Surviving and Thriving as Data Team of One

We know that many in the data analyst and data decision making world are often teams of one. There might be one data analyst in a company, even an SMB or large company, that person is a lone wolf. While they report to many different business units within an organization who want data and need the insights, or utilize the data that you make available in your data stack or your business intelligence tool, you are sometimes just one person. It's up to you to not only keep all the ducks in a row and prioritize all of the requests, but also try to avoid burn out while also keeping your own saw sharp. It can be hard to take a step back and pursue our own professional development so we stay fresh and good at what we do.

Here we talk to three individuals from different parts of the country, who work in different size businesses who are a data team of one to learn their tips and tricks for fighting burn out and staying sharp.

Betsy Morris is a BI Senior Developer at Spanx. She is responsible for BI and analytics on the IT team. Betsy helps them build and put together reports from different teams data sources using Panoply and Tableau. At Spanx, they do not have any developers to work on integrating their paid media sources such as Facebook and Instagram, Double lick, and Pinterest. Instead of their Marketing Manager pulling ad spend per day into a spreadsheet from all of those different sources and then analyzing that, they wanted to integrate all of them into one data base, which is where Panoply came in.

Max Morganfield is a data strategist at marketing agency Room 214. By implementing Panoply, they have defeated the challenge of a very small team getting disparate data sources into one place by implementing panoply into their data stack. Having the tool and their disposal has bee instrumental in onboarding new clients faster.

John Wessel is the Director of IT and Digital at Fresh Water Systems. Their small team has saved countless hours since implementing Panoply to their data stack. Having a tool where other people in the organization know they can easily check to get access to their data without having to constantly ask John has been a huge help.

The following is a live interview with Jason Harris, Panoply’s Evangelist, about surviving and thriving as a data team of one with Betsy, Max, and John.

Jason: Max, can you talk about the data stack Room 214 has in place and what I'd love to hear is the types of data you pull in, how you wrangle all that data and then how do you expose it to the decision makers?

Max: Absolutely. We definitely are working off a pretty simple structure at this point, which has been very great for me because there's not a lot of places I need to navigate between and get an understanding of what's going on. We use a majority of our connections to social and digital marketing tools like Ad Words or Facebook pages, things like that. We're running a lot of that through Panoply with some support from Stitch. Stitch is bringing in some of those other connections that either aren't represented in Panoply or has something that we can add to that.

It all pipes into Chartio. Right now that's all internal facing. We're working with Chartio to make sure that we're addressing the client needs as far as external dashboards that they can access, whether that's through their own portals or through our own portal as an agency. Right now, it's dashboard per client as well as an aggregate view, which is, I think a huge bonus of getting all of our data into a warehouse and not spreadsheets or small databases all over the place. Being able to notice trends across clients, being able to see how our clients are comparing to the overall lot, that's definitely been a huge bonus there.

Jason: Perfect. Thank you very much. Betsy, can you talk about your data stack and tools of choice?

Betsy: Sure, absolutely. For our production system we have a Microsoft stack as I mentioned before. This specific use of Panoply in the paid media and advertising arena is kind of a stand alone tool. We use Panoply, a number of ETL ingestors from Panoply for Facebook, Double Click, and one other I can't think of. Then we also use Fivetran. We use Fivetran for some CSVM and some data ingestion that isn't automated or doesn't live in the data warehouse. We also use Google Sheets.

We do have some forecasts that are constantly changing that don't live in a data warehouse right now. The Google Sheets integration is wonderful because the entire paid media team can be updating and working on their spend versus the data that's coming in from our production system and from our advertising sources on the fly because they may want to spend more or less depending on how the advertisements are actually doing throughout the day. Panoply's ability to ingest on a regular basis gives us a very quick view, much more often than once a day. Then we pull everything together with Tableau and deliver that to the end users with Tableau server.

Jason: John, how about you?

John: Yep, thanks. Likewise, we had a legacy SECO server system and have been moving away from that, really trying to standardize everything to the Panoply data warehouse. We also leveraged Stitch data for ETL for a few sources that Panoply hasn't made into their ingestion engine. Then on top of Panoply we use Mode Analytics, which is an analytics platform which also has some data science capabilities like Python and AR support. We don't really have client facing data per se, but as far as internally we are primarily dealing with a lot of advertising, digital marketing data inside of the stack, but expanding into some more operational data.

Jason: No tool or tool set is perfect. Everyone has workarounds or tips and tricks that you would tell someone if you knew that they were setting up a similar data stack for the first time. Let’s talk about workarounds or ways to overcome limitations of those tools. What’s your favorite workaround or tip that you would give someone using the stack you have in place at Spanx, Betsy?

Betsy: The best workaround for me is Tableau. There are certain data pieces that I wasn't ready to move or to ingest into Panoply and I wanted to leave in the legacy stack. I can be up and running very quickly with data that was manually pulled, that's already on the web because the ETL integrations are fantastic from Panoply, but I do not have to take my production data and put it into the Panoply data warehouse because of Tableau. So, I like the single source of truth concept. We are not quite there yet because the truth serves across very different pieces of information.

I do like that I can basically create any table I want to create in Panoply. It's very easy. I can slap on a VSV and create a forecast or a plan and put it into the data source and have that as a part of my reporting structure very quickly. I just think that the Panoply system is very reliable and it's very quick, which is great. My workaround is basically just utilizing the tools that are there, not trying to move everything at once, but move them slowly and efficiently to ensure that people get a chance to look at them before we jump all the way in and spend a lot of development time.

Jason: Absolutely. John what say you about workarounds, hacks, other supporting tools?

John: I have two workarounds. One of them is the S3 capability. If you have data in some kind of archaic system, maybe an old relational database or some system where it just doesn't have an interface to get data out of natively and maybe it's not even a very common system, so the S3 capability is great because you can write some kind of really short script, or maybe a lot of systems have an FTP style upload. We have, for example, an old phone system, I got the data out of the old phone system with a really tiny little script, get the CSV into S3 and then Panoply ingests from there.

Then Zapier is a cool one. A lot of our more modern systems have web hooks, so there's some event that happens inside your system, like for us a shipment is shipped and that connected to Zapier. You can create a message and then that message can be ingested into, for us, we use Amazon SQS, which is a messaging queue service and then Panoply can ingest from SQS.

Jason: What time saving hacks do you employ to stay effective? Then, a second question that's related, how do you automate? I think it was John Wessel who told me in person back in April that if you're asked the same question more than three times, you do what you can to automate a process to answer the question.

John: Yeah. There's probably a couple. One would be like I was mentioning with S3. That was something that would have been manual or stayed in a spreadsheet.I think another time saving hack is just working directly with end users, understanding their domain. Just actually taking the time to go through the full workflow of report creation or metrics creation and mapping that out and then going back through and doing automation.

As far as personal time saving and developing reports, I end up saving myself the most time when I spend time with the data upfront doing exploratory data analysis because it's easy to just jump in and get excited and write reports and do cool things, but if you don't really understand the domain, you don't really understand the data, then you're going to do a bunch of rework because you didn't actually know what column A or B meant to the full extent.

Jason: Betsy, how about you?

Betsy: I'm going to second the spend a lot of time with the data. One of the things is across the different groups of people here at Spanx, so we've got logistics and we've got people who are producing the products and we've got people who are selling the products, their interests may sound like they want the same types of things, but they don't. Spending time to show them what the data looks like and having fast response times like our response times from Panoply so that they can say, what I meant was X or what I meant was Y.

Jason: A common theme I'm hearing and I relate to this because when I was a reports developer a few jobs ago, definitely you have to have that human element. We're almost like data ambassadors a lot of the times, where we have to dig into the question being asked to us by an individual and say, what are you really trying to get at because maybe the question they ask isn't really what they're truly, truly asking. Max, what about you?

Max: I think something that I've learned throughout our initial set up and definitely incorporating more and more is, not only asking what do you need to see, but sitting down and saying, well how do you get there in your UI? What steps are you taking to get there? Then showing them, okay well here's these four tables to join to then get to this answer and there are these eight dimensions that all are describing similar things, but not quite the same thing. So it's been very helpful to sit down and have those conversations. It definitely helps that exploration when you're going through and digging through schemas.

I think one thing that's been for me, and I'm not sure if this is necessarily time saving, but it's certainly headache saving, is a lot of calendar blocks and giving myself hours when possible to sit down and actually solve a problem. That sometimes doesn't work out the way you planned, but at least it's something that you can start practicing.

Also, automating where I'm going to have to fix something that day. It's just kind of inevitable, I think especially working with digital marketing data that things are going to break. There are API changes every week, there are all sorts of things like Facebook being down, etc. Setting up alerts and dashboards that can basically give me the quickest picture of how are things running today, what fire do I need to go put out? Instead of having to get a note or a Slack message from someone saying, “Hey this chart's broken for some reason, I'm not really sure what's going on.” That I want to avoid at all costs and make it seamless for them on their end.

Jason: How do you ensure your processes are repeatable and scalable? Let’s start with John.

John: I think this is an interesting one for me because I've definitely made the mistake early on in not doing that well. I think where I am now, in this company, I've been here about three years now. I've started thinking, designing with the end in mind and that all of these processes that we introduce, that we're really intentionally not introducing things that aren't repeatable and scalable. Which seems obvious, but if you're not intentional about it, you will inevitably be constantly introducing non-repeatable, non-scalable pieces. Panoply is a huge part of that. Using Mode Analytics for us is a huge part of that because it has reusable definitions as part of their platform. So we say margin and you say margin and you say margin, we all mean the same thing. It's not like accounting has their calculation and marketing has their calculation.

There are things like that and then when you have a shared definition, if we need to update the definition of margin, does that affect 200 reports and are they all linked to that same data object? We're by no means perfect at that, but thinking that way, really thinking more like a developer or programmer from that aspect has been helpful I think, for our organization.

Jason: Great, Betsy?

Betsy: In this particular arena, the way we're using Panoply and Fivetran now, there's not really a lot of repeatable, each data source is new. The scalable part we rely on you for.

Jason: Fantastic, Max?

Max: I think the challenge here for the agency is definitely the different needs of clients, different business goals, different connections that we're dealing with. Different metrics within each of those things. What we started out doing, and I'm really glad we did, because it's made things challenging in some ways as things come up, but definitely overall much easier to digest, is really setting up, especially through Chartio and Panoply as well, setting up these views and these data stories within Chartio that allow us to focus in on the key metrics that are essential to all of our clients and combine all of that data.

Jason: What tips do you have for breeding adoption of data and insights to business intelligence? To expand on this question a little bit, how do you keep stakeholders abreast of what data is available to them and then what tips and tricks do you have for getting them to use the tool of choice that you have inside your organization? Let's start with Betsy.

Betsy: So, all of this is very difficult. It requires a lot of communication. The adoption of data and insights is dependent upon people's needs. I feel like often people do things in a very manual way until, like John brought up, they start saying oh I have to do this every month, robot arms driving me crazy and then their boss says why are you spending seven hours a week putting this together and then it gets escalated back over to myself and the IT team to figure out some solution. The adoption of it comes sometimes not from me pushing, but from me just waiting for somebody to decide that that's not the best way to do it anymore. Of course, I would love to just go to everybody and be like, that's just dumb, why don't you do it my way. That's not what happens in organizations and people hear from a friend, or they hear that there's other ways to do things and then they bring it to us.

I think it really is when people are ready to adopt, to look at things differently and to say hey I kind of need some data to drive this decision. That's been my experience in the retail world.

Jason: Awesome, John?

John: I've had a very similar experience. I think as far as keys for adoption I think it's probably the hardest thing you're going to do. I think the technology and the plumbing and all of that can be hard, but it's gotten a lot easier in the last five or 10 years. It's going to be hard and you have to be prepared for that.

I think second, each group may be different. I know for my accounting group that if it can end up in a spreadsheet somehow and it's accurate and they feel comfortable with the accuracy which is very important to them, then they're good. Then for my sales group, it's like it needs to have flashy charts and animations and stars. It's different, and they don't care if it's accurate at all. You can make up the numbers, they don't care. There's just different needs per group, understanding that, having a solid enough relationship because unfortunately, people will lie to you.

I think trying to have those open, honest communication can be tough because either they feel bad that you spent a whole week on this report they never use. Just trying to keep that really honest and fluid can be tough, but it's worth it.

Jason: Neat, Max?

Max: Yeah, I can feel everyone's pain here. Given the week, about 50% of my role is that adoption piece and I think to almost sum up what everyone is saying, I'm a huge believer in the Switch Framework, which is a great book. If you haven't read it, it's all about change management. Part of what I think everyone has said here is finding the bright spots. That's a big part of it. It's easy to focus on who's not reading the report every week or who is giving you the most feedback that is maybe changing week to week, it's easy to get hung up there, but when you can focus on who's using this, who has been very vocal about appreciating this, what is it that they appreciate about it? What has it helped? We started incorporating those stories to the agency overall and showing this can help you in a way you haven't thought of at this point yet.

Then I think, to build on what Betsy was saying earlier is, one thing I like to do whenever I'm building a new dashboard, whether it's for a new client or a team or whatever it's solving, we have a meeting, we go through everything, we communicate on what needs to be in there, I build it, I build everything that we asked for and then I add one fun thing that I hope will show them that there's a different way to think about the answer or the questions they're asking.

Jason: That's great. I have a question from a viewer. What advice do you have about building views and data stores so that end users have an easier time with them? Then also, how do you educate these end users about views ergo which view to use when?

Max: I guess I can hop in on that one. I think John mentioned this and this is something we've been trying to accomplish. It's hard to do for sure, keeping that dictionary of terminology. It's hard to keep updated. It's hard to make sure you, yourself are referencing that as much as you should be. But really the key part of that is communicating that. We've done that a few ways. Had quizzes for the agency on what does this mean, we work in marketing so there's a lot of terminology and buzz words that are thrown around, so I think another key piece is, in those conversations making sure you're asking when someone says something that might be a little vague, just asking them to reiterate that and maybe get a little more specific.

As far as being in the tools, I have found in Chartio at least, being able to describe what custom columns mean. It does give you that option. Or being very explicit and not worrying about the length of metrics, names and making sure that that's all very, as clear as possible. That's my suggestion for that.

John: Couple things from my standpoint on that, I think a really crucial thing for us was adopting the Extract, Load, Transform (ELT) model. I know a lot of legacy systems have a transform right after extraction. For us, that really increased clarity as to this is the raw unaltered data and it got it way closer to the user. There wasn't this black box where only the one guy in the basement knew how to modify it.

Jason: What is your favorite resource for professional development? Also curious to know if you have a preference or resource, do you prefer in person professional development or if you do online stuff? Let's start with Max?

Max: I think maybe, I'm like some of our listeners and Betsy and John, I don't come from a programming background or any coding background, so when I was starting with all of this, Data Camp was really where I spent a lot of my time. I still use that for refreshers and learning new syntax and ways to work with things. That's definitely great. I think Ed Ex is where I've done a lot more of my deeper learning. It's a free online too. They do have the ability to pay for a certificate, it's like $100 or something like that, but it's great courses that I think have allowed me to dig a little deeper and take what I have learned and keep moving it forward.

Jason: Excellent, John?

John: I think there's a lot of great stuff out there. I've done a lot of Udacity courses. There's a lot of really good free ones. Everything from if you want to get into machine learning or AI or you just want to learn a little bit of Python or brush up on your SQL skills, Udacity is a good one. Then we have a company subscription to Plural Site which has a lot of good courses on it, it's like $30 a month, and that one we do more for day to day stuff. They've got more Microsoft, SQL Server, Cisco, more traditional IT training on that one.

Jason: Betsy, how about you? How do you keep your saw sharp?

Betsy: Well I am a big fan of professional development. I do a lot of courses on Udemy. I also just got my Masters recently. I went through Virginia Tech's Online Masters of Information Technology. It had been 20 something years since I graduated from undergrad in Finance. I’d worked in IT forever, but I wanted to go back and refresh the skill set because my concern was some of the legacy data, data structures, data analytics ways of building business plans, even professional management skills maybe needed a little refresh. That was super fun. I can't believe how interactive and how wonderful these Masters programs are online. It's very doable if you set time aside every week. Now I have so much time I don't know what to do with myself. It was the best experience and the best use of my time.

Jason: That’s great. Let’s move on. How do you avoid burnout? Are there any fun out of work activities you do to stay motivated and focused while you are in the office? Let's lead off with John.

John: Again, I think it's difficult. I think especially, at least for me, I've ended up starting a couple places pretty much being a team of one and then transitioning into a team lead or manager type role. That especially is a really easy place to get burnt out because you're not only an individual contributor, you're also trying to do at least some team lead management type functionality and I think those are such different activities, where if you're trying to block off time to actually be heads down working versus blocking off time to be in meetings and administrative work and emails and stuff.

So I'd say segmenting those two separately, as much as you can. Blocking off a morning and then in the afternoon don't try to cram in more analytics in the afternoon, which you already know is full because the switching back and forth is just exhausting. If you try to sit down, you're like I'm going to knock out this query and then you get interrupted, or then you've got 15 minutes and you try to cram this in, now you've got to go to this meeting, at least for me, I burn myself out that way instead of just trying to good block of time and then when you've got those afternoons or times when you know you're going to have a meeting here, a meeting there, just give yourself a break for those short times and don't try to cram more into the time then the time would allow.

Jason: Betsy, how about you?

Betsy: Well, I work with some really, really great people and we talk about women's clothing. So burnout isn't super bad for me. Spanx is a very healthy work-life balance and we have a lot of events here. There are Fireside Chats with some amazing people, and we had Jane Fonda here recently. We have something called a Be Bold Bootcamp, so we went through debate and public speaking and four different modules as a company to work on our skills as an individual. I think the professional development stuff is really helpful to say, I'm going to spend four hours today professionally developing on the company time and they mix those in quarterly and also monthly. That's very helpful.

Sometimes I'll work with finance and they'll want an Excel spreadsheet and I'll be like, today is not an Excel spreadsheet kind of day. So I try to be honest with people when I'm not in a place where either I have, my mind's in a place where I can be creative and come up with a great solution, or something where I may not be able to add as much value because, as John said, right now my mind is in a financial place, or right now my mind's in a sales place.

Jason: Perfect, Max?

Max: Over the holiday break I read a great book called Nature Fix which I strongly recommend. I will not go on a spiel about it, but one of the take aways I had from that is, every day I go for a walk at 2:00 p.m. anywhere from five minutes to 30 minutes. Being in Boulder, we're 30 feet away from a beautiful tree lined path that turns into dirt and plains. I really try and get outside and away from a screen as much as possible and that pertains to outside of work as well.

I think in the office, it's an open office environment, being in Colorado work life balance is a very valued thing here, so it's not too much of a struggle, but I definitely try to get up after maybe staring at a query or a screen for two hours and just go bother my co-workers and try to talk about whatever show I watched the night before.

Jason: Thanks, Max. Next question: How do you use either outsourced or offshore resources to help scale your department or organization? Then the second part of this question is, how do you keep those people who might be maybe Upworkers or consultants, how do you keep them plugged into what you're doing so that they can immediately contribute rather than you having to coach them too much? Let's start with John.

John: I think rightly so this maybe has a bit of a negative rap as far as outsourcing, basically, you're just taking advantage of the labor market that's cheaper than wherever you are, whether it's outsourcing from another part of the country or overseas. I think the good side of this is if you have a good relationship with whatever the outsourced part is, whether it's Upwork, whether it's an actual outsourcing company. I know for us the key is Asana for project and task management. This sounds super basic, but just ask yourself: What's the task? Who's it assigned to? When is it due? And then communication around the task, give clear direction around what needs to happen.

Jason: Ok, Max, do you outsource any work?

Max: We don’t but I do have a member of my team that is remote. So to John's point, I think there is a lot of communication. She may argue that I may not be doing that enough. It's something I'm trying at, but I think that's something we would be looking towards in the future, especially from a more high powered analytics perspective.

Get a free consultation with a data architect to see how to build a data warehouse in minutes.
Request Demo
Read more in:
Share this post:

Work smarter, better, and faster with monthly tips and how-tos.