Latest / The Data Flowcast: Mastering Apache Airflow ® for Data Engineering and AI / Maximizing Business Impact Through Data at GlossGenius with Katie Bauer

Transcript

Maximizing Business Impact Through Data at GlossGenius with Katie Bauer · The Data Flowcast: Mastering Apache Airflow ® for Data Engineering and AI

• 0:00
  SPEAKER_02: On data science in particular, I think one of the reasons why that job title fell out of favor is that people who had it tended to be very dogmatic.
• 0:07
  SPEAKER_02: And this isn't true of every data scientist.
• 0:09
  SPEAKER_02: It's not necessarily the way that I was, but I worked with a lot of data scientists who they didn't want to write SQL queries.
• 0:15
  SPEAKER_02: They didn't want to build dashboards.
• 0:16
  SPEAKER_02: They were being really rigid about what the role was, as opposed to thinking about like what does this job actually require me to do?
• 0:24
  SPEAKER_02: They were treating it as a boundary where it's like, well, I have this job title data scientist, and therefore I don't write SQL.
• 0:30
  SPEAKER_02: And you just can't do that.
• 0:36
  SPEAKER_00: You're listening to the Data Flowcast, where we unlock the secrets of mastering Airflow for data engineering and AI.
• 0:47
  SPEAKER_01: Airflow users, we have exciting news.
• 0:50
  SPEAKER_01: The annual Apache Airflow survey is live.
• 0:54
  SPEAKER_01: In just seven minutes, you can help shape the future of Airflow.
• 0:57
  SPEAKER_01: And as a thank you for filling it out, you'll get a free Airflow Fundamentals or DAG Authoring certification.
• 1:03
  SPEAKER_01: Those are worth$150 each.
• 1:05
  SPEAKER_01: Plus, you'll also be entered to win an exclusive workshop with Mark Lumberti.
• 1:10
  SPEAKER_01: So make sure you don't miss out.
• 1:11
  SPEAKER_01: The survey link is in the show notes and it closes December 20th.
• 1:17
  SPEAKER_01: Hi everyone.
• 1:18
  SPEAKER_01: Welcome to the Data Flowcast.
• 1:19
  SPEAKER_01: I'm your host, Kenton Dannis, and today I'm joined by Katie Bauer, who's the head of data at Gloss Genius.
• 1:25
  SPEAKER_01: Katie, how are you?
• 1:27
  SPEAKER_01: I'm great.
• 1:27
  SPEAKER_01: How are you?
• 1:28
  SPEAKER_01: I'm good.
• 1:29
  SPEAKER_01: I'm good.
• 1:30
  SPEAKER_01: You are in the Bay Area, correct?
• 1:32
  SPEAKER_01: I am, yes.
• 1:33
  SPEAKER_01: Okay.
• 1:34
  SPEAKER_01: Were you impacted by the recent bomb cyclone?
• 1:38
  SPEAKER_01: I think I heard maybe parts of California also got hit.
• 1:41
  SPEAKER_01: Seattle, it was a huge deal.
• 1:43
  SPEAKER_02: So yeah, much better off than Seattle, but I will be impacted by it later this afternoon when I have to go outside.
• 1:50
  SPEAKER_02: It's raining quite hard, but I'll take that.
• 1:52
  SPEAKER_01: Okay.
• 1:53
  SPEAKER_01: You're getting this one.
• 1:54
  SPEAKER_01: Yeah.
• 1:54
  SPEAKER_01: I think here it's not as bad today, but yeah, earlier this week, it's pretty wild.
• 1:58
  SPEAKER_01: I seem to weirdly have been in this like tiny bubble in Seattle that was like not impacted and didn't lose power.
• 2:04
  SPEAKER_01: I mean, that's great.
• 2:05
  SPEAKER_01: Like everything else in the metro area is just down trees everywhere.
• 2:09
  SPEAKER_01: So I'm like, oh, I'll take it.
• 2:10
  SPEAKER_01: I actually got to work this week.
• 2:12
  SPEAKER_01: So yeah.
• 2:14
  SPEAKER_01: Yeah.
• 2:14
  SPEAKER_01: Perks.
• 2:15
  SPEAKER_01: Awesome.
• 2:16
  SPEAKER_01: Let's go ahead and dive in for our listeners for today's podcast.
• 2:19
  SPEAKER_01: We're going to discuss Katie's principles of running a value-centric data team.
• 2:24
  SPEAKER_01: So actually a little bit different than what we currently do or what we often do, or a little bit zoomed out.
• 2:31
  SPEAKER_01: But let's start with some background and talk about what Gloss Genius does for those who might not be familiar.
• 2:38
  unknown: Sure.
• 2:38
  SPEAKER_02: Gloss Genius, if you haven't heard of it, that's not too surprising because you're probably not a beauty professional if you're listening to this podcast.
• 2:45
  SPEAKER_02: But it's a vertical SAS app that really is a business in a box.
• 2:49
  SPEAKER_02: It does your back-end administration, helps you book clients, manage your calendar, keep track of payments, all sorts of stuff.
• 2:56
  SPEAKER_02: That's used by people in pretty much every vertical of beauty.
• 3:00
  SPEAKER_02: You might expect there to be a lot of hairdressers, but there's also all sorts of off the beaten past sorts of things like tattoo artists and pretty much any beauty vertical you could think of.
• 3:09
  SPEAKER_01: Makes sense.
• 3:10
  SPEAKER_01: And what does the data team do there?
• 3:13
  SPEAKER_02: Yep.
• 3:13
  SPEAKER_02: We're a full-stack data team.
• 3:15
  SPEAKER_02: So we're doing data infrastructure as well as analytics and ML and AI.
• 3:20
  SPEAKER_02: We are a pretty important part of the company, analytics in particular, having a lot of strategic voice and working very closely with our strategic finance function.
• 3:30
  SPEAKER_01: Okay, interesting.
• 3:31
  SPEAKER_01: And I know we're going to talk a lot about data teams today and your, again, principles for running them.
• 3:38
  SPEAKER_01: But since our listeners are mostly Airflow users, I do want to touch very briefly on how you use Airflow within your team and how you got connected with us.
• 3:48
  SPEAKER_01: Yeah.
• 3:48
  SPEAKER_02: I mean, I'll say that I personally have been an Airflow user for a long time since 2017.
• 3:53
  SPEAKER_02: I've used it at previous companies and we use it at Gloss Genius as well.
• 3:56
  SPEAKER_01: Um that would have been a pretty old version in 2017.
• 4:01
  SPEAKER_02: It was a bit nicer these days.
• 4:04
  SPEAKER_02: But we use it for orchestrating our DBT project.
• 4:06
  SPEAKER_02: We actually had a data engineer on our team recently contribute to Cosmos.
• 4:11
  SPEAKER_02: Shout out to Alejandro Ravas.
• 4:13
  SPEAKER_02: And we were also increasingly starting to use it for some of our MLOps workflows.
• 4:19
  SPEAKER_01: Awesome.
• 4:20
  SPEAKER_01: Part of that larger data stack.
• 4:22
  SPEAKER_01: Zooming out and thinking about data teams more broadly, as promised.
• 4:26
  SPEAKER_01: So you've led and built data teams before you were at Glass Genius at other big tech companies like Reddit and Twitter.
• 4:33
  SPEAKER_01: Based on your experience, what do you think is the biggest struggle for most data teams?
• 4:38
  SPEAKER_01: What have you had to address most when you were leading them?
• 4:42
  SPEAKER_02: Probably the biggest thing I see, and the mistake that comes up the most is that data teams get really focused on the how and not necessarily the bigger picture of what they're trying to do.
• 4:54
  SPEAKER_02: And this shows up in different ways.
• 4:55
  SPEAKER_02: It might be that they're overly obsessed with their technology stack.
• 4:59
  SPEAKER_02: It might be that they're overly obsessed with methods.
• 5:01
  SPEAKER_02: And that's definitely something that companies that have more of like the data scientist job title, a lot of times you'll have people who get really into certain statistical methodologies, for example.
• 5:11
  SPEAKER_02: But really what I'd say is data teams don't hone in enough on value.
• 5:15
  SPEAKER_02: And that's amorphous term, but part of that is because it's different in every company.
• 5:20
  SPEAKER_02: It might be growth at some companies, it might be that you're in a stage where you need to manage risk a bit more.
• 5:25
  SPEAKER_02: It might depend on the appetites of the people that you're working with for risk.
• 5:29
  SPEAKER_02: But really, the biggest thing is that data teams are not focused on the bigger picture and the things that the rest of the company is trying to move for or move towards, which is a major source of friction for data teams.
• 5:40
  SPEAKER_02: And a lot of times the reason why they're not seen as more important than they actually are.
• 5:45
  SPEAKER_01: Okay.
• 5:46
  SPEAKER_01: Yeah, I totally believe that has been true in my experience as well, especially before I worked at Astronomer.
• 5:51
  SPEAKER_01: I came from consulting, doing data engineering consulting.
• 5:54
  SPEAKER_01: I worked with a lot of different data teams at different companies.
• 5:57
  SPEAKER_01: And this is for sure a common problem is that disconnect between the rest of the business.
• 6:02
  SPEAKER_01: So, how have you been most successful in addressing that challenge with your own team?
• 6:08
  SPEAKER_02: Yeah, I what I've tried to distill it down to is three key priorities, which to kind of turn them into zippy sound bites, which is helpful both for coaching people and for talking about it externally, is three things that you really need to focus on.
• 6:24
  SPEAKER_02: The first is that your data team should have the same priorities as the rest of the business.
• 6:28
  SPEAKER_02: Basically, your priorities are the same as the business's priorities.
• 6:32
  SPEAKER_02: The second is that the job that you're in, the role you have, the job title you have, you should think of that as a focus rather than a boundary about what you're allowed to do.
• 6:42
  SPEAKER_02: And the third is that you really need to act like your time is valuable.
• 6:45
  SPEAKER_02: Otherwise, people will assume that it's not.
• 6:49
  SPEAKER_01: Okay, I love these three recommendations.
• 6:51
  SPEAKER_01: I want to dig into each of those a little bit more.
• 6:55
  SPEAKER_01: So I'll start with the first one.
• 6:57
  SPEAKER_01: So you said make the data team's priorities the same as the business's priorities.
• 7:03
  SPEAKER_01: And this sounds sort of obvious, but I think often doesn't happen, as you've said.
• 7:07
  SPEAKER_01: So what do you think usually causes that to not be the case?
• 7:10
  SPEAKER_01: Where would they part ways?
• 7:13
  SPEAKER_02: Yeah.
• 7:14
  SPEAKER_02: Some of this is the type of thing that I was talking about earlier where people get really fixated on technology or methodology, but a way that will often manifest is that a data team is really obsessed with getting the right answer and maybe not a useful answer.
• 7:29
  SPEAKER_02: So it might be that they're overly rigorous about something.
• 7:32
  SPEAKER_02: It may be that they they find something and they get super obsessed with it and devote a bunch of time and political capital to trying to get the rest of the company to care about something that's just not related to the top level priorities.
• 7:47
  SPEAKER_02: Really, if you you want to be fixated on sharing the same values as the business and the same priorities as the business, you need to go to the top and look at your company's strategic priorities and use that to choose what you work on, to prioritize what you work on.
• 8:02
  SPEAKER_02: The thing that maybe feels weird about this is it can make you feel like uh you're getting thrashed by stakeholders who they maybe change what they want from you or have like really short fused demands.
• 8:14
  SPEAKER_02: But the thing that you have to keep in mind is you are on the same team as them, you have the same goals as them.
• 8:19
  SPEAKER_02: And when they are successful, you are successful too.
• 8:23
  SPEAKER_02: Uh even if they can be frustrating, you need to react to them and serve them well, maybe not serve, but partner with them well.
• 8:31
  SPEAKER_02: And that sometimes means having really hard conversations with them.
• 8:35
  SPEAKER_02: If someone comes to you with something that you don't really understand why they're asking for it, that's not a good thing, but you should ask them.
• 8:44
  SPEAKER_02: You have to figure out what they actually care about.
• 8:47
  SPEAKER_02: And if it doesn't make sense, you have to be genuinely curious about why and approach what they're asking for with genuine humility as well.
• 8:55
  SPEAKER_02: Um, because maybe you understand one thing to be true, but they have a different perspective and they have different information.
• 9:00
  SPEAKER_02: And probably the answer of the right thing for you to do is somewhere in between your perspective and theirs.
• 9:05
  SPEAKER_02: And sometimes it's not.
• 9:07
  SPEAKER_02: Maybe they're completely right and you're completely wrong.
• 9:10
  SPEAKER_02: You have to approach these people with a true willingness to understand their perspective and the pressures they're under and help them to achieve what they need to do.
• 9:19
  SPEAKER_02: Because a lot of times they have a lot of skin in the game.
• 9:22
  SPEAKER_02: If the product manager ships their wrong product or completely misdiagnoses something, they have to deal with it.
• 9:29
  SPEAKER_02: Like they don't have control over whether customers are gonna adopt the thing that they build necessarily.
• 9:33
  SPEAKER_02: And that's kind of the data team struggle and complaint, too, is that we create things for internal customers and we don't have control over whether they're using them.
• 9:42
  SPEAKER_02: But the thing we do have control over is whether we truly listen and try to respond to what people need.
• 9:48
  SPEAKER_01: Do you think that there's a an aspect of that kind of cutting both ways?
• 9:53
  SPEAKER_01: So both the data team having that better understanding of what the stakeholders are asking for, and also just being having to be good kind of advocates and communicators for your own work since you're not always working with people who are as familiar with whatever the technical products are or something like that.
• 10:11
  SPEAKER_01: I think that's something I've seen as well.
• 10:13
  SPEAKER_01: Do you think that's part of it?
• 10:15
  SPEAKER_02: Definitely.
• 10:16
  SPEAKER_02: And I mean, it's not their job to know.
• 10:18
  SPEAKER_02: So one thing that you have to be careful about is not just throwing a bunch of exhaustive details at someone, but it's really helping them understand the constraints of no, this can't happen instantly.
• 10:28
  SPEAKER_02: And a way to approach that really is just thinking a lot about scope, which should be informed by what they're asking for.
• 10:35
  SPEAKER_02: And maybe a good heuristic, again, is if you don't understand what they're asking for or don't understand the broader context, ideally you do because you're working closely with someone and you understand what the company cares about, but just ask, figure it out, figure out what level of rigor you need, figure out how precise you need to be, where you can take shortcuts, and make sure they understand that you're doing that so that they understand the nature of your work.
• 10:57
  SPEAKER_02: You don't have to do it the proper way every single time.
• 11:01
  SPEAKER_02: You have to do it in a way that moves the company forward.
• 11:04
  SPEAKER_01: Yeah, that makes sense.
• 11:06
  SPEAKER_01: And so your next recommendation, because that was to treat roles as a focus, not a boundary.
• 11:12
  SPEAKER_01: Can you say more about what you mean by that?
• 11:15
  SPEAKER_02: Yeah, this one, if you think about a lot of people's conversations about data job titles, I think this kind of starts helping you understand what it's about.
• 11:24
  SPEAKER_02: I feel like job titles are constantly declared dead.
• 11:27
  SPEAKER_02: I was a data scientist for a long time, that's supposedly dead.
• 11:30
  SPEAKER_02: Um, I'm starting to see stuff about analytics engineering being dead.
• 11:33
  SPEAKER_02: I don't know what's gonna die next, but uh, I'm still doing the same work as I always have, which is interesting.
• 11:39
  SPEAKER_01: My master's is in data science.
• 11:41
  SPEAKER_01: So I moved almost immediately into kind of a data engineering role.
• 11:46
  SPEAKER_01: But I remember even back when I graduated, the data scientists was very controversial.
• 11:51
  SPEAKER_01: It was the new title at the time.
• 11:53
  SPEAKER_01: Nobody knew what it meant.
• 11:55
  SPEAKER_01: Of all of the people that I worked with and graduated with, every single one of us had a different job, but like many had the same titles, or like people who did the same job had different titles.
• 12:05
  SPEAKER_01: It's just a mess.
• 12:06
  SPEAKER_02: Yeah.
• 12:07
  SPEAKER_02: Well, and on data science in particular, I think one of the reasons why that job title fell out of favor is that people who had it tended to be very dogmatic.
• 12:15
  SPEAKER_02: And this isn't true of every data scientist.
• 12:17
  SPEAKER_02: It's not necessarily the way that I was, but I worked with a lot of data scientists who they didn't want to write SQL queries, they didn't want to build dashboards.
• 12:25
  SPEAKER_02: They were being really rigid about what the role was, as opposed to thinking about what does this job actually require me to do?
• 12:32
  SPEAKER_02: They were treating it as a boundary where it's I have this job title data scientist, and therefore I don't write SQL.
• 12:38
  SPEAKER_02: And you just can't do that.
• 12:39
  SPEAKER_02: Companies will need different things at different stages.
• 12:42
  SPEAKER_02: There's a lot of conversation right now about how jobs will change with Gen AI in the works.
• 12:47
  SPEAKER_02: And I do think that's an element here too.
• 12:50
  SPEAKER_02: But the job just it constantly evolves and it's not specific to Gen AI.
• 12:55
  SPEAKER_02: For example, Cloud Data Warehouse has made it possible to be a data engineer without a lot of systems knowledge.
• 13:00
  SPEAKER_02: The technology is always going to change and change the nature of the job, but you're still working on the same thing, which is making sure that people have the right information to be effective, that they have the right data to power operational flows, that they can train ML or AI.
• 13:14
  SPEAKER_02: It's really what does this company specifically require?
• 13:17
  SPEAKER_02: And if that's not something you do, great, you have an opportunity to learn.
• 13:21
  SPEAKER_02: And not being afraid to try new things, to try things that maybe seem like they might be at least on paper beneath you.
• 13:27
  SPEAKER_02: It's caring about driving towards outcomes and being willing to use a lot of different methods and technologies to get there, as opposed to saying, I am this, therefore I use these tools and I use these methods.
• 13:41
  SPEAKER_01: Yeah, interesting.
• 13:42
  SPEAKER_01: And how much of that do you think is on the individual team member to be open to that sort of having a focus, but not necessarily like very rigidly defined swim lanes versus something that you implement as a value for the broader team and how it works within the organization?
• 14:02
  SPEAKER_01: Or is it both?
• 14:04
  SPEAKER_02: Yeah, I would say it's both.
• 14:05
  SPEAKER_02: I definitely, in individual conversations with people on my team, will encourage them to go and use a method that maybe feels a little out of their comfort zone, or or someone will come to me and say, I want to be this thing that maybe they're not based on what their job title is.
• 14:19
  SPEAKER_02: And what I always say is, okay, cool, we'll start using some of those methods.
• 14:22
  SPEAKER_02: There's definitely reasons for you to use them.
• 14:24
  SPEAKER_02: There, there's a little bit of I'm setting the tone there with the team, but it's also a thing where I'm giving people the explicit go-ahead so that they feel like they do have the room to explore and go and do things that the company needs.
• 14:37
  SPEAKER_02: I mean, I think the most important thing to remember is if you do something that's outside of your job description and you do a great job of it, no one's gonna be like, how dare you go out of your lane.
• 14:48
  SPEAKER_02: Like they're gonna be like, awesome, we did that thing that we really needed.
• 14:51
  SPEAKER_02: And like you, you might not want to be on the hook for something permanently that is really outside of your area of expertise, but it's good for people to reach into engineering skill sets or reach into product manager skill sets or operations skill sets.
• 15:07
  SPEAKER_02: There aren't really any rules as long as you're uh making the company better and driving towards the right outcomes.
• 15:14
  SPEAKER_02: People are not gonna punish you for doing something that's not technically your job as a say a data engineer.
• 15:21
  SPEAKER_01: Right.
• 15:22
  SPEAKER_01: It's interesting to hear this in the context of I assume you're applying this to any company and any size of data team.
• 15:29
  SPEAKER_01: I think it's a pretty common mentality amongst like startups or smaller companies where you have few enough people that you can't afford to have people who are not willing to think outside the box or go outside of that comfort zone, versus I often think of like much larger companies as having a little more defined roles in terms of what each individual is doing.
• 15:52
  SPEAKER_01: So maybe do you have an opinion there in terms of like stage or size of company?
• 15:57
  SPEAKER_01: And does it not matter in terms of data teams?
• 16:00
  SPEAKER_01: Do you think they should always operate in that way?
• 16:03
  SPEAKER_02: I think there's definitely value in having people who are clearly responsible for certain things.
• 16:09
  SPEAKER_02: Um, like maybe to give a concrete example from Gloss Genius, uh, we have an analytics engineering team, but everyone contributes to our DBT project.
• 16:17
  SPEAKER_02: The analytics engineering team is more responsible for the analytic the DBT project than other teams, but everyone works in it and contributes to it and builds new models and optimizes models.
• 16:27
  SPEAKER_02: So we're not telling them that they can't.
• 16:30
  SPEAKER_02: We're just making sure that the right tools are available to them, even if their job is one thing on paper.
• 16:35
  SPEAKER_02: On the matter of specialization, I definitely think there's value in it, at least in terms of clarity of who is doing what or who is responsible for what, or who do you talk to if there's an issue with this thing.
• 16:45
  SPEAKER_02: But I do think specialization is something that to really require a hyper-specialized person, you need to be a very large company.
• 16:53
  SPEAKER_02: When I worked at Twitter, it was 8,000 people, and the data science team was not very specialized overall, which maybe sounds surprising, but I don't think you often need um some really niche thing unless you have a huge scale that not even Twitter really had.
• 17:12
  SPEAKER_01: Okay, interesting.
• 17:14
  SPEAKER_01: And then your final recommendation was to act like your time is valuable so others will too.
• 17:19
  SPEAKER_01: I think this I think also hopefully sounds obvious to everybody.
• 17:24
  SPEAKER_01: I would love it if everybody did this, but tell me more about how you think this helps data teams add value to their business beyond just obviously like making an individual's life hopefully more pleasant at work.
• 17:37
  SPEAKER_02: Yeah, I think the biggest way this shows up is the help desk model or the help desk workflow where there's a data team that's sitting there waiting for people to ask for things.
• 17:47
  SPEAKER_02: And then the data team takes a request and interprets it super literally and does literally what is asked for.
• 17:53
  SPEAKER_02: A lot of people who sit on those teams do not like being on a team like that.
• 17:57
  SPEAKER_02: And they complain that people just treat them like vending machines.
• 18:02
  SPEAKER_02: And it can be hard to break out of this trap, to be clear.
• 18:04
  SPEAKER_02: I'm not saying just tread the virtual tickets and do whatever you want.
• 18:09
  SPEAKER_02: But if you are only doing things that people literally ask, they will assume that you have nothing better to do.
• 18:16
  SPEAKER_02: Because if you did have something better to do, you would be doing it already.
• 18:19
  SPEAKER_02: And that's maybe a little bit of a it sounds like a tautology, but again, this is one of those things where if you see a better use of your time and you pursue it and you do it well and you prove to people that you can do something better with your time, no one's gonna be like, that was why did you do that?
• 18:38
  SPEAKER_02: It's not your job on paper.
• 18:40
  SPEAKER_02: And like you you have to sometimes start small and and do little things to prove like if I take away two hours of my week, it'll be worth it because we get this from it.
• 18:51
  SPEAKER_02: You have to build momentum and you have to build velocity.
• 18:54
  SPEAKER_02: But uh if you only ever do what people ask you to, you're not gonna seem like you have uh anything better going on or any bigger ideas.
• 19:03
  SPEAKER_02: And like a big way that engineering teams actually, I think, have uh communicated to people that they have better things to do with their time than say, like a salesperson can't just DM a random engineer and be like, hey, a client, a customer that I was trying to sell to, ask for this.
• 19:20
  SPEAKER_02: Like people would never expect an engineer to just build something because a salesperson DM'd them in Slack or whatever.
• 19:28
  SPEAKER_02: But people do with data teams, and it's because data teams just they listen.
• 19:32
  SPEAKER_02: One of the ways that engineers have done a good job of conveying that they have better uses of their time is that they actually produce essentially artifacts of here is my plan, this is something that requires thought, this is something that requires work.
• 19:46
  SPEAKER_02: The institution of the design doc, I think, is really important for communicating that engineering work is complex and complicated.
• 19:52
  SPEAKER_02: And this kind of gets back to what we were talking about earlier, where if someone comes to you and they ask for something unreasonable on a data team, and you you might need to tell them what is required.
• 20:03
  SPEAKER_02: And maybe you don't write a full-blown design doc, but you like write up some requirements of like, okay, here are the steps.
• 20:08
  SPEAKER_02: I need to pull this data.
• 20:09
  SPEAKER_02: We've never worked with it before, so I need to do some QA.
• 20:12
  SPEAKER_02: I need to write queries that will be slow running because it's a huge unmodeled data set, et cetera, et cetera.
• 20:18
  SPEAKER_02: Or it's like just giving them a sense of like this is actually something that requires some thought and requires some time.
• 20:23
  SPEAKER_02: So they start understanding and respecting what they're asking for a bit more.
• 20:27
  SPEAKER_02: Because a lot of times when people ask for something, it's not something they've really thought through deeply.
• 20:32
  SPEAKER_02: It's just a passing curiosity.
• 20:34
  SPEAKER_02: So when they ask you, they they don't think, yes, I want this person to spend an entire week doing this thing that I asked them for.
• 20:41
  SPEAKER_02: They're just hoping that they can get it quick.
• 20:43
  SPEAKER_02: And if you tell them that's actually really complicated and here's why, they'll start understanding, oh no, that's not a good use of your time.
• 20:49
  SPEAKER_02: These are things that you need to do so that people understand what they're asking for is not cheap and you have to do a bunch of work on it.
• 20:56
  SPEAKER_02: Maybe this would change in a world where everything is done by some Text to SQL bot, but that's not the world that we live in at most companies right now.
• 21:04
  SPEAKER_02: People will be involved in in collecting and fetching data and answering questions.
• 21:07
  SPEAKER_02: So it's something that they need to understand what it costs because they don't want you to waste your time.
• 21:12
  SPEAKER_02: They just assume if you don't say anything, that it's not a big deal for you.
• 21:16
  SPEAKER_02: And that's also why they're disappointed when things take a long time.
• 21:19
  SPEAKER_02: You have to help them understand what they're asking for and act like your time is valuable.
• 21:24
  SPEAKER_01: Yeah, that makes sense.
• 21:25
  SPEAKER_01: I internally for my own team, I always refer to this as the we are not a deli counter.
• 21:30
  SPEAKER_01: We're not just I've said that many times.
• 21:33
  SPEAKER_01: Like we're not just here to take orders without questioning or anything like that.
• 21:37
  SPEAKER_01: And I think I've seen this happen at Astronomer as well.
• 21:40
  SPEAKER_01: I'm sure maybe it's universal with data teams where everybody wants data on something that they can view or a dashboard they can interact with.
• 21:49
  SPEAKER_01: And I think as you touched on, that's not always going to be super valuable to the business, but you could take up a lot of people's time just by every single request that came in.
• 22:00
  SPEAKER_01: So that that totally makes sense.
• 22:02
  SPEAKER_02: Yeah, it's interesting.
• 22:03
  SPEAKER_02: I feel like it's almost the opposite of the world that a lot of us were kind of like we we grew up in, I guess, for lack of a better term, where data was rare and bringing data into the conversation was super different.
• 22:15
  SPEAKER_02: Now I feel like it's everywhere.
• 22:17
  SPEAKER_02: Um, and you have to tell people don't eat too much.
• 22:20
  SPEAKER_02: Like you're gonna make your stomach sick.
• 22:23
  SPEAKER_01: Yeah, yeah, definitely.
• 22:24
  SPEAKER_01: You can totally get out of control with it pretty quickly, especially with yeah, all of the metrics floating around.
• 22:31
  SPEAKER_01: I struggle with this with my own team as well.
• 22:34
  SPEAKER_01: Which ones I'm gonna pull for my own dashboards.
• 22:38
  SPEAKER_01: Cool.
• 22:38
  SPEAKER_01: Well, I think those are three very good recommendations.
• 22:42
  SPEAKER_01: I want to touch on just pulling it all together.
• 22:46
  SPEAKER_01: What do you think data teams stand to gain from implementing all of these things?
• 22:52
  SPEAKER_01: What does that mean for them?
• 22:54
  SPEAKER_02: Yeah, probably the first and foremost tactical thing is you'll have a better time working for your company if you're not just fetching, if you're not working at the proverbial deli counter, perhaps you'll be invited into some more interesting conversations and have a better understanding of how your work connects to what other people are doing as well.
• 23:14
  SPEAKER_02: And I will say if you start acting this way, people start thinking about you differently as well.
• 23:19
  SPEAKER_02: Your team gets more praise, more shout-outs, you get involved in the higher leverage work and the more exciting work, and you really get to feel more like you're building something with everyone else.
• 23:29
  SPEAKER_02: Feeling like a member of the team is a really important and motivating thing for many people.
• 23:34
  SPEAKER_02: And acting like this is something that helps you do that.
• 23:37
  SPEAKER_02: If you're acting like you have the exact same priorities as everyone else, you are just a team member at that point, then you're not just a deli counter worker.
• 23:47
  SPEAKER_01: Yeah, totally.
• 23:48
  SPEAKER_01: I love that.
• 23:49
  SPEAKER_01: I always end with a standard final question, but this time I'm gonna do a bit of a different spin on it.
• 23:56
  SPEAKER_01: And I want to ask, what is one thing that you think open source projects, or maybe I'll even broaden it to tech products in general, should keep in mind for helping users deliver value to their business?
• 24:10
  SPEAKER_02: Yeah.
• 24:12
  SPEAKER_02: I guess the most important thing really is ongoing usage is a really useful signal, even if it's not the entire signal.
• 24:22
  SPEAKER_02: And the reason I say that is I think we have a lot of ideas about people in tech have a lot of ideas about people are using my thing because of XYZ.
• 24:32
  SPEAKER_02: And sometimes you can tell that and and sometimes you can't.
• 24:35
  SPEAKER_02: But if they continue to use it and they continue to come back, they're getting some kind of value from it.
• 24:40
  SPEAKER_02: So that's like the first signal.
• 24:42
  SPEAKER_02: It's not necessarily the only signal or the one you should end with.
• 24:46
  SPEAKER_02: But if you don't even know how often someone uses something that you've built, or even how many people are using it, go look at that first.
• 24:53
  SPEAKER_02: That's one of the most important things and can answer or even create a lot more valuable questions.
• 25:00
  SPEAKER_01: Awesome.
• 25:01
  SPEAKER_01: I love that.
• 25:02
  SPEAKER_01: Well, you know, we're about at time.
• 25:04
  SPEAKER_01: So what is the best way for folks to get in touch with you?
• 25:09
  SPEAKER_02: I'm reasonably active still on social media, not as much as I used to be.
• 25:13
  SPEAKER_02: I am I might be merry on both X and Blue Sky.
• 25:16
  SPEAKER_02: Uh, and you can find me on LinkedIn as well.
• 25:19
  SPEAKER_01: Okay.
• 25:20
  SPEAKER_01: Awesome.
• 25:20
  SPEAKER_01: Katie, it's been a pleasure talking to you.
• 25:22
  SPEAKER_01: This was a great conversation.
• 25:24
  SPEAKER_01: Thanks so much for coming on.
• 25:25
  SPEAKER_01: Yeah, thanks for having me.
• 25:26
  SPEAKER_01: It was fun.
• 25:30
  SPEAKER_00: Thanks for joining us for the latest episode of the Data Flowcast, mastering airflow for data engineering and AI.
• 25:37
  SPEAKER_00: You'll find links in the show notes to any resources mentioned in today's show.
• 25:40
  SPEAKER_00: If you're enjoying our podcast, please subscribe so you never miss an episode and leave us a five star review.