The Emerging Role of a Data Product Manager

all right uh let's get started uh this is yeah welcome to the bi reporting and business use case uh track of dcla 2020 my name is kalyan todorov i'll be your host uh for this session i'm also joined by mark waddell who is who will be the co-host and in just a little bit we'll be hearing from cinderella on the emerging role of the data product manager uh before i introduce our speaker today uh just a quick note um you could as a all attendees are muted but you could post in chat using the hoover app and you could also post questions in the q a section of uh the same app and we'll get to those at the end of the talk um so with that let me let me introduce our our speaker um so uh cinderlin uh well throughout her career cindy has made complex information insights accessible to colleagues and customers uh yielding positive results across numerous industries including technology gaming social media research and manufacturing she has applied a set of data principles and product management tenants to her current role as a data product manager which is the focus of her talk today and personally i would like to say that as a practicing data scientist and engineer i am very interested to hear what cindy has to say so without further ado let's welcome cindy and the virtual uh floor is yours take it away thank you so much cal um yeah thank you everyone um for joining me here today i'm gonna share my screen now all right um i assume everything's working well um you should be seeing my title screen here um yeah so um yes thank you everyone really excited to be here today it's my first day to con and um i'm here to talk to you about a role that i'm starting to see emerge in data organizations called the data product manager i'm really designing this talk to really appeal to two audiences one those in data management and company leadership may be trying to figure out whether or not they need a data product manager on their team and those of you who are maybe thinking is the data product manager role um maybe my next career path or a path i want to pursue so to get started before we even talk about what a data product manager is let's kind of do a quick refresh about what a product manager is um so to me a product manager is someone who solves problems um using a set of skills the first is customer empathy understanding the pain that a customer is going through so that you can figure out which ones are the most acute which ones are causing the issues and which ones you want to solve using creative thinking innovation to solve those problems using an analytical mindset having an analytical mindset so you can not only prioritize what's the most important but start to develop the metrics that'll measure how successful a product is and what is the next step to kind of move forward from there and finally relationship building typically as a product manager you're leading through um influence you're you typically don't have a team reporting to you engineering um typically sits on a separate um under a separate uh manager and so it's important that you're able to build relationships um so that you can get done what you need to get done and so um someone who solves those problems through these skills um creates a product or creates a solution that typically is thought of as a product typically this this can be software hardware but i'd make the argument that anything that solves a problem using these um specific tools is uh it's still a product it might it might not be technology and so what is a data product manager well a data product manager is very similar someone who solves just data problems it's just more specific to data but is still using those hills the same skill sets and they're still creating a solution that is still a product and so um i think we as a data community have experienced we know um some of our most common data problems i've listed some here whether that's poor insufficient data if you look at that data and use that for analysis you're not going to get very accurate results maybe it's a skill issue it's a lack of data analysis skills or lack of ability to communicate with that data um that definitely is something that i've seen and then maybe data is either not utilized or selectively utilized i've been at unfortunately too many places where people want to go with their gut instead of going with data and um that's definitely frustrating and so these are just some typical ones i've heard so much more just in the last day and a half of this conference i'm sure you guys have some i've seen some really great discussions around it but i would argue that these are probably the most common ones that i encounter in my day-to-day as a data professional and i think this is something that often comes up with again data management company leadership and oftentimes what i'm seeing more and more is they recognize that this is an issue and so we want to do something about it and usually that's something is maybe implementing a data culture initiative which is great uh except for i often find that those initiatives are not super well defined um it's pretty ambiguous who owns it what success is going to look like and so i would argue you know solving a problem solving these problems with a solution that causes more problems is maybe a place where a data product manager can step in and have a um a really uh big impact and so today um what i'm going to be talking about for basically the rest of the the talk here is um how my current employer hopskipdrive has used this role um to really solve this problem of um of this to create a data culture so and that's really through looking at the um the issues that they're facing and at least having at a high level a strategy of treating data as a product um so yeah so before we get into that entire story let's just very quickly talk about who hop skip drive is um hopscope drive is a company that's focused on creating opportunity for all through mobility what that really means is that we connect students with our army of of care drivers on our technology platform we typically work with parents schools government agencies and increasingly have been focused on um making sure that we have transportation um set up for our unhoused foster and youth with special needs um we're venture-backed based here in los angeles i know not everyone is based here now but um yeah that's that's who we are um and so uh i started about a year ago and before i started there was um as a growing startup there was a recognized need that um data was a big part of how they wanted the company to grow i think you know we all kind of understand especially as data professionals the impact the value that data can have and the leadership especially knew that but what they were seeing is that they had um the individual employees were really struggling with how to use that data effectively and how to have that data-driven mindset i think a lot of us a lot of leaders want their teams to have and so they really recognized that there were two main problems um surrounding this and had two strategies in place to address them or we wanted to implement two strategies the first um let's let's talk about the problem first the first was the fact that there were over 200 workbooks reviews in tableau or data visualization tool um and many of them were being under utilized um even though and and how that manifested is that the business intelligence team was still fielding questions around where is this data how can i find this what is this number and oftentimes it wasn't complex questions or complex um data that they were looking for it was just like what it what was our you know how much uh how many rides are we expecting to do tomorrow that kind of thing and so um the high-level solution here or the strategy that they wanted to implement and i say i mean company leadership wanted to implement um was treating data as a product um and there are some questions around that right like what are the problems or really the barriers that are preventing that better adoption of those existing data dashboards so that was the first problem and the second problem it was there were certain employees that were able to find that data um but still struggling to develop insights that they can then use to make decisions and so it was really about bringing that product that data to the people was a strategy they wanted to implement which really just means building up the analytical skills in each of those departments so obviously that means some training but you know they they acknowledge that the individuals that should be trained into what level really varied um amongst individuals in just their each of the departments and so um how they decided to go about this is uh they wanted to hire a data product manager and so um we'll just share here um some the mission that they had kind of put out um as what they wanted this role to accomplish um you know i'll just read this out they want to take this role to accelerate adoption of business intelligence self-service enabling the entire staff to take action at a greater scale while maintaining excellent service i think that's great but really you know let's break that down it's just really about making sure that um it's about democratizing data right um what i think is even more powerful is what the key outcomes for this role they were um they were targeting right so this role was going to train members of each team to use tableau curate publish data sources build data products models and enhance data infrastructure so it was clear what what they wanted this role to accomplish and so for the competencies looking for first and foremost product and project management skills someone who could write and communicate effectively and also has really strong presentation skills um definitely had some data skills you can see the different technologies that were being requested here someone who has already worked in data for a bit and also has some you know people management skills right um and then lastly some programming skills linux python um which i'll admit i'm a little weaker there but um yeah so that was um kind of the makeup of the type of person definitely not someone who is starting off in their career but um can definitely someone who maybe has a little bit of bi a little bit of product and so who do they hire obviously they hired me um and so my background i have an education or i have a bachelor's of science in civil engineering from wpi i focused on transportation and traffic engineering which was um interesting when i came across topsail drive because i had actually never done that as part of my career um and so i was really interested in learning more about the company as a result and in my own past experiences my very first role out of university was as a process engineer and a business analyst working at avery dennison a fortune 500 manufacturing company where things like six sigma were drilled into me and i i really understood the value of process improvement and um and ruthlessly prioritizing things as well as um quantifying things that were typically hard to quantify so that was a big part of my experience there in building the foundation that then led me to a business intelligence analyst role at a social media startup victorious based here in los angeles and really just developing those technical skills that's where i learned sql and tableau and understood also what the metrics were for a venture venture back startup from there i moved on to a product manager role at tsia which is a research and advisory firm where the product itself was data information and insights and so it was important that i was able to distill down that information and be able to communicate that in a digital environment to provide value to the customers at tsia and if that wasn't enough in the past few years i've also been a product analytics consultant for a gaming studio here another also based in los angeles um just working on refining those skills becoming a strategic partner for them and really thinking about how do we kind of take the data that they have and make product decisions off of it um and also at the time i was looking for a product manager role at a mission driven technology company and so hopscope drive was a really good fit for me i was a really good fit for hopscotch drive and so i've been there um i started in november of last year so just under a year now i think we're about to hit to uh it's like two weeks to a year so yeah that's why i joined and so what did i do when i started it starts with that customer empathy right exercising that customer empathy so it's important to have a deep understanding of what the customer problem was and so started with customer interviews and the nice thing about this data product manager role is um compared to other product manager roles you're working with internal your internal customers so the people i was working with were also my customers and so these interviews was meant to just meet the team establish rapport and build trust amongst the people that not only was i going to be working with but the people that i was going to be building products for um it was also really important to and i think this is true for every product manager um understanding their key job responsibilities and how the thing that i'm working on in this case data fits into their work lives and that's not just asking them hey what's your problem with data what are your data problems um it's seeing um how are what are the decisions that you're making with data um what are what's keeping you from making those decisions right now or um what information do you still need um being able to identify those problems in a more um accurate way to what's really happening in their in their lives i interviewed 24 people which was in a company of a hundred um was this definitely a significant chunk but i just i think that's super valuable um i'll just set some expectations here i don't know if you're going to be able to do that at every single company but i was very fortunate in that i was able to really touch a lot of people and really understand what the problems were and so as a result develop some pain points create some insight around those pain points right using that customer empathy and using some analytical skills to be able to take that qualitative data and distill it down into really three things um there was just unnecessary time spent looking for answers oftentimes that meant multiple steps were really required to answer a single question or we were seeing teams rebuilding reports or pulling data that other teams had already built there's also a lot of discussion spent clarifying data perhaps because there was inconsistent terminology or there wasn't a consistent like source of truth for certain data and so that obviously causes a lot of confusion i'm sure everyone who's who's been in that position can really relate to the kind of confusion that can cause what do you mean there what do you mean there right and ultimately what that really meant was that actionable data was underutilized it was so difficult and tedious to do data exploration to look at the data with an analytical mind that people didn't do it um we're busy people we're at a startup we're trying to do all these different things so it wasn't happening um and those that could it was only a few so it was really just the business intelligence team and um some of the executives so that's what we were really working on and you might ask yourselves well you know didn't they already figure that out before they even hired you they knew what the pain points were yeah at a high level but we um what you're not seeing is all the ways that those little things manifest and how to um and also hearing like what they've already done right so understanding what has worked what hasn't worked and so those customer interviews were really important in order to understand kind of um holistically what's happening and so uh created a product strategy and a roadmap because i'm a product manager and product managers need to have roadmaps um i'm just kidding um but yeah so what that really means is that i started with a data culture model um wanted to kind of have a way to frame the rest of the work that i was going to do and so it's really three tiers um it starts at the company level and you can see the components that make up the company level here first is trust trust in the data first and foremost people need to understand and trust the data that's coming out that they're using next is process some kind of framework to understand the data transparency and metrics meaning that there are is an agreement across the company about what metrics we're using and transparency from the top about what's happening with those metrics why they're um changing the way they are and what we're doing about it not really covering that stuff up being really honest with those numbers the next is at the team level wanted to develop tools that were designed specifically for those teams and what they were doing their responsibilities and making sure that there was a community just even within the teams that individuals were working on to feel like they were supported not just by the bi team but by others in their in their more immediate kind of working life and finally developing stuff for the employees so i'm thinking about um making sure that employees felt empowered meaning they ask and they can answer they have the capability to answer their own questions which really means developing their talent we're intentionally recruiting developing and retaining data talent and ultimately kind of the pinnacle would be if employees had a mindset that was really data focused meaning they're using data to challenge assumptions um and and have the ability to do so and so this is the model that you know i just came up with um definitely with some help from google but um this helped me kind of reframe everything else that i did and so you'll see on the road map i had kind of planned out starting in december of 2019 through the next few quarters um but it's also organized by company team and employees so again kind of aligned around those areas you can see what i did excuse me on top of that because it is important to measure what you're doing using that analytical mindset to make sure that what you're doing is having an improvement because if it's not you need to you need to refocus right um again metrics along the same kind of areas where it's based on the employee the team and the company level um and for the company level we ran a survey um and the idea was to run this every other quarter or twice a year and so kind of baselining this is where we're at with how people felt about the quality and of the value of the data that was being shared um so not bad but definitely room for improvement um and so let's talk about how we actually um get down to brass tack so what was the execution of this data culture initiative of uh for me so start off with um again we're gonna follow that same kind of data culture model that i used which was we'll first focus on the employee which uh for them really starting to develop trainings around them uh i thought about four levels of data training first being data educated which really establishes a baseline for data literacy which is really um target to everyone in the organization to make sure that they were all kind of on the same page about what we're talking about we just kind of have that baseline and so that was really important um building on top of that we had the data explorer this one was really focused around building knowledge and understanding around hopscotch drive specific data and specific metrics why pick them where you can find them in tableau target audience for this was tableau users so people who had access to data making sure that they had the ability to find that information and could make decisions on it the third level is data evaluator now something i haven't talked about yet is we initially when i had started there was an idea to hire analysts as part of the um to sit in in some of the teams that are a little bit more data focused what we decided to do was actually um identify individuals on those teams that already had some data skills and developed them further and so these are what we call the team analysts that would really mainly interface between the data science team or the business intelligence center for data science um business intelligence team and their own teams and so this data evaluator training was really focused on building um uh skills with within those team analysts to build their own reports in tableau because another thing we found is that um the majority of people who are using tableau didn't need to be building their own reports they just needed help finding the data they're looking for and understanding it so that was why the data explorer one was so important and that's what makes that one different from the data evaluator so this one was really just focused on those team analysts making sure that they understood how to build tableau workbooks off of the published data sources that the bi team have made available and finally the data evangelist training this one was focused on training team analysts to build their own data sources so taking those same individuals and kind of up leveling them when it became apparent that the published data sources that were made available no longer suited their needs so this is really just a review of sql basics and sql and database basics so that's on the training level and now again focus on what's what we're going to do for the employee um next was team dashboard so these are probably the main data product um that individual teams were going to use the goal of these dashboards were really to develop a tool that's specialized on that team's focus area and responsibilities even though if you i didn't say this out loud but the metric for the teams was to increase the views per user and so we wanted to increase um usage of the data but we actually wanted to minimize the amount of time spent looking for that data so um ideally you're you're looking at it you're getting the information you're needing and then you make a decision and you're on your way right so really making it clear what to do next from that information so um that was the goal of that and again we wanted to continuously iterate because it's a data product um things are always changing you know the sales team doesn't have the same forecast every every quarter or anything like that and um as a startup is the rule changes it was also important that we recognized that um we were going to continue changing what this dashboard was going to look like so that we could better tackle the real problems we're going to address right now um we also reorganized tableau so um we had all workbooks kind of sitting in one central folder just needed some basic reorganization um pretty simple here we organized things into one of these six folders we had our team dashboards their first one our operational workbooks which if you click in is kind of organized by operational areas by the teams themselves and then the third is completed analyses if you need to look back on on old things that's at the top level the second level is um and that's available for that should be um what everyone uses and then um that the second level here the templates the personal folders in the boneyard templates are what we've made available for those team analysts to kind of build upon um to lower that barrier to get them started with creating workbooks um five is just a personal space for individuals and sex the boneyard is just an archive so that was just like helping people find what they needed really quickly and then finally at the company level we wanted to have a space for people to train themselves around data so um it's all well and good if we do a training one time right but you need to exercise those skills in order to keep them robust and to make yourself feel more comfortable and so this data gym um is a one-hour weekly session that we make available to every employee at the company and so it's split up into two parts the first is um we the first 30 minutes we review a data topic we publicize this a few days before we've done a number of different things in um in this in the session so the first could be data requests that have come in maybe build those right on the spot we respond to those right on the spot maybe completed analyses either by my team or by other teams or even models completed by other teams we're really a big fan of in inviting other teams to present at data gem because it hammers the point home that anyone could be doing this kind of data analysis and also there are experts in your team that you can go and ask oftentimes it's the executive but that's still someone that they can go to to talk to about some of this data and the in the executives it's great because they have a lot of buy-in um into this program and so they're excited to talk about this stuff and they want their employees to do it and so um hopefully that creates a nice conversation um and then videos from experts we like played a neat silver video i'm sure we'll play some videos from datacon at an upcoming data gym and then the second 30 minutes is just available time to answer typically analyst data questions and so we just we've done we run the gamut of talking about specific tableau issues excel formulas where to find data that kind of thing so that's been kind of the the more like you have a specific need we can help train you there and then finally any good uh any good organization that has a that use the slack david slack channel um so biscuits our slack channel just like every kind of data topic that could come up we put it in there and it's all gravy so that's what we did and so let's talk about the impact kind of what the future state looks like for my role so um the first thing i'm gonna talk again moving back into what's happening at the individual level we'll talk about the dmd trainings and how successful they have been we launched the first three and you can see um the percent that we wind up training um say i wonder training um the the first two but um we realized kind of a little too late is we had included a lot of those individuals lighter we talked about that were already you know adept at using tableau had skills and data in that target audience and so that's the reason that those numbers are so low but you can see the data evaluator we've trained a number of the team analysts and have more training is coming up we haven't launched the last one the data evangelist one because we are just not in a place where um we need it that kind of tells you a little bit about where we are with our data journey so um so yeah not really needed quite yet on the team dashboard side we launched seven team dashboards but also covet and so what this really meant was as an organization who was focused on driving kids to school um as as the the only way um that we were doing things um not the only way really the primary way we were doing things um things really significantly shifted when the pandemic happened and unfortunately we had to lay people off and so um we had launched some uh dashboards prior to the pandemic um that really just didn't weren't useful anymore because of again they were supposed to be really specific to what they were doing at the time and so because we had to really significantly shift the type of work we were doing um the data team really focused on um how do we help bring down costs and make things more efficient with the smaller team that we have however recently we there are schools opening around the country not specifically in la but even california we're seeing schools reopen and so now that things are kind of getting back to normal um we are starting to revamp dashboards we've launched newer versions of those dashboards as you know what we've been working on has been changed has changed and just launched the newest version of the sales dashboard really recently um and then finally talking about what's happening at the company level with the data culture survey um we saw an uptick we ran another survey in q2 of 2020. um saw an uptick here in the perception of people's quality and value of data um seeing the biggest change with the team one of the teams that we work the most closely with which is marketplace but the sales team actually either rated things about the same or less or lower compared to the previous survey and so that's why you know i just mentioned sales we have been working a little bit more closely with them since we got the results of the survey to figure out how we can add more value for them and so in summary do you need a product manager well do you have a data problem do you want that data problem to be solved with customer empathy creativity and analytical mindset and relationship building um in getting a product out of that yeah i you know i think there's a lot of value in having a data product manager especially as has kind of been highlighted across the conference the past day and i'm sure into tomorrow data is becoming a bigger and bigger part of all companies and so um data problems will only grow and so having a data product manager i think really helps um now is it implementing a data culture the only kind of data product or data problem that a data product manager can work on no i've worked on plenty of other data problems things like customer data problems again my last role i was a product manager delivering data to customers external customers and even now in my role at hopscope drive i am spearheading kind of the reporting piece of our product because i really think it could be a big differentiator for us compared to our competitors benchmarking helping the organization understand kind of how they compare against similar organizations that's definitely a big data problem how well am i doing and where are the areas where i should be doing better so the benchmarking and then um you know i think data science project scoping that's another data problem i've seen in some organizations and you know it's definitely a problem uh that just needs a little fine-tuning before the before we can get to a place where um the projects are scoped correctly and we can kind of execute and get things um really or models in place that are really effective there's some examples um yeah so in summary um i want to before i move on to questions i just generally i'm really interested to hear about your stories um around your data culture initiatives i don't think that having a data product manager is the only way that you can get that initiative to be successful i think um it's certainly the way that we approached it but um yeah if you have more stories i'd love to hear them um you can find me at my linkedin at my twitter there but otherwise that's it oh and gonna remember to vote um otherwise are there any questions all right thank you cindy uh i do see some questions so let me read those out to you in the order they were asked the first one is hi cindy how many different products are you supporting in the data product manager role um that's a good question i kind of again think about it more as like problems that i'm supporting um there's definitely the um the problem around data utilization um at the company is kind of still my primary area of focus that's definitely the first one and the second one is is really a growing problem is how do we make sure our customers feel like they have the information they need to make their decisions oftentimes we're working with schools and government agencies and so do they have the information around their students or writers or that kind of thing so that they feel comfortable with you know where they are and their safety and that kind of thing so so yeah that's probably the two that i'm working on right now thank you oh just as a reminder to the attendees uh if you want to ask a question feel free to i think we'll have some time to answer a fair number of questions here um anyways so cindy uh second question thank you for speaking us to us today could you tell us what is the difference between a data evangelist data evaluator and a data product manager data okay let's start with data evaluator i think that's most people should be data evaluators um i don't think you need special training i think you just need to be able to look at data and data is all around us and be able to make sense of it right be able to uh make decisions off of that and i don't think that's just necessarily in your professional life you have data in your personal life um you know i um around like i'll use a very specific example covet data right like um you know you can use that information and make insightful decisions about what you want to do your own personal choice is about wearing a mask or not um and what your personal risk tolerance is so that's data evaluator a data product manager i think is just a is a role specific in organizations that have like a a fair amount of data capabilities but have trial problems maybe um or issues refining and getting a lot of traction was solving them um so i'd say that's probably a data product manager and then um a data evangelist um that was just a name that i picked for that last training but um yeah i don't know someone who just like really believes um the the power of data i think we can all be data evangelists as well agreed great thank you um and one more question which actually is my question um how do you reconcile the kind of inherently uncertain or probabilistic outcomes of building data products with like the very regimented oftentimes way that product management is done using like acceptance criteria story points um and all that jazz um so i i'll say you can still write acceptance criteria um around kind of what you want the product to do um and so is it and again i think you have some of those metrics that i had called out right like how much are people using that product um what kind of the outcomes are you trying to get um i think that can be and so we can apply that to like data data models as well um like what do i want this model to achieve i want this to reduce the amount that we wind up paying for rides i'm going to use obstacle drive specific examples on average a dollar less right so um how do we get that model to to do that and you might not have it before it goes into production but at least you can evaluate it when you're when it's out there um uh yeah and then as far as um what was the first part of your question or did i answer it i i think it did yeah yeah great thank you um i am not seeing any more questions uh so i guess with that we can end the session here a little bit early and um everyone else uh even after after the session is over you could still um attend the chat uh write in the chat uh and post questions and i would very much appreciate it if cindy you could take a look at at this and if there's any other questions answer them um offline other than that i would very much like to thank you um for for this for this presentation and for your insight answering the questions uh and also thanks to everyone else for for attending um that's it um yeah next sessions start in eight minutes so uh catch something good actually plug here for this for this um track here is that you cindy you mentioned something about being a data um data uh analyst was it uh or that everybody has to be there yeah explorer uh and and and covet and the next session is about visualizing covet data which i'm kind of looking forward to so anyways that was my plug again thank you cindy and thanks everyone else and i'll catch you later thank you everyone bye

Channel: Data Con LA

Views: 1,873

Rating: 4.8222222 out of 5

Keywords: Data Con LA, DCLA, Data Con LA 2020

Id: -4ChWRJee10

Length: 38min 31sec (2311 seconds)

Published: Wed Nov 11 2020