy here in Utah. I did that for about two years, then ended up at Overstock.com.
I was a Category and Merchandising Analyst for a little while. I was a dedicated analyst for a vendor, and then was promoted to run the Supply Chain Analytics and Data Science division, which I did for about three years before coming to Purple.
Purple brought me in to start a Data Governance program. Data Governance is as new here as I am, so it’s only about seven months old now. I had done Data Governance at Overstock in an extracurricular fashion. One of my mentors and close colleagues was a Director of Data Governance there, so he brought me in to help establish a strategy, and ultimately stood up Alation. It was cool to have that exposure coming to Purple, and to use that as a benchmark. It’s been about seven months here at Purple, and I love it.
Would you mind describing Purple, and how your data team supports the organization?
I would classify Purple as a Manufacturing/Retail company. We make our own product, and then sell it. We are omni-channel, so we have a brick-and-mortar retail presence, we have a wholesale presence, and then we have our own website through eCommerce.
As for the data team, when I started we were at the back-end of a transition between Senior Directors of Analytics, and there’s still a little bit of a philosophy shift going on.
Our previous leader had an approach that said everybody should have access to all data, and then could inquire about what they needed in a self-service fashion with analysts there to support them. Our new leader has an approach that says everybody should have access to data that they need to successfully do their job with some restriction and standardization. When there’s access to too much data, you can get into this “Wild West” mentality. As you said in Atlan University, “Teamwork makes a dream work, but too many cooks spoil the soup.” That’s where we’re at right now. We’re trying to fix the soup.
We have a bit of a centralized data team. Our Data Enablement team, Analytics, and Data Engineering are centralized. Omni-channel and Operations Analytics are part of that, as is Data Governance, and then the rest of it is federated. eCommerce, FP&A, Innovations, Accounting, Marketing, and Contact Center, all roll up into the business units that they help support.
As far as how the data team supports the business, we’re currently more of a report building team than we are an analytics team, in the traditional sense. Our analytics teams work to produce data elements within Looker, and then provide them in a dashboard to their stakeholders.
We’re trying to switch that and create a culture shift, especially going into 2024, where we can hopefully establish a CEO-CFO relationship between an analyst and their stakeholder, and have that back-and-forth that can drive the business a bit more.”
What does your data stack look like?
NetSuite is our ERP. We have Looker as our reporting platform, and Snowflake as our data warehouse. Data travels through Fivetran and Matillion for our ETL processes, and we also use dbt.
Why search for an Active Metadata Management solution? What was missing?
First and foremost, the “Great Data Debate,” in my boss’s words, was very real and very intense when I got here. We had four different definitions of Gross Sales, for example, with inconsistent sources of truth in terms of how things were documented and defined.
We also had a bit of a Shadow IT issue going on. We’re not necessarily trying to rein it in from a “Data Police” standpoint, but it’s about getting everything out in front of us using a tool like Atlan, and then starting to figure out how we can implement a Data Governance strategy. That’s exactly why we brought Atlan in.
The data culture at Purple is very much in its infancy. Purple’s only about seven years old, so there’s still a lot of very new and “start-up-y” things that we have going on here.
Why was Atlan a good fit? Did anything stand out during your evaluation process?
I come from a background where my father was a builder, so I use the analogy of building the house, and living in the house. Those are two tasks in and of themselves, and they can both be cumbersome in their own ways. So, I use that for Atlan, and Atlan stood out in both ways.
When it came down to it, our final two were Atlan and Alation. With Atlan, we were able to plug in our tech stack for the Proof of Concept, and we were able to do that in a matter of minutes. Our Data Engineering team were already solving issues in Atlan before Alation was even turned on, and while we were waiting for it to be connected.“
The biggest thing is that we have a very new data culture here, so I tried to put myself in the shoes of our stakeholders who would be using this, and I asked, “What’s going to be easier?” When I put together a test team, I presented a very real Purple problem and I said, “Use Alation, and use Atlan to solve this problem. Which one’s easier?”
Our problem had six criteria:
- Nomenclature and context of terms in certain situations, within Looker, Snowflake, the source, and the platform. How many places is this term used and in what context? Consistency in definition is important, how well does the platform show this?
- Can the platform take a shortcut on root cause analysis on why two numbers don’t match? If we are using the same word to describe two different metrics, how do we easily get to the calculation and definition of each? How do we ensure consistency in the calculation and components that derive it?
- Can we track down a very specific metric, and verify accuracy amongst multiple sources and identify a source of truth? Can we clearly identify contextual usage?
- How well does the platform navigate the data, schema, and table structure within the tech stack?
- Does the platform adequately address trust flags?
- ODBC connections and governance: Can the source of the query be traced through the platform?
Our test team was as “blended” as I could manage. We had business stakeholders at all levels, a couple of Data Engineers and Analytics Engineers, Analysts, Analytics Managers, myself, and an Admin who’s now responsible for Atlan.
It was a simple evaluation process, and with roughly 90% of the votes in favor of Atlan.
What do you intend on creating with Atlan? Do you have an idea of what use cases you’ll build, and the value you’ll drive?
We’re trying to roll Atlan out in a phased approach. Right now, we’re at the tail end of phase one. We’re working with our Customer Success Manager, and I’m building out the structure of the glossary.
I’ve already solicited input from potential stakeholders and analytics leadership on what they want that structure to look like. That way, when we evangelize our data stewardship program, which is coming up in the next month or so, we can hand that over to them, and they can start curating. A lot of that legwork has already been done.
Our Analytics Engineering team has been working on re-architecting our data. So they’ve already been using the root cause and impact analysis features of Atlan to help that project along. And our Data Engineering team got ready to work on cost optimization.
In the next month or so, we’re going to bring all that together and start a Data Steward team. We have an Atlan steering committee where we meet to define our goals and what we’re trying to accomplish, and will push our work through that.
My admin has done an awesome job at creating a game that we’re waiting on final approvals for. It mirrors a Lord of the Rings quest, with various milestones of how we get Atlan up and running within the next year. And this quest ends with a day of Topgolf and lunch.
Photo by Slaapwijsheid.nl on Unsplash
