#136 Scaling the Data Culture at Salesforce (with Laura Gent Felker)

What I've seen is some of the greatest innovation that's come out of my team is when we've had to solve a high priority short term business problem and a little bit of pressure is applied to us. We have to all band together to come up with the best solution possible. About three years ago, many of our listeners all experienced this is the business environment was changing so rapidly every and we needed to tell a story with data across Salesforce in a whole new way of ways that we've never looked at data.

Laura, it's great to have you on the show.

It's wonderful to be here as a guest.

Awesome. So I'm excited to discuss with you how you lead data teams at Salesforce, how you approach learning and innovation with your teams, how data leaders can scale data culture and more. But before we begin though, I'd love to talk to you about your background and how you got where you are today.

Absolutely. And I've always taken the path less traveled in life and that continues with my career. I studied economics and Russian in school and actually lived in Yatoslavl, Russia during my college years which were extremely formative and I still bring back concepts I learned today. I got my career start at the Walt Disney Company and and I actually worked in the parks and resorts down in Orlando as a lifeguard at Blizzard beach and truly learned how they provide magic every single day from their operations. Once I graduated, I packed up my little blue car at the time and I headed out to California and I took a traditional finance role at the Walt Disney company where I learned how Disney makes so much money with a little bit of pixie dust on top of that. After a few years I migrated over to tech at Salesforce and I've been at Salesforce for about 10 years. It's been a wild ride because we've grown from not even having our first billion dollar quarter to over $30 billion each and every year of revenue. My first role was in operations at Salesforce where I learned that the right person can make the right the right salesperson make the right amount. I was auditing all of that process and then I built my data team. I really missed some of the finance skills that I learned at Disney and I was able to combine ops and finance. I built my data team from the ground up, looking at the quote to cash process and how Salesforce can scale from people, process and technology. I have three arms of the organization, data engineering and science, risk and business insights as well as data literacy and strategy. I love to learn it's a huge passion of mine. So I'm super excited to be on the podcast today. One of the biggest pieces of Salesforce culture is giving back. I'm on the board of directors at Global Minded. Global Minded is a nonprofit that helps first gen college students from underrepresented minorities graduate gain internships as well as jobs post graduation. And I help advise on technology with Global Minded to help them scale as well.

That's really great. I really appreciate this holistic overview, Lauren. There's definitely a lot to discuss today on the show. One thing we think about a lot on Dataframe and on the podcast and on Data Camp. Something that we think about internally quite a lot is the importance of learning. We're going to talk about how data leaders can scale learning for the wider organization. They can democratize data skills. But what I want to first talk to you about today is, you know, how you as a data leader have approached a learning culture on your team, on the data team, and how you've seen the value of learning play out when driving innovations within data team. So maybe to first set the stage, I'd love if you could provide an overview of why creating a learning culture for data teams when driving innovation is important and how you've seen that play out.

Yeah, absolutely. And as I mentioned in my introduction, I'm a serial learner myself, so. So I want to be a multiplier within my organization. I think one of the first things that I think about with my team and those around me is creating the time and space for the team to be able to have that innovation time. If they're always operating at 100% capacity, they're not able to do so every year. As part of my goals for the last five years, I put down 15% of our time used to be dedicated towards innovation. We check in with our teammates as well, my management team and myself, to really understand what team members want to learn next. Is it more technical skills? Is it more soft skills and being able to open up those opportunities for the teammates and then also making sure the culture is fun. When you're learning, for example, creating a hackathon or being able to solve a complex business problem, it makes it fun. And I think when people are having fun, they tend to learn more.

And we're going to unpack a lot of these different items that you mentioned here. And I'm really interested in kind of that 15% innovation time. Right. Kind of walk me through how you put that in the fiscal year. Right. What's the process by which you kind of put that in the quarter and the calendar. And how does that play out in practice?

Yes, absolutely. And I think looking at 15%, if you broke it down by the week, it would be about six to eight hours per week. And making sure that almost a day, a week, somebody's learning something new is really important. Now, it doesn't mean that you have to take formalized classes. So I can kind of talk a little bit about how I approach learning. Now, of course, formalized classwork is the base of it. So really making sure that I kind of anchor on a model. 10% formalized learning, 20% of working with our community and clients, and 70% on the job. 10% of formalized learning is stuff like data camps, which I'm so excited to be on this, or getting your master's program or just a one off training. Maybe Salesforce provides a training for softer skills. For example, 20% is learning from our community and creating the right space within our teams and across all the different units within our community. And then 70% is really focused on the job, really making sure that we every day are trying to bring what we learned from that formalized learning and then transferring it to what we do in our job every single day.

You know, I love that breakdown of like kind of 10% formalized learning, 20% learning innovation, 70% on the job. You know, one thing that I think is always a sticking point for a lot of data leaders, when they try to kind of create a learning culture on their team and then they want to measure the trade off between short term business value and kind of long term innovation projects. Right. You know, while technical skills are not the only skill set data teams need to grow. As we'll discuss later on kind of longer term moonshot projects are really an incredible mechanism to develop the skill sets of different teams and new domains. So as a data leader, how do you approach prioritizing short term business value while projects and moonshot projects are kind of accounted for in the longer term strategy? What's the mental model that you use here to split priorities?

I'm going to actually rock the boat here a little bit. Adele. Why do we have to separate out the two? You know, what I've seen is some of the greatest innovation that's come out of my team is when we've had to solve a high priority short term business problem and a little bit of pressure is applied to us. We have to all band together to come up with the best solution possible. And I think something that our audience can maybe identify as is COVID19 about three years ago, many of our listeners all experienced this is the business environment was changing so rapidly and we needed to tell a story with data across Salesforce in a whole new way of ways that we've never looked at data. And the amount of learnings that from data science, from data storytelling as well as security is something that I bring forward as a leader almost three years later. And we were able to put this data story together in a matter of weeks. We could not wait months to be able to produce something. And we were able to provide insights to many leaders across Salesforce to make critical decisions.

That's a great example and maybe kind of walk me through, you mentioned here the COVID example. What are kind of other projects where you that also fit that bill of a short term project but that had to push the limits of a team and kind of create innovative projects, innovative solutions given the boundaries that you've had.

Yeah, absolutely. So I mean, I think another one is when we're looking in quotacash, we look a lot with the sales process and how to improve it. And if we are trying to scale for the future, we really have to quickly identify problems in the process where we can improve as an organization. Salesforce is continually changing, whether it's growth by acquisition or organic growth. And so we have to pivot quickly whether it's process improvement or technology improvement. And so we've had to implement a lot of data projects again within a few weeks or a month to be able to identify that, especially during critical times such as fiscal quarter end.

And do you find that moonshot projects need to fit a specific criteria before they get signed off? And so kind of what are those criteria?

One is based on business need and executive alignment. So if it's just a moonshot project that's nice to have. I kind of find that we never get it done. And so when the team truly understands why we are doing something, the real why behind it as well. So it's not just leadership understanding why we're doing it, but also the team. Then we're able to get rolling and provide some outcomes and we are able to move so much faster that way.

I want to kind of harp on what you mentioned here on executive alignment and the understanding of the why. Right. Because oftentimes I feel a hallmark of an immature data organization is an organization that just wants a shiny use case without necessarily thinking about the business value or the why or the how of a data project. Right. How important do you think is having this executive alignment and you know, having Leadership that is very much so thinking about the business value of data projects and how do you bring that about in your conversations with leadership when getting that executive alignment on data projects?

It's one of the most critical things and I'm really lucky to work for an organization where, where I have that leadership support and what I've been able to see is making sure it's two ways. Right. Executives understand the why and the stories that we uncover and what we can do about it. And then they're able to also create some hypotheses around what data problems they want to solve. And that has been imperative because it gives us more what do we need to work on right now and what do we need to prioritize? Because otherwise we'd have an infinite backlog. We'd never get through as a data team.

And if, you know, kind of flipping the switch here, like for many data leaders who are, you know, maybe struggling with trying to reach that consensus with their executives and trying to be able to build out a strategy and a, you know, a project plan that is, that is aligned with the business value, what would you, what would be your advice for kind of, you know, reaching consensus with different like functional stakeholders and business lines leaders on the data science project roadmap?

I think really understanding and hearing different stakeholders, especially if it's not executives in your direct line and really bring forward ideas and for people to understand again, why do you want to pull some, some sort of data and tell that story and propel forward? So, you know, a big theme of what I always say is listen first and hear people out if you have that opportunity, whether it's through executives or other business stakeholders, there's many other business stakeholders outside of executives that we have to get alignment as well.

Connecting back to something else that you said earlier is the importance of like learning community. Right. I think that's the 20% aspect of the learning formula that you've seen. So one thing you mentioned here is the importance of learning community when driving a learning culture within data teams. I'd love if you can just discuss how that plays out in practice. What is a great model for a learning community within a data function?

I think the multi layered approach and not everything has to be very formalized is important. So I kind of view right now with our maturity, three layers within our data team around community first is within my direct organization and building that culture of collaboration and trust so we're able to learn from each other and creating those safe spaces where our team feels comfortable being able to present Bring in new ideas and being able to collaborate. It's not all formalized programming. Some of the most beautiful things I see is when it's organic, innovative collaboration. Second, my organization has built this incredible community called all things Data across quotacash or revenue Operations. And we bring monthly meetups where we talk about technology that we're implementing as well as new project ideas to be able to really jumpstart other data analysts within our ecosystem. And then last, finance has this incredible data community ran by the finance Data Office. And there are many components of this community. They also do their own monthly meetups. And I encourage my team to present their new projects during this. It's a great opportunity for them to practice their communication skills, but also for other community members to learn with what we're doing. I find that so valuable because we're able to learn what other parts of finance are doing. And then last, on the finance data community, they have this cool data doctor program where they are able to have finance professionals. It's an opt in program. Either people that know data specific expertise or a person that wants to learn a new skill. And you can book an appointment with a doctor if you get stuck in a certain part of your data process.

I love that. I love how many programs there are at Salesforce. And I think there's really an interesting takeaway here is that there's also communities for the entire data function within the organization, but also kind of sub communities that are focused on specific problems. Right. Do you find that layered approach, especially for large organizations with large data teams that have many sub functions to be effective at free Specific, specific, certain business problems. And how have you seen that? For example, walk me through an example of the finance data community and how have you seen that in action?

I mean, I think for one, we're all at different starting points on our data journey. And so to enhance data literacy, we must have a layered approach. It can't be one size fits all. Some people have never built a data visualization and they want to start learning about what's best in class from a design. That's a very different starting point than somebody that has a master's in data science. Right. So if you did that one size fits all, it wouldn't be useful. You wouldn't be talking to the right stakeholders. And I think how I've seen the finance data community when we've had to move really fast on data projects is my own team has reached out to data doctors and been like, hey, we're stuck on this certain formula or hey, I need to figure out a model related to data science. Has anyone ever done this before? And to have those active people as sounding boards that you trust has actually helped spur innovation in a quick way.

That's great. And what would your advice be for a data leader looking to kickstart a learning community within their function?

One is listen to your community to see what type of help they need. And like I stated, being at different parts of your learning journey. That's why it's important to first listen, allow the community to provide input, to be able to provide the right enablement and the right tools for those members. And then last we spoke a little bit about this is getting that executive sponsorship so people know why this is a priority. Why is sharpening your skills within the data community important for not only the future of data teams, but the future of the entire organization?

So we spoke about executive alignment when it comes to determining the project plan for a data science team and kind of the roadmap and how to incorporate innovation there. But kind of walk me through how data leaders can also gain executive sponsorship when it comes to creating time for learning and dedicating resources and bandwidth. Just focused on innovation and learning activities as well.

At Salesforce, we have what's called the V2 mom, and it is the goals that we have every year. And I think the first step to allow that time and space that I spoke about is truly for executives to have it on their V2 mom, because if people know that the learning is a key part of the culture, they will see it by watching those folks do it. And so I'm lucky enough my executive does really sponsor that. And it's not just learning those technical skills, but it's also using data to make data informed decisions as well. So it's not just the build phase, but it's at every layer of the organization and leveraging that. And my executive really makes sure that we're doing that by monitoring our accountability metrics every year.

So I think this also marks a great segue to not just discuss how data teams can learn, but actually what they need to learn. Right? What they need to be focusing on. Data science is inherently multidisciplinary. Data scientists are required to blend broad skill sets to be able to deliver value, namely two broad skill sets. Business acumen. So this is like product sense, communication skills, ability to tell data stories, collaboration and technical skills, which is like really the nuts and bolts of data science. We talked about how data scientists can learn technical skills, for example, with moonshot projects, with the different kind of community programs that you have. But I'd love to focus on business acumen. You know, business acumen is arguably the more challenging skill set to develop as a data scientist because it blends communication skills, collaboration, product sense, and kind of things that are a bit, you know, not very hard that you can acquire just from reading a textbook. Right. What are frameworks and mental models that you found useful to improving that skillset continuously?

Absolutely. I'm so glad you brought business acumen up today because I feel like it's often forgotten in the data world. We're so hunkered down in learning the next technology or a new coding language that I feel like we don't dedicate enough time to the softer skills and business acumen. I think if you have this incredible model or a data visualization you have built but can't communicate it to your stakeholders effectively, then you're going to most likely fail that project. Going back to that same learning model I discussed before, of 10% learning, 20% community, and 70% on the job framework, with this topic, we really focus on being intentional with professional development plans around business acumen. So first, the 10% learning actually on the actual learning through classes. Salesforce offers a ton of coursework, and of course you can go outside of the company as well, but it could be stakeholder management, presenting up data, storytelling. So formalized learning, I feel like, is fully covered there. Going back to community, that looks a little different. Of course, within our data communities, we can practice the communication and presenting skills, but I think finding that safe spot where you can get that really good constructive feedback is important. You can select your manager, a peer, or even a mentor. And so another thing which this is less community focused, but more of how to practice. I encourage my team to even record themselves before a big meeting and rewatching it and taking notes of how they could improve. Now, on the job, we practice all the time with this. Whether we're working with stakeholders to build requirements or we're running a metrics call or an insights call call. And you can always ask for that feedback from your stakeholders and say, hey, am I hitting the mark? Is there a way I can improve? How did the meeting go? But the other indicator is whether or not there's engagement for the outcomes that you're trying to drive. At the end of the meeting, if you're just talking the whole time and there's no kind of connection, then it means that maybe we should go back to the drawing boards and really think about how are we going to present to create Some data informed decision making.

So I love the different resources that you have to provide data teams the ability to kind of scale and improve their data storytelling skills. But let's harp on the job segment. A lot of data scientists you mentioned here at the end, kind of engagement and people being able to kind of empathize and understand what the data team is working on and driving that adoption of the data product. How important do you find embedding data scientists in different teams, for example, and collaborating with others from different background conducive to growth and business acumen? I'm thinking here of kind of data scientists who are working with their functional counterparts directly to be able to understand the business problem at hand, rather than being just a support function and hunkered down in the center of X.

Absolutely. I think working with our stakeholders, whether it's within the business executives or other stakeholders across the company, is really learning the business and being able to sit down and understand some of the process areas we're evaluating from a data perspective. And there's two things that we're trying to do with our stakeholders. One is being able to build trust. Trust is our number one value at Salesforce and it's also our number one value in data. You want to make sure data is accurate, which is my second thing I want to talk about. If you don't understand the process or if a data scientist doesn't understand the process, they'll most likely get that data assumption wrong. And so that's why sitting down, collecting those requirements and really starting to understand the process is important. And something a fun exercise I recently did with my management team was to create a relationship map with our different stakeholders and to have varying categories around the different stakeholders we work with. And then being able to evaluate do we have strong relationships with those stakeholders, are we developing those, or do we have no relationship at all? And, and then writing it down truly helped us to be able to understand where do we need to be and how in the next quarter or two quarters can we develop some of those working relationships and what are kind of.

First steps to developing those relationships, especially with those stakeholders that you don't necessarily have an active working relationships with right now.

You know, in the remote world, I've really focused on reaching out and hearing, hearing the business problems and hearing how data can help and trying to develop those and being very intentional with those relationships. And then also building a brand across my organization of continuing to deliver can help inspire trust as well. If we continue to deliver great end products that are accurate, then other Stakeholders want to be a part of that solution.

Now we're talking here for the past kind of 20 minutes. We're talking about how to improve the skill sets of the data team. But I'd also love to talk about extending learning to be on the data team. Right. We often talk about data cap of the importance of data culture and data literacy as a force multiplier that helps the organization scale its capabilities. How much do data culture and organizational data literacy for non technical stakeholders play a role in creating this fertile environment for collaboration that we've been speaking about? And how do you foster these values as a data leader?

Super imperative. And I started looking at a Persona model and so there's two different Personas I look at when I'm thinking of non technical stakeholders within the kind of domain of people I work with. One are data consumers, people that are leveraging that data to make business informed decisions. This Persona could be leadership or they could be at a decision part of a program or a project. We need to ensure that this Persona is data literate, understands the metrics that we are providing. And the last thing is we monitor their behavior. And so we really look at are they looking at our assets that we've promoted, how often, what's their behavior? If they're not understanding why was the asset that we provided from a data perspective not helpful? Did it not hit the mark and getting that voice of customer all over again? The second one is really data analyst or have a desire to upskill and become, you know, a data expert. And we have a lot of that within the organization. They could use it for data visualization, so audits or even data science. And this goes back to the importance of community. And different analysts are starting at very different parts within the organization. So providing the right enablement as well as data governance and security model to ensure that those data analysts can be successful in their domain. And so unlocking these capabilities can be successful across the entire organization. If it's just my data team that looks at data visualizations or other elements or are the only ones building, at the end of the day, we'd have an infinite backlog for building.

Indeed, in a lot of ways, democratizing data skills is like a force multiplier for the data team in general. Let's harp on that. Data consumer Persona people need to be able to make data informed decisions. They need to be able to to consume the data insights that your team is creating to effectively navigate the organization. What do you think are key skills the data consumer Persona needs to have within any Organization today, I think one.

Is being able to understand the data definitions and numbers and how trends work and how it's going around. If you pick up an asset and have no idea what the definition of a, of a data element is, they're probably not going to use it at the end of the day. Second is being able to enable them on different technologies that exist. We're in such a rapidly changing environment. Back in the day we used to use Excel for everything and some people might be comfortable in Excel, but that has shifted and how we can see set up technology to help drive those decisions is so cool. I mean we can get more just in time decision making versus a lag of data that we had in the past. So being able to also work with those data consumers around what is the art of possible? Where do they need help like say they don't even know where to start or log in to a certain technology. Let's sit down and make sure we have that content. Whether it's a one to one meeting or if we're looking to scale, we can build a playbook.

Yeah, that's really exciting. I assume as well. With the Tableau acquisition, Salesforce has definitely transitioned from Excel to Tableau here now. Also walking us through the data analyst Persona. Right. I love the fact that you mentioned data governance and data quality. You know we had Shane Murray on the show, he, he's the field CTO of Monte Carlo Data. Walk us through from your perspective, how you've seen how important is data quality for the data team today in terms of like an organizational priority to fix it, who should own it within the organization and how does it connect to the data literacy conversation?

I mean I think there's multiple layers when it comes to data governance and security. And at the end of the day making sure those data analysts have access to the right data at the right time is super important. And so they can do their job. And so making sure that the monitoring mechanisms are in place to make sure that they're aware of how to keep data secure and what are the best in class ways that we should monitor that. And so my team has had that kind of responsibility for the last year especially. And you mentioned Tableau. As we roll out more and more Tableau and get additional use users, it's important. And the second thing is making sure that we're all drawing from the same source. So if data analyst in another org is looking at the same data that my org is that we're all drawing from the same numbers in the same source at the end of the day and having those right guardrails in place to do so, but also optimizing and making sure they have access to a to do their job and make data informed decisions.

What are ways that you've sound successful in scaling data democratization and data culture within an organization?

Yes, absolutely. And you're going to start seeing themes in what I kind of talk about, but anchoring on three different areas. One is enabling our end users, whether it's one of those data consumers or data analysts, to build this culture outside of my organization. Second is creating the right material and guardrails to make these Personas successful. And that material keeps evolving as we learn more about these Personas. So it's written in pencil, not pen, which I think is really important because the data world keeps evolving and we have the opportunity, there's new releases of data products and we want to be able to enable these folks. And then third is monitoring success for these stakeholders on both facets, whether it's a data consumer or a data analyst, to really understand their behavior and make sure that they are following the right guardrails, like I mentioned, and being able to use data on data, as I call it, to be able to have that monitoring mechanism in place.

What are key metrics that you monitor here when it comes to making sure that data is being adopted?

I think one is the adoption of the asset. How often are people looking at it and should we continue to have those assets in place? Something that we built three years ago might not be relevant today or even six months ago because maybe it's from a project that we delivered to monitor. So that's like the first one. And really understanding how often are people using different data visualizations, I think on the data analyst side is really understanding what source are they looking at and how are they deriving the data as well as looking at, are they following the standards when it comes to building out? Whether it's building out in SQL code, whether it's building out in python code or even data visualizations, are they making sure that they're approaching it in an unbiased way?

Okay, that's great. I really appreciate this perspective, Laura, and especially kind of the measurement aspect. We don't talk about, you know, what to measure when thinking about data culture. Now, Laura, as we end, as we near the end of our chat, I'd be remiss not to talk to you about some of the data science use cases you've worked on at Salesforce that helped you as a data leader grow. And you know Expand your extend your skills. Walk us through some of these use cases and how they helped you stretch your skill sets.

One example comes to mind that actually had two use cases. Every year, QuotaCash sends out a survey to one of our internal stakeholders, sales. And we want to hear from them around their point of view of our process, people and technology of how we're supporting them. And of course within that survey there's very easily derived questions such as like how satisfied are you? One to five for instance. But the one I want to hone in on is the 10 to 15 open text questions we send to our sales organization. And there are three thousands upon thousands of responses that end up coming back to us because they want to tell us and give us feedback and we don't want our stakeholders to be reading through every single response. First, if you read through it, it promotes bias of how our brain is wired. We will anchor on potentially something negative even if there's 10 positive answers to this commentary. I think second is it's really time consuming to read through all of that content and difficult to be able to tell a story at the end of the day. So my organization wanted to tell a data story on these open text questions. And so they looked at the data in two ways. One is they wanted to look at sentiment analysis to truly understand how sales is feeling about certain questions and certain products and processes. And so if they use words such as awesome, excellent, we know that's good sentiment. If they use words such as terrible or horrible, we know that they might have a pain point there. And so we're able to kind of see that sentiment for each question that we ask. Second, we look at topic modeling and also in gram analysis to be able to bring together some of that commentary. And this is really, really powerful because we're able to activate a plan around this. So if they something groups together about a product improvement related to within our process, we're like phenomenal. Let's take a look to see if it's on our roadmap for this upcoming year and be able to say awesome. We got some qualitative data from one of our internal stakeholders, sales, and this is a great thing that we should put prioritize on our roadmap for next year. If it's not, maybe we should explore that particular enhancement a bit more. And we have done this for several years. And something that I've learned is when we be able, when we're setting up these models across data science, we, we are able to then become multipliers, not just within my team or my organization, but across the company to be able to share our learnings going back to that collaboration and community, making sure that people know that these types of opportunities exist, whether or not they're in the data world, but that this is the art of possible to be able to tell that story because they might be able to come up with additional uses cases and can reproduce this type of analysis again and again to drive business value.

I love this example so much because not only does it leverage, you know, relatively simple techniques, you know, such as sentiment analysis and anger analysis, but it's also extremely aligned with the business value and it's very useful for the rest of the organization, as you mentioned here, as a multiplier for, you know, the product roadmap sales team feel like their feedback speed being heard. It's really a great way to showcase the value of relatively simple data science. But applied at extremely poignant business problems.

Absolutely. And that's why I kind of wanted to give this example because it's easy to get your feet wet with it. Plus it's very quick to implement. It's not like it takes months and months to do this research you get can quickly implement and quickly drive business value.

Yeah, that's great. Now finally, Laura, as we wrap up today's episode, do you have any final call to action for our listeners?

Absolutely. So for all the leaders out there, I encourage you to build a culture of fun and innovation and data literacy. I think it's so important. And so my three calls to action is creating that time and space for innovation. The data world is changing so rapidly. We need to make sure that we're on top of it as leaders to help drive the organization to make data informed decisions. Second is generating that programming, whether it's formalized or informal, for multiple audiences and getting those ideas. And last is how can you you inspire a culture of fun within the learning journey and make sure to continue to hunker down that maybe it's a competition. Everybody loves a good healthy competition at the end of the day and it'll really make learning stick more.

Thank you so much, Laura for coming on Data Framed.

Absolutely. Thank you.